A Magnetoencephalographic (MEG) Study of Brain Mechanisms
in Temporomandibular Disorder
A DISSERTATION SUBMITTED TO THE FACULTY OF THE GRADUATE SCHOOL OF THE UNIVERSITY OF MINNESOTA BY
Aurelio Abdalla Alonso, DDS, MS
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY
DR. APOSTOLOS P. GEORGOPOULOS, ADVISER
December 2009
© Aurelio Abdalla Alonso, December/2009
Acknowledgments
I am thankful to my supervisor, Dr. Apostolos Georgopoulos, whose encouragement, guidance, support, patience and extraordinary mentorship throughout this journey enabled me to develop an understanding of this work.
I would like to thank also the facial somesthesia team members: Dr. Art
Leuthold for sharing his knowledge on MEG and his help in general; to Dr.
Ioannis Koutlas for inviting me to participate in this amazing project, and help during patient collection and processing data; to Dr. Elissaios Karageorgiou for his technical and general help through this journey.
My special thank to Dr. Scott Lewis for his time helping during MRI acquisition and consent forms; to Dr. May Tan for her help and patience during my MEG questions.
To all Brain Sciences Center members that direct or indirect help me conquer this journey, especially my sincere appreciation to Mrs. Penny Becker and Mrs. Gail Hollstadt that were always there for me when I needed.
Dr. Donald Simone, Dr. Alvin Beitz, Dr. Matt Chafee, and Dr. Darryl
Hamamoto it was an honor for me to have you as part of my committee, my sincere gratitude.
My many thanks to the Oral Biology Program crew: Dr. Mark Herzbeg for his inspirational conversation, and advices; to Michelle Lamere and Ann Hagen for their support during these years.
i
I also thank Dr. Eric Schiffman and Dr. Donald Nixdorf for helping me selecting my patients.
And to all my friends that followed and supported me through this journey.
ii
Dedication
For my parents Aurelio and Suraya, and my brothers Alexandre and
Alberto, who offered me unconditional love and support throughout the course of my education.
For the love of my life, my wife Claudia, who shared all the good, the bad and the ugly moments, laughed and cried together, I am thankful for your love, support, and motivation throughout these years.
For the new member of our family, our daughter Mel, who does not know what a thesis is, but knows how her dad feels about it. Just be patience and I will be joining you in your “rolling” sessions.
For my parents in laws, Claudio and Maria, who shared their love and support always.
For my friends Dr. Sillas Duarte and Dr. Fabiana Varjao, who persistent called me almost on a daily basis to learn the status of my thesis, job search, or simple to say ‘what’s up?, my infinite gratitude for your friendship.
iii
Table of Contents
Acknowledgments i
Dedication iii
Table of contents iv
List of Figures viii
List of Tables xi
Abbreviations xii
Chapter 1: General Introduction 1
Temporomandibular Disorders (TMD) 1
History of TMD 1
Epidemiology of TMD 2
TMJ Joint Pathology 4
Muscle Pathology 7
Tactile Sensory Mechanisms 9
Cortical Mechanisms of Tactile Stimulus Processing 12
Cortical Layers 14
Cortical Cells 16
Cortical Areas and Pain 17
Primary Somatosensory Cortex (SI) 18
Secondary Somatosensory Cortex (SII) 19
Magnetoencephalography (MEG) 20 iv
Forward and Inverse Problem 25
The Questions 26
Figures 27
Chapter 2: Magnetic Source Imaging 46
Background 46
Functional Neuroimaging Studies of Orofacial Pain 46
PET 46
fMRI 48
MEG 52
Summary 54
Methods 55
Subjects 55
Stimulus delivery 55
Data collection 56
Data processing and dipole extraction 56
Assignment to brain areas 58
Statistical Analysis 59
Results 59
ECD counts 59
ECD duration 60
ECD onset time 60
v
Discussion 61
Tables and Figures 62
Chapter 3: Time-Frequency Analysis 74
Background 74
Methods 78
Data processing 78
Statistical Analysis 79
Results 79
Frequency effects 79
Time course 80
Discussion 82
Figures and Tables 83
Chapter 4: Multivariate Analyses 99
Background 99
MDS 102
Method 102
Results 103
HTC 103
Method 103
Results 103
vi
Discussion 105
Figures 106
Chapter 5: General Discussion and Conclusion 110
Bibliography 116
vii
List of Figures
1.0.1 Normal TMJ Anatomy 27
1.0.2 Closer view of a normal TMJ anatomy 28
1.0.3 TMJ disc-displacement 29
1.0.4 TMJ osteoarthritis 30
1.0.5 Sensory inputs to the trigeminal system 31
1.0.6 Homunculus 32
1.0.7 Cerebral cortex layers 33
1.0.8 Ramon y Cajal cortical layers drawing 34
1.0.9 Pyramidal neuron 35
1.1.0 Generators of MEG/EEG signals 36
1.1.1 Open and closed field cell configuration 37
1.1.2 Comparison of biomagnetic and environmental fields 38
1.1.3 MEG unit at Brain Sciences Center 39
1.1.4 Magnetic shielded room 40
1.1.5 Examples of flux transformers 41
1.1.6 Structure of the axial gradiometer 42
1.1.7 Helmet layout with sensors 43
1.1.8 MEG signal processing chain 44
1.1.9 Types of brain imaging 45
2.0.1 ECD duration group plot 66
viii
2.0.2 Location of the airpuff stimulators 69
2.0.3 Example of ECD location 70
2.0.4 Average number of dipoles per brain areas per group 71
2.0.5 Dipole duration (ms) both groups 72
2.0.6 ECD onset time 73
3.0.1 Time-frequency plot 83
3.0.2 TF plot average changes over baseline 84
3.0.3 Group differences at >20Hz 86
3.0.4 TF plot (4-20 Hz) 87
3.0.5 TF plot (4-20 Hz) sensor #10 88
3.0.6 TF plot (4-20 Hz) sensor #13 89
3.0.7 TF plot (4-20 Hz) sensor #22 90
3.0.8 TF plot (140-160 Hz) all sensors 91
3.0.9 TF plot (140-160 Hz) sensor # 2 92
3.1.0 TF plot (140-160 Hz) sensor # 17 93
3.1.1 TF plot (140-160 Hz) sensor # 22 94
3.1.2 TF plot (340-360 Hz) all sensors 95
3.1.3 TF plot (340-360 Hz) sensor # 12 96
3.1.4 TF plot (340-360 Hz) sensor # 14 97
3.1.5 TF plot (340-360 Hz) sensor # 20 98
4.0.1 MDS plot control group 106
4.0.2 MDS plot TMD pain group 107
ix
4.0.3 HTC plot control group 108
4.0.4 HTC plot TMD pain group 109
x
List of Tables
2.0.1 Descriptive statistics for the square-root on ECD counts 62
2.0.2 P-values of t-test (dipoles counts, individual areas) 63
2.0.3 ECD duration (ms) control group 64
2.0.4 ECD duration (ms) TMD pain group 65
2.0.5 ECD onset times (ms) control group 67
2.0.6 ECD onset times (ms) TMD pain group 68
3.0.1 Group mean effect for TF 85
3.0.2 Frequency main effect and Group x Frequency interaction 85
xi
Abbreviation
ANOVA Analysis of variance
TMD Temporomandibular disorder
TMDs Temporomandibular disorders
TMJ Temporomandibular joint
TMJDD Temporomandibular joint disc displacement
TMJOA Temporomandibular joint disorder osteoarthritis
OA Osteoarthritis
PET Positron emission tomography
MRI Magnetic resonance imaging fMRI Functional magnetic resonance imaging
DTI Diffusion tensor imaging
SAIs Slow adapting type I receptors
SAIIs Slow adapting type II receptors
RAs Rapid adapting receptors
FAIs Fast adapting receptors type I
FAIIs Fast adapting receptors type II
SI Primary somatosensory cortex
SII Secondary somatosensory cortex
GABA Gamma-aminobutyric acid
xii
PPC Posterior parietal cortex
Vc Trigeminal nucleus caudalis
LEP Laser evoked potential
ECD Equivalent current dipole
ECDs Equivalent current dipoles
SQUID Superconducting quantum interference device
MN Minimum norm wMN weighted minimum norm
BMS Burning mouth syndrome
BMD Burning mouth disorder rCBF Regional cerebral blood flow
EMG Electromyography
ISI Inter-stimulus interval
QST Quantitative sensory test
PSI Pressure per square inch
BESA Brain electrical source analysis
PoCGy Post central gyrus
PoCSul Post central sulcus
TemOper Temporal operculum
STGy Superior temporal gyrus
Ins Insula
xiii
PrCGy Pre central gyrus
PrCSul Pre central sulcus
MTGy Middle temporal gyrus
AnGy Angular gyrus
SMGy Superior marginal gyrus
STSul Superior temporal sulcus
LinGy Lingual gyrus
SPL Superior parietal lobule
Csul Central sulcus
IFGy Inferior frontal gyrus
PCun Pre cuneus
Cun Cuneus
LatSul Lateral sulcus
CingGy Cingulate gyrus
IntParSul Internal parietal sulcus
ITGy Inferior temporal gyrus
OccGy Occipital Gyrus
SFGy Superior frontal gyrus
SEF Somatosensory evoked fields
TFA Time-frequency analysis
TFR Time-frequency representation
Hz Hertz
xiv
MDS Multidimension scaling
MVA Multivariate analysis
HTC Hierarchical tree clustering
ACC Anterior cingulate cortex
IC Insular cortex
xv
CHAPTER 1: GENERAL INTRODUCTION
Temporomandibular Disorder (TMD)
History of TMD
In 1934, an otolaryngologist, Dr. James Costen, first suggested that changes in an individual’s dental condition, such as edentulous mouth and overbite, were the causes of several symptoms affecting the ear including pain
(Costen, 1934). Costen’s initial theory has been revised several times over the years.
From the late 1930s through the 1940s only a few dentists were interested in managing these conditions. During that period, “bite raising appliances” such as vulcanite splints, partial and total dentures, and crowns were the most common treatment strategies (Bleiker, 1938; Costen, 1934; Pippin, 1940). In the late 1940s and 1950s, “bite raising appliances” began to be questioned as a therapy of choice for mandibular dysfunction.
By the late 1950s, the first descriptions of masticatory dysfunction were being published in dental textbooks (Nourallah and Johansson, 1995; Sarnat,
1951; Schwartz, 1959; Shore, 1959). Malocclusion, and later emotional stress, were accepted as the major causes of masticatory system disorders through the 1960s and 1970s. However, during this time the dental profession also
1 began to develop interest in pain disorders thought to arise from intracapsular structures (Farrar and McCarty, 1979).
By the 1980s, dentists began to understand more about the pathophysiologic processes currently thought to be involved in the development of TMDs (de Leeuw et al., 2005), including trauma, parafunctional activities, emotional stress, trigeminal pain processing, and certain occlusal conditions
(Okeson, 2003).
Epidemiology of TMD
In 1989, Brattberg and colleagues reported that 66% of 827 randomly selected individuals from the general population reported pain or discomfort in different parts of their bodies (Brattberg et al., 1989). Approximately 10% of these individuals reported pain in their head, face, or neck. A second study by
Lipton and colleagues, investigating the prevalence of orofacial pain in the
United States, revealed that nearly 22% of the general population had experienced at least one of the five most common types of orofacial pain in the previous six months (Lipton et al., 1993). The most common complaint was toothache which was reported by 12.2% of the population, followed by oral sores with 8.4%. Temporomandibular joint (TMJ) pain was reported by 5.3% of the population, with face or cheek pain being reported by 1.4% of respondents.
Burning mouth was reported in only 0.7% of the population. These results
2 suggested that nearly 40 million Americans over 18 years old had experienced orofacial pain in the previous 6 months (Lipton et al., 1993).
Many epidemiologic studies have examined the prevalence of TMD
(Glass, 1993; Hiltunen et al., 1995; Nourallah and Johansson, 1995) and found the signs and symptoms of TMD to be common. Forty to sixty percent of the population is estimated to have some type of TMD with approximately 41% reporting at least one symptom associated with TMD and 56% showing at least one clinical sign (Okeson, 2003). The prevalence of TMD signs and symptoms was studied by Glass and colleagues (Glass, 1993). These investigators described six TMD-related symptoms reported by a randomly-selected, non- clinic subjects who were contacted through a telephone survey. Of 534 people interviewed, 246 reported one or more of the following six TMD-related symptoms: nocturnal bruxism, diurnal clenching, soreness in the jaw on awakening, soreness with use, joint noise with use, and joint noise at other times. Bruxism, tooth clenching, jaw soreness, and jaw sounds were prevalent in 10 to 19%. Pain was found to be more common in people with multiple symptoms.
Most TMD symptoms are reported in patients 20 to 40 years of age (De
Kanter, 1983; Dworkin et al., 1990b; Von Korff et al., 1988). Women report these symptoms about twice as frequently as men (Lipton et al., 1993). The prevalence of these signs and symptoms of TMD have been reported to be comparable in Europe and the Middle East (Nourallah and Johansson, 1995).
3
Temporomandibular Joint Pathology
The TMJ serves as the articulation between the mandible and the cranial base. As a compound joint, it is capable of both hinging (ginglymoid) and sliding (arthroidal) movements. It is composed of the glenoid fossa, mandibular condyle and the articular disc that separates the TMJ into upper and lower compartments (Figures 1.0.1 and 1.0.2).
Temporomandibular joint disorders (TMJDs) are common musculoskeletal disorders of the jaw characterized by a dysfunction of the condyle-disc complex. This dysfunction is associated with the presence of clicking and or catching of the joint that is sometimes associated with pain.
Epidemiologic studies indicate that in the general population the prevalence of these sounds ranges from about 15%-40% (Glass, 1993; Greene and Marbach,
1982; Locker and Slade, 1988).
Disorders of the TMJ can be subdivided into three major categories including derangements of the condyle-disc complex, structural incompatibilities of the articular surfaces, and inflammatory disorders of the TMJ (Okeson,
2003). TMJ disc displacements (TMJ DD; Figure 1.0.3) and TMJ osteoarthritis
(TMJ OA; Figure 1.0.4) are the most common joint disorders (Bewyer, 1989;
Boering et al., 1990; Dolwick and Riggs, 1983; Schiffman et al., 1990; Solberg et al., 1979; Stegenga et al., 1989; Stegenga et al., 1991). TMJ DD is characterized by an abnormal relationship of the articular disc and condyle complex. Although the etiology of this disorder is still controversial, an
4 increased laxity of associated ligaments seem to be the involved in but not the causative factor for the displaced disc. There are two types of disc displacements: disc displacement with reduction and disc displacement without reduction. Displacement with reduction is described as an alteration of the condyle-disc complex during mandibular translation with mouth opening and closing. It is characterized by the presence of reciprocal clicking during opening and closing the mouth. Disc displacement with reduction is not pathognomonic for TMJD since more than one third of an asymptomatic sample can have moderate to severe derangement and as many as one fourth of the clicking joints show normal or slight TMJ disc displacement. On the other hand, disc displacement without reduction is described as an altered condyle-disc relationship that is maintained during mandibular translation. The disc is permanently displaced and does not change its relation with the condyle on translation. Pain can occur especially if the patient attempts to open the mouth.
With time, pain tends to decrease and the problem becomes more chronic, also arthritic changes may be present in this condition (Okeson, 1996).
Inflammatory conditions such as capsulitis, synovitis and polyarthrides can also affect the TMJ (Okeson, 1996). Synovitis and capsulitis are associated with trauma, irritation, or infection and often follow other TMJDs.
Synovitis is an inflammation of the synovial lining. It is characterized by localized pain that is exacerbated by function or joint loading. Swelling may be present causing inability to occlude the teeth on the ipsilateral side. Capsulitis
5 is an inflammation of the capsule that is almost impossible to differentiate from synovitis. Capsulitis is also referred to as arthralgia, discitis, and retrodiscitis.
Polyarthritides are relatively rare and are associated with rheumatologic disease. Those affecting the TMJ include rheumathoid arthritis, juvenile rheumathoid arthritis, spondyloarthropathies (psoriatic arthritis, infectious arthritis, Reiter’s syndrome), and crystal induced disease (gout, chondrocalcinosis). Other rheumatological diseases that may affect the TMJ include autoimmune disorders and connective tissue diseases such as scleroderma, Sjögren’s syndrome, and lupus erythematosus. Polyarthritides are characterized by pain during acute and subacute phases, possible crepitus and decreased range of motion due to pain, degeneration, and bony changes
(Okeson, 1996).
Osteoarthritis is considered an inflammatory arthritic condition that is found in synovial joints. It can be classified as primary or secondary. Primary
OA is characterized by deterioration and abrasion of the articular tissue and remodeling of the underlying subcondral bone due to overload of the remodeling mechanism. Secondary OA has the same features as primary; however an associated prior event or disease that overloaded the remodeling mechanism can be identified. Etiologic factors may include direct trauma to the TMJ, TMJ infection, or history of active arthritis (Okeson, 1996).
6
Muscle Pathology
Several types of muscle disorders can affect the muscle of mastication and the cervical muscles, and are associated with TMD. The mechanisms that produce pain in the skeletal muscles are not well understood. Some authors suggest that muscle overuse or ischemia of normally working muscle may cause pain (Layzer, 1986; Mense, 1990; 1991; 1993). The nociceptive endings can be sensitized very easily by endogenous substances such as bradykinin, serotonin, prostaglandin (PGE2), neuropeptides and substance P. These painful muscle conditions not only result in increased sensitivity of peripheral nociceptors, but also produce hyperexcitability in the central nervous system and localized hyperalgesia (Dworkin et al., 1990a; Mense, 1990). The majority of TMD patients are found to have tenderness of the elevator muscles during palpation, approximately 40% of these patients report pain during chewing (Dworkin et al.,
1990a).
The following muscle disorders including myofascial pain, myositis, myospasm and myofibrotic contracture, are associated with excessive or abnormal function. Local myalgia does not fulfill the criteria for the other categories above and masticatory muscle neoplasia will be discussed as follows:
Myofascial pain is characterized by a regional, dull, aching muscle pain and the presence of localized trigger points in muscle, tendon or fascia. When palpated these trigger points may produce a referral pain and/or autonomic
7 symptoms (Clark, 1985; Dao et al., 1994a; Fricton, 1990; Schiffman, 1984;
Solberg, 1986).
Myositis is characterized by clinical signs and symptoms associated with tissue inflammation. It usually begins after trauma to the muscle or spreading infection. It presents as a constant, acutely painful muscle(s), and may be accompanied by redness, swelling, and increase in temperature. Clinically, the patient may present wit limited range of motion, and, due to inflammation muscle ossification may occur, a conditions known as myositis ossificans (Okeson,
1996). Myospasm is an acute muscle disorder characterized by sudden involuntary, tonic contraction of the muscle. It presents as acutely shortened, limited range of motion, and painful. This condition is relatively rare in orofacial pain population (Okeson, 1996).
Myofibrotic contracture is a chronic resistance of muscle to passive stretch as a result of fibrosis of the supporting tendons, ligaments, or muscle fibers. This condition is not usually painful unless the muscle is extended beyond its functional length. It is common to find a history of trauma or infection in the muscle (Okeson, 1996).
Local myalgia includes muscle pain secondary to ischemia, bruxism, fatigue, metabolic alterations, delayed onset muscle soreness, autonomics effects, and protective splint (co-contraction) (Dao et al., 1994b; Sahlin et al.,
1981). Although there is significant evidence that these conditions exist, there
8 are only a few reliable clinical characteristics that can be used to distinguish them from each other.
Finally, masticatory muscle neoplasia is defined as a new, abnormal or uncontrolled growth of muscle tissue. It can be malignant or benign and pain may or may not be present (Okeson, 1996).
Tactile Sensory Mechanisms
The scientific study of the relation between stimulus and sensation is named psychophysics (Gescheider, 1985). Many different afferent somatosensory fiber types innervate the skin (Lumpkin and Caterina, 2007).
They can be divided according to their conduction velocity into A-beta, A-delta, and C-fibers. Nociceptors and temperature receptors are primarily A-delta and
C-fiber subtypes, whereas light touch sensation is mediated mostly by A-beta fibers. The A-beta fibers are further subdivided by the adaptation characteristics of the fibers and the size of their receptive field, resulting in a two-by-two classification system as follows: (a) slowly adapting with small receptive fields
(SAI), innervating Merkel cell-neurite complexes (Woodbury and Koerber, 2007);
(b) slowly adapting with large receptive fields (SAII); rapidly adapting with small receptive fields, thought to innervate Ruffini corpuscles; (c) rapidly (or fast) adapting with small receptive fields; and (d) rapidly adapting with large receptive fields (Hollins et al., 1991; Johnson, 2001; Woodbury and Koerber, 2007). These latter fibers are associated with the Pacinian corpuscles and are known as PCs.
9
Some areas of the body, such as the perioral region, may lack Pacinian corpuscles (Hollins et al., 1991; Johnson, 2001; Woodbury and Koerber, 2007).
In the skin of the human face the majority of the mechanoreceptive afferents are slowly adapting with small and well defined receptive fields. Also, only slowly adapting mechanoreceptors are found in the oral mucosa and the transitional zone of the lip (Johansson et al., 1988).
Georgopoulos studied the functional properties of high-threshold primary afferent fibers innervating primate glabrous skin of the hand (Georgopoulos,
1976; 1977). Mechanical and thermal stimulation were used to determine the response characteristics of the isolated fibers. He found that both A-delta and C fibers responded to high-threshold mechanical and thermal (hot and cold) stimuli.
Simone et al. (1996) demonstrated responses to cold stimuli from mechanosensitive nociceptors (A-delta and C fibers) innervating the hairy skin of the rat hind paw. The A-delta nociceptor response was significant higher at cold temperatures than that of C-fibers. The response of both fiber types increased as temperature decreased, suggesting the ability of these nociceptors to respond to a wide range a cold stimuli (Simone and Kajander, 1996). Finally, the properties of A-delta and C fiber mechanoreceptors and nociceptors in mouse glabrous skin were characterized after application of heat and cold in different ranges. It was found that a large proportion of A-delta and C fibers were sensitive to these stimuli (Cain et al., 2001).
10
Most primary afferents innervating cutaneous, oral mucosa, tooth pulp, joint and muscle, and cerebrovascular tissues have their cell bodies in the trigeminal ganglion and project centrally to the trigeminal brainstem sensory nuclear complex, where they ascend or descend in the trigeminal spinal tract. In the tract they give off collaterals that terminate in one or more subdivisions of the trigeminal brainstem sensory nuclear complex and activate second order neurons within or adjacent to this trigeminal brainstem complex (Figure 1.0.5). In addition, the trigeminal brainstem complex receives afferents inputs from other cranial nerves, as well as from upper cervical nerves which also send afferent inputs into the upper cervical dorsal horn. Some of the trigeminal afferents may project to other parts of the brainstem such as the solitary tract nucleus, reticular formation, and supratrigeminal nucleus (Dubner et al., 1978; Sessle, 2000). The trigeminal brainstem complex can be subdivided into main sensory nucleus and the spinal tract nucleus that is formed by three subnuclei named oralis, interpolaris, and caudalis (Dubner et al., 1978; Sessle, 2000). It is in the subnucleus caudalis that most of the sensory including nociceptive information is received and further transmitted to higher centers (Sessle, 2000). Most sensory neurons in the orofacial region relay in the brainstem nuclei. The trigeminal nucleus extends from the medulla, through the pons, and into the midbrain.
Orofacial trigeminal inputs impinge on neurons which decussate in the brainstem and ascend toward the thalamus; these neurons also make contacts with interneurons and motoneurons. The next relay station for the sensory afferents
11 is the thalamus, where they contact with third-order neurons which project to the sensory cortex. Cell bodies for neurons in the spinal system are located in the ventral posterolateral (VPL) nucleus of the thalamus, whereas cell bodies for the neurons in the orofacial system are located in the ventral posteromedial nucleus
(VPM). The highest level of projection of sensory pathways is the postcentral gyrus, or primary somatosensory cortex (SI). The pattern of projection of sensory inputs from various regions of the body on SI is represented by the
Penfield’s “homunculus” (Figure 1.0.8) (Junge, 1998).
Cortical Mechanisms of Tactile Stimulus Processing
Tactile information is processed at different stages across the neuraxis, starting from the peripheral receptors and culminating in the cerebral cortex.
Indeed, it is in the cortex that the higher functions of the nervous system come into play. It is only in the cerebral cortex of mammals that the various higher centers associated with all the sense organs are found. Thus, the cerebral cortex of mammals contains visual, auditory, somatosensory, olfactory, and gustatory areas that are involved in gathering and processing stimulus information from the corresponding sensory systems into sensations. Flechsig
(quoted by Cajal 1995) suggested that, in addition to these sensory areas, mammals with folded cortex have intercalated cortical areas involved in associating and combining in countless ways sensory “residues” from the various cortical sensory areas to produce highly complex psychological experiences.
12
Flechsig’s theory implies two roles of the projection from the sense organs to the cortex: one corresponding to the initial projection to a specific sensory area, and another, higher order one, related to cortical integration pathways. It is this latter association cortex that is in the better position to process the conscious information performed first by the sensory cortex (Cajal, 1995). Remarkably, very similar views were prevalent in the 1970’s based on the anatomical work of cortical connectivity by Jones and Powell (Jones and Powell, 1970).
The cerebral cortex consists of two superimposed formations: the gray matter which is a highly vascular layer on the very surface of the brain, just deep to the pia mater, and the white matter, which is considerably thicker and lies between the layer of the gray matter on the one hand and the ventricles and corpus striatum on the other. The cortex is smooth in lower vertebrates, as well as in small mammals, whereas it is folded in large mammals, especially primates, with gyri separated by sulci. The organization of cerebral cortex is basically the same, be it smooth or convoluted (Cajal, 1995).
The cerebral cortex, approximately 2.5 sq. ft. in area, contains ~25 billion neurons, interconnected by more than 100,000 km of axons and making an incredible ~3 x 1014 synapses (Nolte, 2002).
13
Cortical Layers
In the past, the cerebral neocortex was divided into seven layers, named as follows, from the surface to the white matter: 1) plexiform layer, 2) small pyramidal cell layer, 3 and 4) medium and large superficial cell layers, 5) granular layer or dwarf pyramidal cell layer, 6) large deep pyramidal cell layer, 7) medium- sized deep pyramidal cell and fusiform cell layer (Cajal, 1995). It was not until
Vladimir Betz (1874) described the giant pyramidal cells in layer 5 of Meynert in the motor cortex that Bevan Lewis (1878) confirmed and suggested a division of the cortex in six layers (Lorente de No, 1949).
The cells of the neocortex are arranged in a series of six layers, more apparent in some areas than in others. Most superficial is the cell-poor molecular layer, and the deepest is the polymorphic (or multiform) layer with predominantly fusiform-shaped modified pyramidal cells. In between these two layers, there are four layers alternately populated mostly by small cells or mostly by large pyramidal cells. The layers are commonly designated by Roman numerals and by the following names: I) Molecular layer, II) External granular layer, III) External pyramidal layer, IV) Internal granular layer, V) Internal pyramidal layer, and VI) Multiform layer (Figures 1.1.4 and 1.1.6).
Layer I is an acellular layer called the molecular layer. It is occupied by dendrites of the cells located deeper in the cortex and axons that travel through or form connections in this layer.
14
Layer II is comprised mainly of small spherical cells called granule cells and therefore are called the external granule cell layer, also layer 2 is composed with small pyramidal cells.
Layer III is called the external pyramidal cell layer and contains a variety of cells types, many of which are pyramidal cells; the neurons located deeper in layer III are typically larger than those located more superficially.
Layer IV, similar to layer 2, it contains primarily granule cells (especially in primary sensory cortices) and is called the internal granule cell layer.
Layer V, the internal pyramidal cell layer, contains mainly pyramidal cells that are typically larger than those in layer 3.
Layer VI is a fairly heterogeneous layer of neurons and is thus called the polymorphic or multiform layer. The Martinotti cells in this layer have globular, ovoid or fusiform elements with varicose ascending and descending dendrites and no apical dendritic trunk (Cajal, 1995). Layer VI blends into the white matter that forms the deep limit of the cortex and carries axons to and from the cortex.
The neurons of the cortex have a variety of shapes and sizes. Raphael
Lorente de No, a student of Santiago Ramon y Cajal, used the Golgi method to identify more than 40 different types of cortical neurons based only on the distribution of their dendrites and axons. In general, the neurons of the cortex, as elsewhere, can be broadly defined as projection neurons and local interneurons.
Projection neurons typically have pyramidally shaped cell bodies. They are located mainly in layers 3, 5, and 6 and use the excitatory amino acid
15 glutamate as their primary neurotransmitter. Local interneurons use the inhibitory neurotransmitter GABA, constitute 20 – 25% of the neurons in the neocortex, and are located in all layers. Pyramidal cells in layer II typically project to the same hemisphere, those of layer III to the contralateral hemisphere, and those in layer V to subcortical structures, including the spinal cord. Corticothalamic projections arise from layer VI.
Cortical Cells
The pyramidal cells are the most numerous neurons of the neocortex.
They have a conical cell body from which a series of spine-stunned dendrites emerge – a long apical dendrite that leaves the “top” of each cell and ascends vertically toward the cortical surface, and a series of basal dendrites that emerge from nearer the base of the cell and spread out horizontally. Pyramidal cells range in size from 10µm in diameter to 70 – 100µm in the giant pyramidal cells
(Betz cells) of the motor cortex, which are among the largest neurons in the central nervous system (CNS). The majority of pyramidal cells have long axons that leave the cortex to reach other cortical areas or subcortical sites, where they make excitatory synapses (Figure 1.1.5). The remaining cortical neurons are known as nonpyramidal cells. Many of these cells are small (often less than
10µm), multipolar stellate (or granule) cells, but a variety of other types and sizes have been described. With a few exceptions, nonpyramidal cells have short axons that remain within the cortex, and can have an excitatory or inhibitory
16 effect. The small granule cells (excitatory) are found mostly in layer IV of the sensory cortices. Inhibitory interneurons comprise small stellate, basket cells and cells with a double bouquet that ends on the axon hillock.
The dendritic spines of pyramidal cells are preferential sites of synaptic contacts and have been the source of considerable interest and study. They are not merely a device for increasing dendritic surface area because the portions of the dendrite located between spines are sparsely populated with synaptic contacts. It has been suggested that dendritic spines may be the sites of synapses that are selectively modified as a result of learning, because small changes in the geometry of a spine could cause relatively large changes in its electrical properties and therefore in the efficacy of that synapse (Nolte, 2002).
The pyramidal cells of the cerebral cortex have relatively large apical dendrites
(of 2 µm diameter and 2 mm length) that are oriented parallel to one another and perpendicular to the cortical surface. The pyramidal cells are clustered into tightly interconnecting columns whose orientation relative to the external surface of the head varies from parallel, (i.e., tangential, columns in a sulcus to perpendicular, i.e., normal, for columns on gyrus).
Cortical Areas and Pain
The role of cortical areas involved in pain perception has been significantly advanced using positron emission tomography (PET) which measures the blood flow response to application of painful stimuli (Andrew and Craig, 2001; Casey et
17 al., 1994; Craig et al., 1996; Jones et al., 1991; Talbot et al., 1991). These studies have identified four areas with increased blood flow in response to the application of painful stimuli: SI, SII, insula, and cingulate cortex (Bagley et al.,
2006).
Primary Somatosensory Cortex (SI)
The peripheral somatosensory systems are specialized in touch, proprioception, temperature, and pain. Somatosensory information is transmitted through the spinal cord and the thalamus to the contralateral SI; tactile and proprioceptive information are transmitted through the medial lemniscus, and temperature and pain through the spinothalamic tract. Afferent pathways also reach directly (from the thalamus) SII bilaterally and the contralateral posterior parietal cortex (PPC) (Kass, 1990), although the major input to these areas is from SI.
In humans, SI is located in the anterior parietal cortex, on the posterior bank of the central sulcus and the postcentral gyrus. It consists of several cytoarchitectonic areas, including (from anterior to posterior) Brodmann’s areas
3a, 3b, 1, and 2, which are all interconnected. (Hyvarinen, 1982). SI is the cortical target of the majority of subnucleus caudalis neurons (Van Buren and
Borke, 1972).
The SI cortex is activated by innocuous and noxious stimulation in humans, as demonstrated by MEG, evoked potentials, PET, and most recently,
18 fMRI studies (Bushnell et al., 1999). Study have demonstrated that rCBF was
15 measured using PET H2 O, the SI is significantly augmented by increases in the magnitude of perceived pain evoked by greater stimulus amplitude (Han et al., 1998), or by hypnotic suggestion (Bushnell et al., 1999; Hofbauer et al.,
2001). This capacity to encode the intensity of painful stimuli supports the notion that the SI is involved in the sensory-discriminative aspect of pain.
Finally, the human PPC, also known as the parietal association area, is situated in the post-central sulcus, medial and posterior to the SI hand area.
The PPC integrates sensory and motor processing, combines tactile and proprioceptive input with other sensory modalities, and is associated with higher cognitive functions (Hyvarinen, 1982).
Secondary Somatosensory Cortex (SII)
The human SII is located in the parietal operculum, immediately posterior to the most lateral aspect of SI. SII receives inputs from both the ventral posterior thalamus and SI (Burton, 1986). SII is activated bilaterally by a unilateral tactile stimulus (Jousmaki, 2000). Surrounding SII, are several other electrophysiologically defined somatotopic maps, which have been identified as
PV (parietal ventral, just anterior to the SII in the parietal operculum) (Krubitzer et al., 1995), area 7b (lateral to the SII) (Craig, 2003a; b), posterior insula (medial to the SII) (Craig, 1995), and retroinsula (in the lateral fissure posterior to the insula
19 proper (Burton, 1986). Together, these areas are identified as parasylvian cortical areas.
Neurophysiological studies in primates have identified cells responsive to painful stimuli in parasylvian areas SII and area 7b (Burton, 1986; Dong et al.,
1989; Dong et al., 1994; Robinson and Burton, 1980). Human studies suggest that the subnuclei in and around the trigeminal subnucleus caudalis project to the posterior parietal operculum (including the SI); (Van Buren and Borke, 1972).
Human laser evoked potential (LEP) studies also provide evidence of nociceptive inputs to this area (Zaslansky et al., 1996).
Imaging studies have reported blood flow -related responses in SII and other parasylvian cortical areas (Davis, 2000; Rainville et al., 2000). Recent fMRI studies have demonstrated multiple loci of activation in the parietal operculum– posterior insula region with both innocuous and noxious stimuli (Davis, 2000).
There is evidence from human stimulation studies that insula, notably its posterior superior portion, is involved in pain-related processes (Bagley et al.,
2006).
Magnetoencephalography (MEG)
MEG has been developed over the last 30 years as an attempt to better trace and analyze magnetic currents generated by the brain. The MEG signal is the magnetic analog of the electroencephalographic (Steeghs et al., 2002) signal and apparently derives from the same source, namely the postsynaptic current
20 elicited in aligned dendrites of pyramidal cells.(Figure 1.1.7). The brain activity as induction of magnetic fields is modeled as a current dipole following the right- hand rule. The dipole is defined by a positive and negative charge, with a primary current flowing in a straight line between these charges and return currents flowing in a dipolar pattern through the conductive medium. The exact orientation and configuration of currents associated with dendritic activity depend on the excitatory, versus inhibitory, nature of the relevant synapse and its location along the dendritic tree (Lewine and Orrison, 1995). In contrast to the
EEG where signals are distorted by the conductivity pattern of the head, magnetic signals pass largely unaffected through the bone allowing for better and more accurate signal localization (Lewine and Orrison, 1995). Usually the current dipole moments required to explain the measured magnetic field outside the head are on the order of 10 nA. It has been suggested that approximately one million synchronous synapses are necessary to produce a detectable magnetic field (Hamalainen et al., 1993; Lewine and Orrison, 1995). Given that there are approximately 100, 000 pyramidal cells/mm2 of cortex, each with numerous synapses, as few as one synapse per thousand must be active to produce a magnetic flux that can be detected by MEG. The neuromagnetic field recorded outside of the head mostly reflects the dendritic currents of pyramidal cells that are oriented parallel to the skull surface (Lewine and Orrison, 1995). In addition, the morphology of individual cells plays a role in the magnetic field contribution. For example, stellate cells possess mostly symmetric dendritic
21 configurations, and, as a consequence, dendritic activation produces electrical potential variations only within the region of the dendritic field, without potential changes at a distance; in other words, these cells have a closed field configuration (Lorente de No, 1947). In contrast, pyramidal cells have asymmetric dendritic tree, typically characterized by an apical dendrite and multiple basilar dendrites. The synaptic activation of the apical dendrite causes electrical potential variations throughout the extracelular medium (open field configuration) (Figure 1.1.1). Pyramidal neurons constitute almost 70% of the neocortical neurons, and the current flow in the apical dendrites of these cells is believed to be the main source of extracranial neuromagnetic signals (Lewine and Orrison, 1995).
Biomagnetic fields directly reflect electrophysiologic events of the heart and the brain. The most salient property of these fields is that neuromagnetic signals penetrate the skull and scalp without significant distortion. MEG provides the best temporal resolution (in order of ms or less) and good spatial resolution
(in order of mm) depending on the strength of the source (Lewine and Orrison,
1995). MEG provides for real-time direct assessment of dynamic brain function.
In order to consider the magnetic fields associated with neuronal currents and how these fields are detected, it is useful to model the intracellular flow of current in a neuron as a current dipole, that is, a current flowing over a short distance.
The direction and strength of the dipole is expressed by a vector quantity called the dipole moment (Q). The vector points in the direction of the current flow and
22 has a magnitude equal to the product of the strength of the current (I) in amperes and the length of the current flow (ΔI) in meters; Q = I ΔI, nanoamperes meter
(Neuroimaging).
A current dipole generates a magnetic field that encircles the dipole just as a magnetic field encircles a short segment of a current-carrying wire. The direction of the field around the dipole is dependent on the direction of the current flow within the dipole, and the directional relationship follows the right hand rule.
The strength of the field depends on the magnitude of the dipole moment, Q
(Neuroimaging).
The magnetic fields that develop are extremely weak, in the order of femtotesla (1x10-15 Tesla) (Figure 1.1.0) and are measured by special sensors called SQUIDs (Superconducting Quantum Interference Device). SQUIDs convert the magnetic flux to voltage at superconducting, very low temperatures,
~270°C. The SQUIDS are located inside a dewar that is filled with liquid helium which maintains a superconducting environment (Lewine and Orrison, 1995)
(Figure 1.0.7). Because of the extremely weak magnetic signals, MEG experiments are conducted in specially designed, magnetically shielded rooms
(Figure 1.0.6) (Hamalainen et al., 1993; Lewine and Orrison, 1995). SQUIDs can be configured as raw magnetometers, which record the whole magnetic field, or as gradiometers, which measure magnetic field gradients (i.e. difference in magnetic field strength) by using oppositely wrapped coils (Figures 1.1.1 and
1.1.2). Thus, constant environmental magnetic fields cancel out. Planar
23 gradiometers align the opposing coils in a plane whereas axial gradiometers do so in a radial orientation. Gradiometers are generally much less noisy than raw magnetometers.
As mentioned above, the magnetic fields are defined as current dipoles and the localization of these dipoles is the ultimate task for MEG. This source localization depends on different assumptions such as the presence of single or multiple sources, and head anatomy. Head coordinates can be individually standardized by using special magnetic markers at specific fiducials such as the ears, eyes, nasion and inion. These positions can be localized in an individual head MRI. Regarding the localization of the dipole and its orientation, certain assumptions have to be postulated, such as the number of generators, the cortical areas involved, and their depth and point source current representation.
Since multiple dipoles develop during a task, multiple brain responses are averaged together to increase the signal to noise ratio (SNR).
Multiple individual dipole locations are tested using the least squares fitting algorithm to find the single source location that best explains the data at hand. In this analysis, cortical activity is represented as a point source of current
(Current Source Dipole or CSD). The accuracy of CSD has been examined by using various techniques based on spherical as well as realistic head geometries. It has been found that accuracy between 2 to 4 mm is possible for the localization of strong, shallow sources, indicating that MEG possesses excellent spatial resolution for sources of adequate magnetic field strength.
24
However, one of the greatest advantages of MEG is the very high temporal resolution that is obtained, in the order of milliseconds, far surpassing the temporal resolution of other functional neuroimaging techniques such as fMRI and PET.
Forward and Inverse Problems
The extraction of the MEG signal from its sources (i.e. the electrical properties of their biological environment) and the configuration of the measuring devices is known as the forward problem. The estimation of the generator strength, location, and time course from the MEG (or EEG) signal is known as the inverse problem (Ioannides, 2006).
The forward problem is linear and has a unique solution. The electric and magnetic field generated by any combination of instantaneous current elements is uniquely defined for each current element, and the total is simply the sum of individual contributions, one from each element. In the example of continuous primary current density, the instantaneous electric and magnetic field can be computed by integrating the contributions from each-volume element in the source space. The source space for MEG includes only regions in which gray and possibly white matter exist; any intervening regions and boundaries are not part of the source space as long as they do not generate currents (Ioannides,
2006).
25
The inverse problem has no unique solution. A solution to the inverse problem can be obtained as a probabilistic estimate under suitable constraints.
Until recently, the most popular constraint reduced the generators to one or more point-like sources, or current dipoles. These points like sources are interpreted as representative of their neighborhood and are referred to as equivalent current dipoles. Another constraint assumes that the continuous current density can be written as a linear sum of functions, each defining the sensitivity profile, or lead field, of sensors. The lead fields decay fast away from each sensor, so a compensating weight function usually is used to ameliorate biases toward superficial sources. These methods lead to a solution with smallest length and are known as minimum norm (MN) or weighted minimum norm (wMN) solutions
(Ioannides, 2006).
The Questions
The main goal of this study was to investigate, using MEG, the dynamic neural mechanisms underlying facial tactile stimulation in two groups of subjects, namely a control group (without pain) and a TMD pain group (arthromyalgia), by stimulating the facial skin with a non-painful air-driven plastic membrane. Our first specific aim was to investigate and compare the spatial and temporal features of the ECDs following innocuous tactile stimuli in both groups. Andt he second specific aim was to investigate the differences in dynamic brain function between these two groups using a time-frequency analysis of the MEG data.
26
Figures
Figure 1.0.1: Temporomandibular joint (TMJ) anatomy, showing the condyle- disc relationship with associated structures in the open and closed mouth positions. (http://www.is.wayne.edu/mnissani/bruxnet/Image47.jpg).
27
Figure 1.0.2: Closer view from the anatomy a normal TMJ and its structures. From TMD patient brochure (www.aaop.net).
28
Figure 1.0.3: TMJ disc-displacement and its structures. From TMD patient brochure (www.aaop.net)
29
Figure 1.0.4: TMJ osteoarthritis. Notice the degeneration of the disc and related bone structure. From TMD patient brochure (www.aaop.net).
30
Figure 1.0.5: Major somatosensory pathway from the mouth and face. Trigeminal (V) primary afferents have their cell bodies in the trigeminal ganglion and project to second-order neurons in the V brainstem complex. These neurons may project to neurons in higher levels of the brain (e.g. in the thalamus) or in the brainstem regions such as cranial nerve motor nuclei or the reticular formation (RF) (Sessle, 2000).
31
Figure 1.0.6: Homunculus. The body representation in the somatosensory cortex: A) Somatosensory cortex in right cerebral hemisphere; B) Motor cortex in right cerebral hemisphere. From http://www.harmonicresolution.com/Sensory%20Homunculus.htm.
32
Figure 1.0.7: Here is a cross section revealing the six layers of the cerebral cortex, from layer I at the top, to layer VI at the bottom of the image. Note the large layer V and VI pyramidal neurons (in red), with their apical shafts ascending to layer I. The inhibitory fibers (in blue) wrap around these apical shafts in order to control the level of excitability in the cortex (thereby preventing seizures). (From the Blue project: http://bluebrain.epfl.ch/page18701.html).
33
Figure 1.0.8: Transverse section through the gray matter of a rat cortex; the thickness of the sheet from top bottom, is in humans about 3 mm. A prominent pyramidal neuron (H) is illustrated in the middle and two stellate cells (F) nearby. The tissue was prepared by the Golgi method, which stains about 1% of the neurons. Cortical layers with it respectively cells. (Cajal 1888).
34
Figure 1.0.9: Cortical neuron. A) Schematic illustration of a pyramidal neuron and three magnified synapses (modified from Iversen, 1979); B) pyramidal neuron, redrawn from Ramon y Cajal (Hamalainen et al., 1993).
35
Figure 1.1.0: Networks of cortical neural cell assemblies are the main generators of MEG/EEG signals. Left: Excitatory postsynaptic potentials (EPSPs) are generated at the apical dendritic tree of a cortical pyramidal cell and trigger the generation of a current that flows through the volume conductor from the non- excited membrane of the soma and basal dendrites to the apical dendritic tree sustaining the EPSPs. Some of the current takes the shortest route between the source and the sink by traveling within the dendritic trunk (primary current in blue), while conservation of electric charges imposes that the current loop be closed with extracellular currents flowing even through the most distant part of the volume conductor (secondary currents in red). Center: Large cortical pyramidal nerve cells are organized in macro-assemblies with their dendrites normally oriented to the local cortical surface. This spatial arrangement and the simultaneous activation of a large population of these cells contribute to the spatio-temporal superposition of the elemental activity of every cell, resulting in a current flow that generates detectable EEG and MEG signals. Right: Functional networks made of these cortical cell assemblies and distributed at possibly multiple brain locations are thus the putative main generators of MEG and EEG signals (Baillet et al., 2001).
36
(a (b ) )
Figure 1.1.1: Schematic drawing showing (a) open field cell configuration, and (b) close field cell configuration.
37
Figure 1.1.2: Comparison of selected biomagnetic fields and environmental disturbances (Vrba, 1996. SFT Systems, Inc).
38
Figure 1.1.3: Closer view from MEG unit at Brain Sciences Center. This unit has 248 sensors (axial gradiometers).
39
Figure 1.1.4: The magnetic shielded room and the MEG machine inside located in our facility at the Brain Sciences Center.
40
Figure 1.1.5: Examples of flux transformers. (a) magnetometer, (b) 1st-order gradiometer, (c) 2nd-order gradiometer, (d) 3rd-order gradiometer (Vrba, 1996. SFT Systems, Inc).
41
Figure 1.1.6: Structure of the axial gradiometer. (a) Schematic diagram of the gradiometer and (b) photograph of the fabricated gradiometer (Lee et al., 2009).
42
Figure 1.1.7: Layout of a helmet-type 128-channel axial gradiometer array (Lee et al., 2009).
43
Figure 1.1.8: Overview of the MEG signal processing chain. The MEG signals originate in the brain neurons. Activation of the individual neurons is not detectable and only the collective activations of large number of neurons are detected by the primary SQUID sensors (Section 1). In addition to the brain signals, the SQUID sensors are also exposed to the environmental and body noise. To eliminate the environmental noise, references sensors, positioned farther from the scalp, are often used. The reference signals are subtracted from the primary sensor outputs to reduce the detected noise; the process can be understood as spatial high pass filtering. After the noise reduction, the detected signals are processed to the required bandwidth and the data are acquired. The acquired data represent magnetic field on the scalp surface and must be interpreted to yield information about the brain sources. This process requires additional information about the anatomical structure, forward models of the brain sources, and methods for source estimation from the measured fields. The brain magnetic fields were generated by a specific distribution of the neuronal currents as shown in the upper left side of this figure. After the measurement, processing, and interpretation, a smoothed estimate of the neuronal activity is obtained, as shown in the lower right side of the figure (Vrba and Robinson, 2001).
44
Figure 1.1.9: Types of brain imaging techniques used in the study of pain MEG and EEG provide the most direct measures of neuronal activity, whereas the two most commonly used techniques (PET and fMRI) measure an indirect haemodynamic response (Kupers and Kehlet, 2006).
45
CHAPTER 2: MAGNETIC SOURCE IMAGING
BACKGROUND
Functional Neuroimaging Studies of Orofacial Pain
Studies using somatosensory stimulation in human subjects showed that orofacial structures are represented on the lateral inferior primary somatosensory cortex (Baumgartner et al., 1992; McCarthy et al., 1993; Penfield and
Rasmussen, 1950; Picard and Olivier, 1983). In addition, several studies have been performed to map and explore cortical function in orofacial pain using various functional neuroimaging techniques, including positron emission tomography (PET), functional magnetic resonance imaging (fMRI), diffusion tensor imaging (DTI), electroencephalography (Steeghs et al., 2002) and magnetoencephalography (MEG), as follows.
PET
Derbyshire et al. (1994) studied rCBF response in patients suffering from chronic atypical facial pain and healthy control subjects following delivery of an innocuous (i.e. non-painful) and painful thermal stimuli. Both groups showed significant differences in rCBF response to painful vs. non-painful heat in the thalamus, anterior cingulate cortex (area 24), lentiform nucleus, insula and prefrontal cortex. Increased rCBF was found in the anterior cingulate cortex and decreased in the prefrontal cortex in the atypical facial pain group, documenting
46 the importance of the cingulate cortex and its reciprocal connections with the prefrontal cortex in processing of pain (Derbyshire et al., 1994).
Kupers et al. (2000) measured the regional cerebral blood flow (rCBF) in chronic orofacial pain patients. They found significantly increased rCBF in the prefrontal cortex, anterior cingulate cortex, anterior insular cortex, hypothalamus and periaqueductal gray matter. In contrast, rCBF was decreased in the substantia nigra, red nucleus and anterior pulvinar. During thalamic stimulation, rCBF increased in the amygdala and anterior insular cortex (Kupers et al., 2000).
Kupers et al. (2004) also investigated the cerebral activation patterns elicited by experimental jaw-muscle pain induced by injections of 5% hypertonic saline into the right masseter muscle. This led to rCBF increase in the dorsal-posterior insula (bilaterally), anterior cingulate and prefrontal cortices, right posterior parietal cortices, brainstem, and cerebellum. No changes were found in SI and
SII cortices. In contrast, stimulation by von Frey hairs increased rCBF in the contralateral face representation in the SI (Kupers et al., 2004).
The studies reviewed above were based on rCBF measurements using
PET. A different study, also using PET, was carried out to examine the striatal dopamine D1 and D2 receptors in patients with burning mouth syndrome (BMS).
The results showed that the ligand11C-NNC 756 did not differ between controls and patients and the 11C-raclopride was statistically significant higher in the left putamen in the patients with BMS. After analyzing the region of interest they found low ratio in D1/D2 (lower in the right and left putamen) when compared to
47 controls suggesting that increased 11C-raclopride uptake and the decrease in the D1/D2 ratio may indicate a decline in endogenous dopamine levels in the putamen in these patients (Hagelberg et al., 2003).
fMRI
Several fMRI studies have attempted to evaluate brain function in patients with trigeminal pain. Borsook et al. (2003) assessed the activation of the trigeminal ganglion using innocuous mechanical (brush) stimuli and noxious thermal stimuli applied to the face. Changes in BOLD signal were observed in the ipsilateral ganglion with either stimulus type; specifically, the signal decreased after brush stimuli and increased after noxious stimuli, suggesting the ability of fMRI to differentiate between innocuous and noxious stimuli (Borsook et al., 2003).
In a study of cortical activation, de Leeuw et al. (2006) found that a painfully hot stimulation of the skin overlying the left masseter muscle activated discrete brain regions within the insula, cingulate gyrus, thalamus, inferior parietal lobe/postcentral gyrus, right middle and inferior frontal gyri, cuneus, precuneus, and precentral gyrus. This pattern of activation resembles that involved in pain processing of painful peripheral stimuli. (de Leeuw et al., 2006).
In contrast, electrically-induced tooth pain activated SI, SII and insula bilaterally
(Jantsch et al., 2005), a distinctly different activation pattern from that elicited by painful stimulation of the hand.
48
Nash and colleagues (2007) investigated the cortical and subcortical patterns of activation elicited by experimentally induced jaw-muscle pain in the right masseter muscle. The results revealed significant activation in the spinal trigeminal nucleus (the SI relay nucleus serving the orofacial region) as well as in the left (contralateral) SI and posterior insular cortex, and the right (ipsilateral) anterior insular cortex. In contrast, the perigenual cingulate cortex showed significantly decreased signal intensity (Nash et al., 2007).
Blatow et al. (2009) tested whether pain is associated with an alteration in somatosensory function in patients with trigeminal neuralgia (TN) using fMRI. A tactile nonpainful stimulus was delivered in the lips and fingers. There were three groups, as follows. The first group consisted of patients with TN, the second group consisted of patients TN relieved after surgery, and the third group consisted of healthy control subjects. It was found that the activation of SI and
SII did not differ significantly from that of operated or non-operated groups. On the other hand, TN patients showed a significant reduction in SI and SII activation, as compared to controls, and these differences were more prominent during stimulation of the fingers. With respect to lip stimulation, SI and SII showed a reduced activation on the contra-lateral but not on the ipsilateral side to the stimulus application. This reduced activation probably reflects reduced neural processing, possibly, due perhaps to long term adaptation to painful stimuli
(Blatow et al., 2009).
49
Jiang et al. (2006) compared the CNS activation patterns after evoked clenching, among three groups of subjects, namely patients with atypical facial pain, patients who had occlusal treatment for synovitis pain of TMD, and normal subjects. They found that different regions were activated between the two patient groups. Specifically, the atypical facial pain activated the thalamus and anterior cingulate cortex, whereas the synovitis pain of TMD activated SI and prefrontal cortex. It was concluded that the pain network is different between the two pain groups and that the pain network is more sensitive in patients with atypical pain (Jiang et al., 2006).
Albuquerque et al. (2006) studied patients with burning mouth disorder
(BMD) and normal controls after thermal stimulation of the right masseter. The
BMD group showed greater fractional signal changes in the right anterior cingulate cortex (areas 32/24) and the precuneus (bilaterally). The control group showed larger signal changes in the thalamus bilaterally, the right middle frontal gyrus, the right precentral gyrus, the left lingual gyrus, and the cerebellum. In general, BMD patients showed brain activations patterns similar to those observed in patients with other neuropathic pain conditions and appear to process painful trigeminal stimulation qualitatively and quantitatively differently than normal. These results suggest that brain hypoactivity may be an important feature in the pathophysiology of BMD (Albuquerque et al., 2006). The same authors reviewed the supraspinal mechanisms in pain derived from neuroimaging studies and concluded that chronic orofacial pain conditions may be related to a
50 dysfunctional brain network and may involve a deficient descending inhibitory system. The somatosensory cortices, anterior cingulate cortex, thalamus, and prefrontal cortex may play a vital role in the pathophysiology of chronic pain and should be the main focus of future neuroimaging studies in chronic pain patients
(de Leeuw et al., 2005).
In patients with chronic trigeminal neuropathic pain (right maxillary), changes in cortical thickness was found, including both cortical thickening and thinning in sensorimotor regions (SI and SII precentral gyrus), and predominantly thinning in other regions (dorsolateral prefrontal cortex, frontal cortex, anterior insula, and cingulate cortex). The patterns of alteration in cortical thickness suggest a dynamic functionally-driven plasticity of the brain. Interesting these structural changes, which correlated with the pain duration, age-at-onset, pain intensity and cortical activity, may be specific targets for evaluating therapeutic interventions (DaSilva et al., 2008), were frequently co-localized and correlated with functional allodynic SI and SII activations documented by fMRI.
Finally, investigation of the cerebral activation in cluster headache before, during and after an attack, showed significant hypothalamic activation of the hypothalamus ipsilateral to the pain side (DaSilva et al., 2007; Morelli et al.,
2008).
51
MEG
Several studies have been performed using MEG to localize trigeminal structures in SI and SII. Equivalent current dipoles (ECDs) were extracted and localized after tongue protrusion. It was found that the ECDs were localized in the central sulcus anterior and inferior to locus activated by electrical stimulation of the median nerve (Nakasato et al., 2001; Sakamoto et al., 2008).
Bessho et al. (2007) found peak latencies of 15, 65 and 125 ms after an electrical stimulation in three sites of the hard palatal area. For the early magnetic field, an ECD was found along the posterior wall of the inferior part of the central sulcus, documenting the ability of MEG to localize the palatal area in
SI, even the area being small (Bessho et al., 2007).
Nevalainen et al. (2006) used innocuous tactile stimulation to localize facial areas such as the lower lip, cheek, chin, and forehead. Contralateral responses were evoked after lip stimulation, right (91%) and left after cheek stimulation (64%), right (73%) and left after chin stimulation (82%), and right
(64%) and left (27%) after forehead stimulation. The authors concluded that tactile stimulation of the lip area activated the contralateral SI in normal subjects, but for the other trigeminal areas the response was lower (Nevalainen et al.,
2006). Somatosensory evoked magnetic fields were studied after air-pressure tactile stimuli were applied to the scalp in normal human subjects. The result
found that tactile stimulation of the scalp activated contralateral SI, followed by bilateral activation of SII (Hoshiyama et al., 1995).
52
The face representation in the SI and SII was investigated after tactile stimulation of six points on the face, lower lip and thumb (Nguyen et al., 2004;
Nguyen et al., 2005). It was found that the lips occupy a large area of the face representation in SI, whereas only a small area is devoted to the skin of the face.
Regarding SII, the lip (and the face close to the lip) showed a clear representation in the lateral region of SII (Nguyen et al., 2004; Nguyen et al.,
2005).
Electrical stimulation of the lip was also used to elucidate cortical responses in normal subjects. A contralateral SI response was detected at a mean latency of 14.6 ms, and the ECD localized at the posterior bank of the central sulcus with anterior-superior orientation and inferior to the location of the
ECD elicited by median nerve stimulation (Nagamatsu et al., 2001).
A quantitative sensory test (QST) was performed in subjects with idiopathic facial pain and the results showed the QST was not altered in patients with idiopathic facial pain that showed abnormal dishabituation of the R2 component of the blink reflex. Also the distance between the cortical representations of the lip and index finger did not differ between patients and controls, suggesting that idiopathic facial pain is maintained by mechanisms which may not involve somatosensory processing of stimuli from the pain area
(Lang et al., 2005).
Finally, studies of cranial nerves in normal subjects, MEG has mostly been used for its localization accuracy to map the cortical representation areas of
53 different facial and oral structures (Karhu et al., 1991; Nakahara et al., 2004;
Nakamura et al., 1998; Nguyen et al., 2004; Yang et al., 1993).
In summary,
Based on the information above, several imaging techniques have been used to map not only the cortical function but also the subcortical regions and their respective function after certain type of stimulation, and to demonstrate the ability of the imaging techniques to measure their spatial and temporal aspects.
In this study we used a process of measuring the magnetic sources outside the head known as magnetoencephalography (Hamalainen et al., 1993).
We investigated the dynamic neural mechanisms underlying facial tactile sensibility by means of MEG and evaluated the interactions of cortical areas during facial tactile stimulation in two different groups; a control group (no pain) and an TMD group (arthromyalgia), by bilaterally stimulating the facial skin with a non-painful air-driven membrane.
MEG detects the weak extracranial magnetic fields produced by currents generated in the aligned pyramidal cells in the cerebral cortex. This technique is unique in a way that these currents are not affected by the differences in conductivity between the active brain source and the measuring device
(Hamalainen et al., 1993).
54
METHODS
Subjects
Eighteen subjects were studied as paid volunteers. They were divided into two groups, one without pain (n = 10; 7 women and 3 men; mean age = 30.8
± 11.1 y) and the other suffering from TMD pain (n =8; 7 women and 1 man; age
= 37.5 ± 11.6y). The subjects from the TMD pain group participated in a previous study for Research Diagnostic Criteria (RDC) for TMD and already have a established diagnosis for TMD according to RBC/TMD guidelines. Inclusion criteria included subject age from 18 to 65 years old, with history of TMD pain.
Exclusion criteria include subjects younger than 18 years of age, subjects with implanted metal devices that could affect the MEG acquisition. For the group without pain, inclusion criteria included no history of orofacial pain, or any other type of pain the occurs on a regular basis. The study protocol was approved by the appropriate institutional review boards and informed consent was obtained from all subjects before the study according to the Declaration of Helsinki.
Stimulus delivery
Subjects rested in a supine position with their eyes open. A circular elastic membrane, 1 cm in diameter, was attached to each side of the face on the top of the masseter muscle (Figure 2.0.1). The motion of each membrane was triggered by a computer-generated air pressure stimulus of 17 pounds per square inch (psi). Two hundred non-painful tactile stimuli of 50 ms duration each
55 were delivered with an interstimulus interval of 1-2 s. There was a delay of 41 ms from the onset of the trigger to the application of the stimulus (Biermann et al., 1998). The stimulus were not painful in any of the groups per subject report.
Data collection
MEG data were collected using a 248-channel axial gradiometer system
(Magnes 3600WH; 4D-Neuroimaging, San Diego, USA). The cryogenic helmet- shaped dewar of the MEG was located within an electromagnetically shielded room to reduce noise. Data (0.1–400 Hz) were collected at 1017.25 Hz. For anatomical source localization, T1-weighted gradient-echo images of structural
MRI for each subject were acquired.
Data preprocessing and dipole extraction
The acquired time-series data were corrected for the cardiac artifact using event synchronous subtraction (Leuthold, 2003). The integrated BESA (v. 5.06,
MEGIS Software GmbH, Gräfelfing, Germany), BrainVoyager (Electrical
Geodesics, Inc, Eugene, OR, USA) package was used for single equivalent current dipole (ECD) analysis (Figure 2.0.2). The ECD method, is widely used because it very easy to apply and accurate for most problems (Uutela et al.,
1999; Wolters et al., 1999). This method assumes dipolar current sources to approximate the flow of electrical current in a small brain area. Periods of noisy sensors and eyeblink artifacts were manually detected and eliminated from
56 further analysis. Subsequently, the BESA artifact rejection tool was utilized for more precise artifact characterization (rejecting trials and channels with a range of amplitude (2,000 fT). Epochs were defined from −100 to 1,000 ms from stimulus onset and averaged aligned to stimulus onset. The averaged file was then bandpass filtered (high-pass filter: 1 Hz second order; lowpass filter: 44 Hz fourth order; zero-phase). Possible ECD locations with dipolar distribution and correlated temporal evolution of pole intensity, were determined from derived flux maps of sensor waveforms by visual inspection up to 1,000 ms following stimulus onset. Only sensors (minimum 12) relevant to these locations were used for further dipole analysis of the respective ECD. For each subject, the Brain
Voyager software (integrated with the BESA software) was used to create a 3D reconstruction of the brain according to the acquired MRI images. Alignment between the Talairach coordinates of BESA and BrainVoyager was based on fiducials and surface points digitized during acquisition (Langheim et al., 2006).
Using the BESA algorithm for dipole source modeling, ECDs were determined for each of the flux maps and projected onto the 3D-MRI reconstructions of their cortical location (Langheim et al., 2006). Acceptance criteria for ECD localization included (a) goodness of fit of >90%, (b) maximum amplitude of 25 nA, (c) location in cortical gray matter, and (d) duration of 10 ms. In waveforms that had multiple peaks, individual peaks were analyzed separately. The time of occurrence of the peak of each and a selection of a subset of sensors allowed a more accurate characterization of each ECD, additionally avoiding interference
57 from other fields. Finally, the onset time of derived dipoles was adjusted to account for the transmission delay (41 ms) of the stimulus (air puff) through the tubes.
Assignment to brain areas
The assignment of dipoles to specific 23 cortical regions found in the study was based on Talairach coordinates (Tailarach and Tournoux, 1988) and anatomical landmarks. The following abbreviations are used below: ParOper
(parietal operculum), PoCGy (postcentral gyrus), PoCSul (postcentral sulcus),
TemOper (temporal operculum), STGy (superior temporal gyrus), Ins (insula),
PrCGy (precentral gyrus) PrCSul (precentral sulcus), MTGy (middle temporal gyrus), AngGy (angular gyrus), SMGy (supramarginal gyrus), STSul (superior temporal sulcus), LinGy (lingual gyrus), SPL (superior parietal lobule), CSul
(central sulcus), IFGy (inferior frontal gyrus), Pcun (precuneus), Cun (cuneus),
LatSul (lateral sulcus), CingGy (cingulate gyrus), IntParSul (internal parietal sulcus), ITGy (inferior temporal gyrus), OccGy (occipital gyrus), SFGy (superior frontal gyrus).
58
STATISTICAL ANALYSIS
Standard statistical analyses (analysis of variance [ANOVA], t-tests, non- parametric tests, etc.), including parametric and nonparametric tests, were performed on ECD attributes to evaluate differences between the two groups within specific brain areas. These attributes comprised ECD counts, duration and onset time. ECD counts were square-root transformed to normalize their distribution and stabilize the variance (Snedecor and Cochran, 1989).
RESULTS
ECD counts
A total of 324 ECDs were found in 23 different cortical areas in the 18 subjects studied (Figure 2.0.4). The descriptive statistics for the square-root transformed ECD counts in each area are given in Table 2.0.1. An analysis of variance (ANOVA) of the overall counts found no statistically significant difference between the two groups (P = 0.264); in addition, neither the subject gender nor age were statistically significant (P = 0.829 and 0.467, respectively).
A t-test on the dipole counts for each area separately did not reveal any statistically significant difference between the two groups (Table 2.0.2).
59
ECD duration
The descriptive statistics for the ECD in each area are given in Tables
2.0.3 and 2.0.4 for the control and TMD pain groups, respectively. An ANOVA of the ECD duration across areas yielded a highly significant difference between the two groups (P = 0.001, F-test); neither the subject gender nor age were statistically significant (P = 0.439 and 0.431, respectively). A t-test on the ECD duration for each area separately revealed a statistically significant longer duration in the TMD pain group in the following areas: PrCGy (p = 0.001), Csul (p
= 0.034), SPL (p = 0.017), ParOp (p = 0.027), Ins (p = 0.021) (Figure 2.0.5).
ECD onset time
The descriptive statistics for the ECD time onset in each area are given in
Tables 2.0.5 and 2.0.6 for the control and TMD pain groups, respectively. We used the log rank test to assess the statistical significance of differences in the onset time distributions between the two groups (Cox and Oakes, 1984).
Specifically, we focused on early onset times (up to 300 ms) and analyzed data from the following areas with 5 or more ECDs per group: Ins (12 control, 5 pain),
ParOp (22 control, 21 pain), PoCGy (15 control, 12 pain), STGy (19 control, 6 pain) and TempOp (18 control, 13 TMD pain). A statistically significant difference with earlier onset times in the TMD pain group was found only in
ParOp (P = 0.043) (Figure 2.0.6).
60
DISCUSSION
To our knowledge, this is the first time that control and TMD pain groups have been compared with respect to brain responses to a non-noxious tactile stimulus. In this study, we localized dipoles in 23 different cortical areas in both groups. The majority of the dipoles were localized in SI and SII, and the ECD counts did not differ significantly in any area between the two groups. We were able to more precisely parcellate parietal and temporal areas for more accurate localization of activation.
In our study we also analyzed the duration of the ECDs in 23 cortical areas after tactile stimulation. We found statistically significant differences between the two groups in five areas (PrCGy, CSul, SPL, ParOp, Ins), suggesting a persistent cortical activation in the TMD pain group. In addition, we found that ECDs onset time was significant different and occurred earlier in the
ParOper in the TMD pain group, and continued to differ after 50 ms. One possible explanation is that the cortical reorganization may have lowered their threshold for tactile stimuli causing this area to be more sensitive to the stimuli.
In conclusion, the most dramatic changes observed in this study was in a time domain and specific SI and SII regions. These changes possibly are due to cortical reorganization that occurred over the years of suffering from TMD pain.
61
TABLES AND FIGURES
Table 2.0.1: Descriptive statistics for the square-root transformed ECD counts in each area.
62
Table 2.0.2: P-values of t-test (dipole counts, individual areas).
AREA P-value (t-test) ParOper 0.589 PoCGy_PoCSul 0.969 TemOper 0.595 STGy 0.151 Ins 0.125 PrCGy_PrCSul 0.494 MTGy 0.827 MFGy 0.398 AngGy 0.474 SMGy 0.461 STSul 0.658 LinGy 0.555 SPL 0.832 Csul 0.426 IFGy 0.111 Pcun 0.934 Cun 0.104 LatSul 0.387 CingGy 0.387 IntParSul 0.276 ITGy 0.276 OccGy 0.387 SFGy 0.387
63
Table 2.0.3: ECD duration (ms) statistics for the control group.
64
Table 2.0.4: ECD duration (ms) statistics for the TMD pain group.
65
Figure 2.0.1: ECD duration group plot
66
Table 2.0.5: ECD onset times (ms) for the control group.
Table 2.0.5: ECD onset times (ms) for the control group.
67
Table 2.0.6: ECD onset times (ms) for the TMD pain group.
68
(a)
(b)
Figure 2.0.2: Location of the airpuff stimullators above the skin of the masseter muscle. Figure (a) left side view, red line is the stimulator on the left side. Figure (b) frontal view, bilateral view of the stimulators (red and blue lines)
69
(a )
(b )
(c )
Figure 2.0.3: Example of ECD location. Parietal Operculum in the TMD pain group. (a) sagital view, (b) coronal view, (c)transverse view
70
Figure 2.0.4: Plot showing the average number of dipoles in various brain areas in both groups (blue = control, brown = TMD pain group). There were no significant changes in the average number of dipoles between the two groups in any brain areas.
71
Figure 2.0.5: Plot showing the dipole duration in both groups with their respective p-values (blue = control, brown = TMD pain group). The TMD pain group showed statistically significant changes in the duration of the dipole in relation to the control group.
72
Figure 2.0.6: Plot showing the early ECD onset time for the parietal operculum. Notice that at 100 ms, approximately 65% of the stimulus was already processed in the TMD pain group (brown), and only 45% in the control group (blue).
73
CHAPTER 3: TIME-FREQUENCY ANALYSIS
BACKGROUND
Time-frequency analysis (TFA) identifies the time at which various frequencies are present, by calculating a spectrum at regular intervals of time
(WaveMetrics.com). A time-frequency representation (TFR) is a view of a signal
(taken to be a function of time) represented over both time and frequency. TFA means analysis of a TFR. TFRs are complex-valued fields over time and frequency, where the modulus of the field represents “energy density” (the concentration of the root mean square over time and frequency) or amplitude, and the argument of the field represents phase (http://en.wikipedia.org/wiki/Time- frequency_representation).
A signal, as a function of time, may be considered as a representation with perfect time resolution. In contrast, the magnitude of the Fourier transform of the signal may be considered as a representation with perfect spectral resolution but with no time information because the magnitude of the FT conveys frequency content but it fails to convey when, in time, different events occur in the signal.TFRs provide a bridge between these two representations (time x frequency). They provide some temporal information and some spectral information simultaneously. Thus, TFRs are useful for the analysis of signals containing multiple time-varying frequencies (http://en.wikipedia.org/wiki/Time- frequency_representation).
74
Oscillatory activity is a hallmark of neuronal network function in various brain regions, including the olfactory bulb thalamus, hippocampus and neocortex.
The frequency of network oscillations covers more than three orders of magnitude, from slow oscillations in delta (0.5 -3 Hz) and theta (3-8Hz) to fast oscillations in the gamma (30 – 90Hz) and ultrafast (90-200Hz) ranges (Buzsaki and Draguhn, 2004).
The evoked gamma-band activity is an event-related rhythmic response which persists within the first 100 ms after the stimulus onset. It shows spectral peaks between 30 and 40 Hz in the auditory, between 45 and 55 Hz in the somatosensory, and between 100 and 110 Hz in the visual system (Pantev,
1995).
TFA has been applied to EEG (Ray et al., 2008; Steeghs et al., 2002;
Tierra-Criollo and Infantosi, 2006) and MEG data (Boonstra et al., 2007; Cimatti et al., 2007; Hauck et al., 2007; Jaros et al., 2008; Nikouline et al., 2000; Riggs et al., 2008; Ross et al., 2005). Nikouline at al. (2000) studied the somatosensory evoked responses influenced by ongoing rhythmic activity in the 8-13 Hz frequency range originating in the sensorimotor cortex (mu rhythm) after median nerve stimulation. The results showed that N20m deflection did not depend on pre-stimulus activity, while the amplitude of the P35m deflection, and to a lesser extent that of the P60m deflection, showed a small positive correlation with the amplitude of the pre-stimulus mu rhythm. The authors concluded that the first excitatory cortical response (N20m) is independent of the oscillatory state (8-13
Hz) of the sensorimotor cortex. Later parts of the response (P35m and P60m)
75
are also relatively stable compared with the large variations in the mu rhythm
(Nikouline et al., 2000).
Hauck et al. (2007) studied the influence of direct attention on pain- induced oscillations and synchronization process using MEG in combination with an oddball paradigm. After delivering a series of electrical painful stimuli in two fingers, subjects were told to count the stimuli delivered in only one finger while ignoring the other. They found that oscillations in the beta band increased with high stimulus intensity and direct attention at the contralateral sensorimotor sites, followed by suppression and rebound of beta activity after the painful stimulation was stopped. In addition, oscillatory activity was induced in the high gamma band which increased with direct attention. It was concluded that pain-induced high-frequency activity in sensorimotor areas may reflect an attentional augmentation of processing, leading to enhanced saliency of pain-related signals and thus to more efficient processing of this information by downstream cortical centers (Hauck et al., 2007).
Cimatti et al. (2007) characterized the spectral-temporal properties of high frequency oscillatory (HFO) patterns evoked by median-nerve stimulation of the dominant hand in healthy subjects and in patients with writer's cramp. They found that HFO power was strongly reduced and disorganized in time, supporting the hypothesis that abnormal HFOs reflect pathophysiological mechanisms occurring in focal dystonia, possibly due to dysfunctional somatosensory processing (Cimatti et al., 2007).
76
Hsiao et al. (2008) studied the frequency characteristics of pain-evoked neuromagnetic responses in SII cortex. The results indicated increased theta (4-
8 Hz) and alpha (8-13 Hz) power in SII bilaterally 180-210 ms after stimulus onset (Hsiao et al., 2008).
Jaros et al. (2008) studied the influence of double stimulation on the human somatosensory system by analyzing both the low-frequency activity and the high-frequency oscillations (at about 600 Hz) during two consecutive median nerve stimulations separated by an inter-stimulus interval between 2.4 and 4.8 ms. Analyses of precortical (thalamus) dipole activation curves showed a correlation between the ISI and the latency of the response to the second stimulus in both the low- and high-frequency bands. In the second cortical (SI) response, this correlation between ISI and the latency could not be found in
HFOs, but in low frequency and HFOs they found amplitude fluctuations that were dependent on the ISI, indicating nonlinear interference between successive stimuli. In conclusion, low-frequency and HFOs have different generators and that precortical and cortical HFOs are independently generated (Jaros et al.,
2008).
Finally, Ray et al. (2008) studied the role of gamma oscillations (>30Hz) in selective attention by means of subdural electrocorticography (ECoG) in human subjects during application of auditory and tactile stimuli. They found that event- related gamma activity was greater over the auditory and somatosensory cortex when subjects attended to auditory and vibrotactile stimuli, respectively. Gamma activity was also observed over the prefrontal cortex when stimuli were applied in
77
either modality, but only when they were attended. The increase in gamma activity was greatest at frequencies between 80 and 150 Hz, in the so-called high-gamma frequency range. In conclusion, there appeared to be a strong link between activity in the high-gamma range (80-150 Hz) and selective attention
(Ray et al., 2008).
In this study, our goal is to use TFA to evaluate differences in dynamic brain function between the control and TMD pain group.
METHODS
Data Preprocessing
The MEG data come from the same experiment described in chapter 2.
The acquired time-series data were corrected for the cardiac artifact by using event synchronous subtraction (Leuthold 2003). Periods of noisy sensors and eye-blink artifacts were, initially, manually characterized and eliminated from further analysis. Subsequently, the BESA artifact rejection tool was used for more precise artifact characterization (amplitude > 2000 fT). Epochs were defined between −100 to 1000 ms from stimulus onset. Time-frequency analysis was performed per subject, condition and sensor using the BESA package (v.
5.06, MEGIS Software GmbH, Gräfelfing, Germany). A bandwidth from 4 - 400
Hz was analyzed (every 2 Hz) for a period of 1100 ms (−100 ms to 1000 ms, every 25 ms; the period of −100 to 0 ms was stimulus-free and served as the baseline). The analysis yielded normalized changes in power over the baseline at a given frequency and time bin.
78
STATISTICAL ANALYSES
Standard statistical analyses (ANOVA, t-tests, etc.) were employed to assess differences in TF between the two groups.
RESULTS
An example of a TF plot for a single sensor and a subject is shown in Fig.
3.0.1. It can be seen that changes from baseline differed for different frequencies and at different times. In general, there were statistically significant differences between the two groups. These changes were long-lasting and distributed in time, covering wide frequency range, present in individual sensors as well as in all sensors. On the average, 40% (38-44%) of the 248 sensors showed a statistically significant effect of Time x Group interaction in a given frequency band. These results are presented in more detail below.
Frequency Effects
The average changes over the baseline (across sensors and group subjects) at different frequencies (binned every 20 Hz) are shown in Fig. 3.0.2.
The Group main effect was not statistically significant but the Frequency main effect and the Group x Frequency interaction effect were highly significant (Table
3.0.1 and 3.0.2). However, as can be seen in Fig. 3.0.2, the changes from baseline for both groups were negative at the low frequencies (4-20 Hz) and positive at frequencies >20 Hz. This prompted us to perform two additional
79
ANOVAs, one for 4-20 Hz and another for >20 Hz. We found that in both cases the power changes from baseline were stronger in the TMD pain vs. the control group, i.e. more negative at 4-20 Hz and more positive at >20 Hz; these differences were statistically highly significant (Tables 3.0.1 and 3.0.2). The latter finding raises the issue of more specific differences between the two groups depending on the frequency; i.e., the presence of statistically significant
Group x Frequency interactions. Fig. 3.0.3 shows the two frequencies (21-400
Hz) for each group, superimposed. An ANOVA yielded a statistically significant interaction effect (Table 3.0.2). We then performed 19 separate comparisons between the 2 groups (independent sample t-test) for each frequency bin. We found that the 2 groups different statistically significantly in 4 frequency bins corresponding to 21-40 Hz, 61-80 Hz, 101-120 Hz and 141-160 Hz (nominal P <
0.05, adjusted for the 19 multiple comparisons by dividing the observed P value by 19; actual P < 0.05/19, or P < 0.0026) (Fig. 3.0.3.).
Time Course
The analysis above dealt with group differences in a frequency range collapsed across time bins. In this section we examine and compare the frequency changes between the 2 groups along time. We begin by plotting these
2 time courses averaged across all 248 sensors for the 4-20 Hz low frequency band (Fig. 3.0.4). A repeated measures ANOVA showed a highly statistically significant effect of Group (between subjects factor), Time (within subjects factor) and Group x Time interaction (P < 10−10 for all F-tests above). Plots from
80
individual sensors are illustrated in Figs. 3.0.5 and 3.0.6. It can be seen that the curves are similar to those of all sensors combined (Fig. 3.0.4) in that initially there is higher power in the control vs. the pain group. This was observed consistently in all sensors.
Statistically significant effects were also found for higher frequencies.
However, specific changes differed between sensors, in that initially higher power was found in either group, depending on the frequency band. Examples from individual sensors are illustrated in Figs. 3.0.7, 3.0.8 and 3.0.9 for the frequency band 140-160 Hz, and in Figs. 3.1.0, 3.1.1, and 3.1.2 for the frequency band 340-
360 Hz.
81
DISCUSSION
The results above demonstrated substantial changes in power across a wide range of frequencies (4-400 Hz) and along the whole time span of 1 s following stimulus onset. The magnitude and frequency-specific time course of these changes were distributed among the 248 MEG sensors and typically differed between the two groups. In general, greater changes in power predominated in the TMD pain group for high and ultra-high frequencies, in contrast to the control group, where power changes predominated at lower frequency bands, including theta, alpha and gamma (4 – 20 Hz).
82
FIGURES AND TABLES
TF Plot (Single Sensor)
352H z 320H z 30 B 0 297H a z Frequency (Hz) s e 20 l 0 i n 163H z 142H e 141H z z 10 0
21H 0 z
0 200 400 600 800 ms Time
Figure 3.0.1: Time-frequency plot: y-axis shows the frequency 4 - 400 Hz, x-axis shows time -100 to 1000ms. The period -100 to 0ms is designated as a baseline period. The difference in power over the baseline is shown by the colors red (more power) and blue (less power) in relation to baseline. This plot shows the difference in power over baseline at different frequency band over time.
83
Figure 3.0.2: The average changes over the baseline (across sensors and group subjects) at different frequencies (binned every 20 Hz). The Group x Frequency interaction effect was highly significant. Changes from baseline for both groups were negative at the low frequencies (4-20 Hz) and positive at frequencies >20 Hz.
84
Table 3.0.1: At low frequencies (4-20 Hz) the Group main effect was not statistically significant.
Table 3.0.2: At high frequencies (21- 400 Hz) the Frequency main effect and the Group x Frequency interaction effect were highly significant.
85
Figure 3.0.3: Shows the frequencies (21-400 Hz) for each group, superimposed. We found that the 2 groups different statistically significantly in 4 frequency bins (marked with an asterisk) corresponding to 21-40 Hz, 61-80 Hz, 101-120 Hz and 141-160 Hz (nominal P < 0.05, adjusted for the 19 multiple comparisons by dividing the observed P value by 19; actual P < 0.05/19, or P < 0.0026).
86
Figure 3.0.4: Plot of these 2 time courses averaged across all 248 sensors for the 4-20 Hz low frequency band. It can be seen that the curves are similar to those of all sensors combined in that initially there is higher power in the control vs. the pain group. This was observed consistently in all sensors.
87
Figure 3.0.5: Plot from individual sensor (sensor 10). It can be seen that initially there is higher power in the TMD pain group vs. the control, and as time progresses higher power prevail in the control group.
88
Figure 3.0.6: Plot from individual sensor (sensor 13) at lower frequency. It can be seen that the curves are similar to those of all sensors combined (Fig. 3.0.4) in that initially there is higher power in the control vs. the pain group, and this difference maintained through the 1000 ms.
89
Figure 3.0.7: Plot from individual sensor (sensor 22) at lower frequency. Initially there is higher power in the control vs. the pain group, and the difference in power was continue to change between TMD Pain group (green) and control through the 1000 ms.
90
Figure 3.0.8: Time frequency-plot showing the changes of power over the baseline over time in the two groups TMD pain (green) and control (blue), in all sensors and high frequency range (140-160Hz). Here we see consistent more power in the TMD pain group in relation to the control group throughout the 1000 ms period.
91
Figure 3.0.9: At higher frequencies (140 -160 Hz), specific sensor (sensor 2) had higher power in the TMD pain group.
92
Figure 3.1.0: In sensor 17, the changes in power over baseline differ significant in the control group during the 1000ms period.
93
Figure 3.1.1: In sensor 22 the control group had significant more power than the TMD pain group during the 1000ms period and in frequencies from 140 – 160 Hz.
94
Figure 3.1.2: Time frequency-plot showing the changes of power over the baseline over time in the two groups TMD pain (green) and control (blue), in all sensors and ultra high frequency range (340-360Hz). Here we see again consistent more power in the TMD pain group in relation to the control group.
95
Figure 3.1.3: Single sensor (sensor 12) at frequencies ranges 340 – 360 Hz statistically significant effects were also found for higher frequencies.
96
Figure 3.1.4: Single sensor at frequencies ranging from 340 -360 Hz. Statistically significant effects were also found for higher frequencies. However, specific changes differed between sensors, in that initially higher power was found in either group, depending on the frequency band.
97
Figure 3.1.5: Statistically significant effects were also found for higher frequencies (340 – 360 Hz). Here a single sensor (sensor 20), shows statistically significant more power after 75 ms, and continue to differ until 1000 ms.
98
CHAPTER 4: MULTIVARIATE ANALYSES
BACKGROUND
Multivariate analyses on the equivalent current dipoles counts (ECD) were performed to gain an insight into the organization of brain areas following stimulus delivery and assess possible differences in this organization between the control and TMD pain groups. Specifically, we used two kinds of multivariate analyses, namely multidimensional scaling (MDS) and hierarchical tree clustering
(HTC). Both methods analyze distance-like data in several variables
(dissimilarities or similarities, called proximities in general) between two items.
The goal of MDS is to detect meaningful underlying dimensions that allow the researcher to explain observed similarities or dissimilarities (distances) between the investigated objects. On the other hand, HTC creates a hierarchy of clusters which may be represented in a tree structure called a dendrogram. The root of the tree consists of a single cluster containing all observations, and the leaves correspond to individual observations. Algorithms for hierarchical clustering are generally either agglomerative, in which one starts at the leaves and successively merges clusters together; or divisive, in which one starts at the root and recursively splits the clusters. Any valid metric may be used as a measure of similarity between pairs of observations. The choice of which clusters to merge or split is determined by a linkage criteria, which is a function of the pairwise distances between observations. Cutting the tree at a given height will give a clustering at a selected precision. MDS and HTC have proved very useful 99
methods in behavioral (Whang et al., 1999), neurophysiological (Merchant et al.,
2003), and functional neuroimaging (Merchant et al., 2003; Tagaris et al., 1997;
Tagaris et al., 1998; Tzagarakis et al., 2009) studies. To our knowledge, this is the first time that these analyses have been applied to MEG data.
Whang et al. (1999) studied the representations involved in two construction-related tasks. Analyses of numerical subjective ratings and response times for these judgments showed that within the same set of geometric objects, different shape-related properties were emphasized under different task conditions. The similarity judgment depended most on a representational dimension related to enclosure of space, while the fit judgment depended to a greater extent on a dimension related to the objects' symmetry properties. This pattern of results was found in both subjective ratings and response times, as analyzed by MDS and by confirmatory classical statistics.
The findings suggest that construction-related tasks depend on representations that are context-dependent, and that MDS may be useful in a variety of settings as an intermediate-level tool for analyzing representations related to context- specific abilities (Whang et al., 1999). Tagaris et al. (1998) used MDS to analyze fMRI data obtained from four cognitive tasks, namely mental rotation of visual shapes, mental rotation of movement direction, memory scanning of visual stimuli and memory scanning of movement direction. It was found that MDS separated successfully the mental operation (mental rotation vs. memory scanning) from the operand themselves (visual shapes vs. movement direction).
100
Tzagarakis et al. (2009) used MDS to characterize the integrative neural mechanisms during viewing and subsequently copying nine geometrical shapes and conclude that this fundamental spatial motor aspect of drawing geometrical shapes is the critical variable, independent of the particular shape drawn, that dominates cortical activation during copying (Tzagarakis et al., 2009).
Tagaris et al. (1997) used HTC to analyze the organization of brain areas involved in mental rotation using fMRI data. It was found that performance measures clustered with parietal cortex activation, whereas the rate of mental rotation clustered with the right precentral gyrus. Merchant et al. (2003) analyzed the dissimilarity matrix of neuronal responses to moving visual stimuli. Tree clustering analyses showed that the [rightward, leftward], [upward, downward], and [clockwise, counterclockwise]] motions were clustered in three separate branches (i.e., horizontal, vertical, and rotatory motion, respectively). In contrast, expansion was in a lone branch, whereas contraction was also separate but within a larger cluster. The distances among these clusters were then subjected to an MDS analysis to identify the dimensions underlying the tree clustering observed. This analysis revealed two major factors in operation. The first factor separated expansion from all other stimulus motions, which seems to reflect the prominence of expansion during the common activity of locomotion. In contrast, the second factor separated planar motions from motion in depth, which suggests that the latter may hold a special place in visual motion processing
(Merchant et al., 2003).
101
In the present case, the items are the various brain areas and the data are the dipole counts for each area. MDS attempts to derive composite dimensions (of the variables) such that the corresponding linear combinations of data possess closely similar distance relations. Typically, two dimensions are used. Output measures include goodness of fit (e.g. percent of variance accounted for by the recombined data) and a plot of the variables in the derived
2-D space. In the present application, two such plots were obtained, one for each group. The potentially different placement of the areas in the two plots provided a comparison between the two groups. The HTC is complementary to
MDS. Its purpose is to derive an explicit clustering of the areas; then, a comparison of the potentially different clusters in the control and pain groups would be informative with respect to different brain organization between the two groups.
MDS
Method
The input data consisted of the square-rooted ECD counts in 23 cortical areas. The analysis was performed using the ALSCAL procedure of the SPSS statistical package (version 15) with the following parameters: the dissimilarities were Euclidean distances, the model was ordinal, the matrix was conditioned, and the solution was -2-dimensional.
102
Results
The analysis was successful, yielding an excellent goodness of it (R2 =
0.984). The areas configurations obtained for the control and pain groups are shown is shown in Fig. 4.0.1. and 4.0.2, respectively. It can be seen that areas closely involved with the stimulus processing (PoCGy_PoCSul, ParOper,
TempOper, Ins, STGy, enclosed within the blue ellipse in Fig. 4.0.1) are farther apart in the pain group (within green ellipse in Fig. 4.0.2), indicating a more divergent, and less coordinated information processing.
HTC
Method
The input data consisted of the square-rooted ECD counts in 23 cortical areas. The analysis was performed using the CLUSTER procedure of the SPSS statistical package (version 15) with the following parameters: the dissimilarities were squared Euclidean distances and the cluster method was between-groups linkage.
Results
The trees (dendrograms) obtained for the control and pain groups are shown in Fig. 4.0.3 and 4.0.4, respectively. It can be seen that the trees have similar structure in both groups, reflecting the same applied stimulus.
Specifically, there are two main branches, an upper and a lower one that subdivides in two sub-branches. The former is separate whereas the latter
103
contains partially overlapping smaller branches. Remarkably, and in spite of the overall formal similarity, these branches have different area composition, which reflects different stimulus processing in the two groups. A major difference is that area STGy has moved out of the lower branch (in the controls) to the upper branch (in the pain group), whereas the other areas in that branch
(PoCGy_PoCSul ,ParOper, TempOper, Ins) remain the same . Thus, the lower branch contains one area less in the pain group than in the control group.
Finally, there is a reorganization of the areas in the various sub-branches of the upper branch.
104
Discussion
The results of the multivariate analyses above document various changes in the internal organization of the brain areas activated by the task in the two groups. These changes involve both a less tight association in the pain group
(compared to the control group) of the areas related to the stimulus processing
(Figs. 4.0.1 and 4.0.2 - MDS), as well as differential composition of their internal clustering (Figs. 4.0.3 and 4.0.4 HTC). However, it is remarkable that those changes are seen within a pattern that is overall very similar in the two groups, as reflected in the general distributions in MDS and the general tree structure in
HTC. These findings lend to the hypothesis that (a) the application of the stimulus elicits a specific cortical activation pattern, which, however,(b) is altered in its finer structure in the pain group in a way suggesting a less coordinated information processing.
105
Figures
Figure 4.0.1: MDS plot showing the 23 cortical areas in the 2 dimensions in the control group. Notice that 5 areas distinguished from the others: PoCGy_PoCSul, ParOper, TempOper, Ins, STGy. These areas are close together suggesting a more coordinate processing stimulus.
106
4.0.2: MDS plot showing the 23 cortical areas in the 2 dimensions in the TMD pain group. Notice that 5 areas distinguished from the others: PoCGy_PoCSul, ParOper, TempOper, Ins, STGy. These areas are farther apart suggesting a less coordinate, more divergent stimulus processing information.
107
4.0.3: HTC plot showing the 23 cortical areas in the tree clustering, in the control group. Two main branches are seen here. An upper and lower that subdivide in another 2 sub-branches upper and lower. The top upper one subdivides in smaller overlapping branches and the lower ones subdivides in a couple smaller branches. Notice also the composition of the branches that differ in both groups.
108
4.0.4: HTC plot showing the 23 cortical areas in the tree clustering, in the TMD pain group. Similar composition to the control group. Two main branches are seen here. An upper and lower that subdivide in another 2 sub-branches upper and lower. The top upper one subdivides in smaller overlapping branches and the lower ones subdivides in a couple smaller branches. Notice the difference in branches composition in both groups.
109
CHAPTER 5: GENERAL DISCUSSION
In this study two groups (TMD pain vs. Control) were compared in order to establish the differences in their brain organization. It is known that TMD has an idiopathic origin. However, contributing factors are well identified, thus making the diagnosis and treatment in most cases very successful. On the other hand, in other cases of chronic TMD pain, pain continues to be a problem even after the rehabilitation process. Based on this information, we tested the strong hypothesis that in TMD the patterns of brain responses to innocuous tactile stimulation will be altered; and this was demonstrated in our results. Specifically, we observed similar activation patterns in both groups but the dynamics of the responses differed temporally and spatially in the TMD pain group. It was demonstrated that the TMD pain group had a longer tactile stimuli processing time in five areas mentioned above; and at one specific area (ParOper) activation occurred faster than the control group. These findings may lead us to hypothesize that cortical plasticity had been induced in the TMD pain group due to the chronic pain.
Previous studies using MEG have attempted to measure the cortical activity and determine the locus of activity after trigeminal nerve stimulation
(Bessho et al., 2007; Hoshiyama et al., 1995; Nagamatsu et al., 2001;
Nevalainen et al., 2006; Nguyen et al., 2004; Nguyen et al., 2005). Electrical stimulation of three different areas of the hard palate, elicited ECDs that were localized along the central sulcus and were oriented along an antero-inferior axis
(Bessho et al., 2007). Nagamatsu et al. (2001) found similar results after
110
electrical stimulation of the lower lip. In this case, the short-latency (14.6 ms)
ECD was also localized in the central sulcus, along an antero-superior axis.
The somatosensory evoked magnetic fields in SI and SII have been also investigated using tactile stimulation. Facial skin areas were stimulated by air- driven membrane in six regions of the face and the resulted ECDs were localized in SI (Nguyen et al., 2004). Subsequently, three facial skin areas were stimulated by an air-driven membrane in order to map the facial skin representation in SII.
The results showed that facial skin and lip are also represented in SII, although the representation of facial skin in SII is small (Nguyen et al., 2005). The representation of the facial skin in the cerebral cortex are located between the thumb and lip in the SI (Nguyen et al., 2004). Also, the facial area of the SI is mostly represented by the lips as the lips have a great mechanoreceptor density and plays roles in speaking and eating while the facial skin has a small representation in SI and SII (Nguyen et al., 2004). In keeping with the results of those studies, we also observed an activation of SI and SII. However, those studies localized cortical areas in a more general way (e.g. SI, SII, posterior parietal cortex) whereas we were able to more precisely parcellate parietal and temporal areas for more accurate localization of activation. Overall, these results are in agreement with previous neuroimaging studies in which it was found that
SI and SII participated in the perception of sensory features of pain (Apkarian et al., 2005; Bushnell et al., 1999).
In our study we also analyzed the duration of the ECDs in 23 cortical areas after tactile stimulation. We found statistically significant differences
111
between the two groups in five areas (PrCGy, CSul, SPL, ParOp, Ins).
Specifically, activation lasted longer than in the control group. It is possible that the TMD painful condition has altered the cortical mechanisms responsible for perceiving and processing non-noxious tactile stimuli resulting in this persistent cortical activation. This cortical plasticity was previous demonstrated in patients suffering from chronic pain by Jones (2000), in the somatosensory cortex after nerve injury, and by Tronnier et al (2001) in chronic trigeminal neuralgia patients
(Jones, 2000; Tronnier et al., 2001). In addition, we found that ECDs onset time was significant different and occurred earlier in the ParOper in the TMD pain group, and continued to differ after 50 ms. One possible explanation is that the postulated cortical reorganization may have lowered their threshold for tactile stimuli. Such changes in plasticity were also manifested as differences in the internal organization of areas involved in stimulus processing, identified using
MDS and HTC, suggesting a different architecture in processing the tactile information between these groups.
It is known that cortical reorganization or plasticity occurs in different pain conditions such as fibromyalgia and chronic neuropathic pain. In one study Wood and colleagues using voxel-based morphometry found significant reduction in the cortical areas including bilateral parahippocampal gyri, right posterior cingulate cortex, and left anterior cingulate cortex in subjects suffering from fibromyalgia
(Wood et al., 2009). In chronic pain caused by herpes simplex virus a neuromagnetic response was recorded using MEG after a finger tactile stimulation in both patients and a control. The results showed that the distances
112
between the thumb and the little finger in the contralateral painful side in the chronic pain group were smaller than the control group indicating a cortical reorganization in this chronic neuropathic pain condition (Vartiainen et al., 2009).
Flor et al. (2006) observed nerve terminals sprouting in inactive somatosensory areas in phantom limb pain, and found out that the overexcitation of the anterior cingulate cortex (ACC) and insular cortex (IC) contribute to phantom limb pain
(Flor et al., 2006).
One hypothesis for this cortical reorganization is explained in a review paper by Zhuo (2008), where is stated that periphery injuries triggers plastic changes or long term potentiation (LTP) in the cortical synapses. These plastic changes persist for a long period of time, generating abnormal neuronal activity even after the periphery injury has healed, suggesting that chronic pain utilizes highly selective synaptic connections and molecular signaling within the cortical areas related to pain. Furthermore, these neuroplastic changes can occur in cortical synapse, cortical excitation, GABA mediated inhibitory transmission, and long term depression (Zhuo, 2008).
In our study, based on our ECDs findings, the changes may occur mainly in the cortical synapses and cortical excitation and cortical excitation could also be observed in a time-frequency domain.
In the frequency domain, there was a long-lasting (up to 1 s) increased power at high and ultra high frequencies in the TMD pain group. This finding is suggestive of cortical excitation in a time-frequency domain. In an EEG study with neuropathic pain subjects, it was observed spontaneous presence of
113
enhanced activation within high theta (6-9 Hz), and low beta (12-16 Hz) frequencies located in the insular cortex (IC), anterior cingulate cortex (ACC), prefrontal, inferior posterior parietal cortices, SI, SII, and supplementary somatosensory cortices suggesting frequency specific ongoing excitation may contribute to the subject’s chronic pain (Stern et al., 2006). Also, Hsiao et al.
(2008) studied the frequency characteristics of pain-evoked neuromagnetic responses in SII cortex. The results indicated increased theta (4-8 Hz) and alpha
(8-13 Hz) power in SII bilaterally 180-210 ms after stimulus onset (Hsiao et al.,
2008).
In regards to the multivariate analyses, to our knowledge, no studies were performed in order to measure the brain dynamics in orofacial pain or pain subjects using MEG. As we mentioned earlier, MDS and HTC have proved very useful methods in behavioral (Whang et al., 1999), neurophysiological (Merchant et al., 2003), and functional neuroimaging (Merchant et al., 2003; Tagaris et al.,
1997; Tagaris et al., 1998; Tzagarakis et al., 2009) studies. We showed that with these methods we were able to differentiate the two groups based on their brain dynamics of how they differ in processing the same innocuous tactile information.
In summary, innocuous tactile stimulation proved to be a successful way to measure brain spatio-temporal dynamics in two group population. In contrast to previous studies that used electrical, heat, and laser stimulation to produce a painful stimulus, we were able to demonstrate very clear the differences in
114
brain organization and dynamics between these two groups using an innocuous stimulus and without causing an unpleasant feeling. The results obtained allow for a paradigm shift in future research of brain mechanisms in pain by the use of non-painful tactile stimuli to evaluate brain function in various orofacial (or other) pain conditions, including neurovascular and neuropathic pains and other complex orofacial pain disorders.
115
BIBLIIOGRAPHY
Albuquerque RJ, de Leeuw R, Carlson CR, Okeson JP, Miller CS, Andersen AH (2006). Cerebral activation during thermal stimulation of patients who have burning mouth disorder: an fMRI study. Pain 122(3):223-34.
Andrew D, Craig AD (2001). Spinothalamic lamina I neurones selectively responsive to cutaneous warming in cats. J Physiol 537(Pt 2):489-95.
Apkarian AV, Bushnell MC, Treede RD, Zubieta JK (2005). Human brain mechanisms of pain perception and regulation in health and disease. Eur J Pain 9(4):463-84.
Bagley CA, Ohara S, Lawson HC, Lenz FA (2006). Psychophysics of CNS pain- related activity: binary and analog channels and memory encoding. Neuroscientist 12(1):29-42.
Baillet S, Mosher JC, Leahy RM (2001). Electromagnetic Brain Mapping. Signal Processing Magazine, IEEE. 18(6):14 - 30.
Baumgartner C, Barth DS, Levesque MF, Sutherling WW (1992). Human hand and lip sensorimotor cortex as studied on electrocorticography. Electroencephalogr Clin Neurophysiol 84(115-126.
Bessho H, Shibukawa Y, Shintani M, Yajima Y, Suzuki T, Shibahara T (2007). Localization of palatal area in human somatosensory cortex. J Dent Res 86(3):265-70.
Bewyer DC (1989). Bimechanical and physiological processes as cause of internal displacement with adhesion. Quintessence 40(11):2025-34.
Biermann K, Schmitz F, Witte OW, Konczak J, Freund HJ, Schnitzler A (1998). Interaction of finger representation in the human first somatosensory cortex: a neuromagnetic study. Neurosci Lett 251(1):13-6.
Blatow M, Nennig E, Sarpaczki E, Reinhardt J, Schlieter M, Herweh C, et al. (2009). Altered somatosensory processing in trigeminal neuralgia. Hum Brain Mapp.
Bleiker R (1938). Ear disturbance of the temporomandibular joint origin. J A D A & D Cos 25(1390- 1399.
116
Boering G, Stegenga B, de Bont LG (1990). Temporomandibular joint osteoarthrosis and internal derangement. Part I: Clinical course and initial treatment. Int Dent J 40(6):339-46.
Boonstra TW, Daffertshofer A, Breakspear M, Beek PJ (2007). Multivariate time- frequency analysis of electromagnetic brain activity during bimanual motor learning. Neuroimage 36(2):370-7.
Borsook D, DaSilva AF, Ploghaus A, Becerra L (2003). Specific and somatotopic functional magnetic resonance imaging activation in the trigeminal ganglion by brush and noxious heat. J Neurosci 23(21):7897-903.
Brattberg G, Thorslund M, Wikman A (1989). The prevalence of pain in a general population. The results of a postal survey in a county of Sweden. Pain 37(2):215- 22.
Burton H (1986). Second somatosensory cortex and related areas. In: Cerebral cortex , Sensory-motor areas and aspects of cortical conectivity. EG Jones and A Peters editors. New York: Plenum., pp. 31-98.
Bushnell MC, Duncan GH, Hofbauer RK, Ha B, Chen JI, Carrier B (1999). Pain perception: is there a role for primary somatosensory cortex? Proc Natl Acad Sci U S A 96(14):7705-9.
Buzsaki G, Draguhn A (2004). Neuronal oscillations in cortical networks. Science 304(5679):1926-9.
Cain DM, Khasabov SG, Simone DA (2001). Response properties of mechanoreceptors and nociceptors in mouse glabrous skin: an in vivo study. J Neurophysiol 85(4):1561-74.
Cajal SR (1995). General structure of the cerebral cortex. In: Histology of the nervous system of man and vertebrates. New York: Oxford University press.
Casey KL, Minoshima S, Berger KL, Koeppe RA, Morrow TJ, Frey KA (1994). Positron emission tomographic analysis of cerebral structures activated specifically by repetitive noxious heat stimuli. J Neurophysiol 71(2):802-7.
Cimatti Z, Schwartz DP, Bourdain F, Meunier S, Bleton JP, Vidailhet M, et al. (2007). Time-frequency analysis reveals decreased high-frequency oscillations in writer's cramp. Brain 130(Pt 1):198-205.
Clark GT (1985). Muscle hyperactivity, pain and dysfunction. In: Orofacial Pain and Neuromuscular Dysfunction: Mechanisms and clinical Correlates. I Klineberg and BJ Sessle editors. Sidney: Pergamon., pp. 103 - 111.
117
Costen JB (1934). A syndrome of ear and sinus symptoms dependent upon disturbed function of the temporomandibular joint. Ann of Otology, Rhinology and Laryngology 43(1):1-15.
Cox DR, Oakes D (1984). Analysis of Survival Data Boca Raton, Fl: Chapman & Hall/CRC.
Craig AD (1995). Supraspinal projections of lamina one neurons. In: Forebrain areas involved in pain processing. JM Besson, G Guillbaund and H Ollat editors. London: Libby, pp. 13 - 25.
Craig AD, Reiman EM, Evans A, Bushnell MC (1996). Functional imaging of an illusion of pain. Nature 384(6606):258-60.
Craig AD (2003a). A new view of pain as a homeostatic emotion. Trends Neurosci 26(6):303-7.
Craig AD (2003b). Pain mechanisms: labeled lines versus convergence in central processing. Annu Rev Neurosci 26(1-30.
Dao TT, Lavigne GJ, Charbonneau A, Feine JS, Lund JP (1994a). The efficacy of oral splints in the treatment of myofascial pain of the jaw muscles: a controlled clinical trial. Pain 56(1):85-94.
Dao TT, Lund JP, Lavigne GJ (1994b). Comparison of pain and quality of life in bruxers and patients with myofascial pain of the masticatory muscles. J Orofac Pain 8(4):350-6.
DaSilva AF, Goadsby PJ, Borsook D (2007). Cluster headache: a review of neuroimaging findings. Curr Pain Headache Rep 11(2):131-6.
DaSilva AF, Becerra L, Pendse G, Chizh B, Tully S, Borsook D (2008). Colocalized structural and functional changes in the cortex of patients with trigeminal neuropathic pain. PLoS ONE 3(10):e3396.
Davis KD (2000). Studies of pain using functional magnetic resonance imaging. In: Pain imaging. KL Casey and MC Bushnell editors. Seattle, WA: IASP Press, pp. 195 - 210.
De Kanter RJ, Truin, G. J. , Burgersdijk, R. C., Van 't Hof, M. A., Battistuzzi, P. G., Kalsbeek, H., Kayser, A. F. (1983). Prevalence in the Dutch adult population and a meta-analysis of signs and symptoms of temporomandibular disorder. J Dent Res. 72(11):1509-18.
118
de Leeuw R, Albuquerque R, Okeson J, Carlson C (2005). The contribution of neuroimaging techniques to the understanding of supraspinal pain circuits: implications for orofacial pain. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 100(3):308-14. de Leeuw R, Davis CE, Albuquerque R, Carlson CR, Andersen AH (2006). Brain activity during stimulation of the trigeminal nerve with noxious heat. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 102(6):750-7.
Derbyshire SW, Jones AK, Devani P, Friston KJ, Feinmann C, Harris M, et al. (1994). Cerebral responses to pain in patients with atypical facial pain measured by positron emission tomography. J Neurol Neurosurg Psychiatry 57(10):1166- 72.
Dolwick MF, Riggs RR (1983). Diagnosis and treatment of internal derangements of the temporomandibular joint. Dent Clin North Am 27(3):561-72.
Dong WK, Salonen LD, Kawakami Y, Shiwaku T, Kaukoranta EM, Martin RF (1989). Nociceptive responses of trigeminal neurons in SII-7b cortex of awake monkeys. Brain Res 484(1-2):314-24.
Dong WK, Chudler EH, Sugiyama K, Roberts VJ, Hayashi T (1994). Somatosensory, multisensory, and task-related neurons in cortical area 7b (PF) of unanesthetized monkeys. J Neurophysiol 72(2):542-64.
Dubner R, Sessle BJ, Storey AT (1978). The Neural Basis of Oral and Facial Function. New York.: Plenum Press.
Dworkin SF, Huggins KH, LeResche L, Von Korff M, Howard J, Truelove E, et al. (1990a). Epidemiology of signs and symptoms in temporomandibular disorders: clinical signs in cases and controls. J Am Dent Assoc 120(3):273-81.
Dworkin SF, LeResche L, Von Korff MR (1990b). Diagnostic studies of temporomandibular disorders: challenges from an epidemiologic perspective. Anesth Prog 37(2-3):147-54.
Farrar WB, McCarty WL, Jr. (1979). The TMJ dilemma. J Ala Dent Assoc 63(1):19-26.
Flor H, Nikolajsen L, Staehelin Jensen T (2006). Phantom limb pain: a case of maladaptive CNS plasticity? Nat Rev Neurosci 7(11):873-81.
Fricton JR (1990). Myofascial pain syndrome: Characteristics and epidemiology. In: Advances in Pain Research and Therapy. JR Fricton and E Awad editors. New York: Raven, pp. 107 - 127.
119
Georgopoulos AP (1976). Functional properties of primary afferent units probably related to pain mechanisms in primate glabrous skin. J Neurophysiol 39(1):71-83.
Georgopoulos AP (1977). Stimulus-response relations in high-threshold mechanothermal fibers innervating primate glabrous skin. Brain Res 128(3):547- 52.
Gescheider GA (1985). Psychophysics method, theory, and application. 2nd ed. Hillsdale, NJ: Lawrence Erlbaum Associates, Inc., Publishers.
Glass EG, McGlynn, F. D., Glaros, A. G., Melton, K., Romans, K. (1993). Prevalence of temporomandibular disorder symptoms in a major metropolitan area. Cranio. 11(3):217-20.
Greene CS, Marbach JJ (1982). Epidemiologic studies of mandibular dysfunction: a critical review. J Prosthet Dent 48(2):184-90.
Hagelberg N, Forssell H, Rinne JO, Scheinin H, Taiminen T, Aalto S, et al. (2003). Striatal dopamine D1 and D2 receptors in burning mouth syndrome. Pain 101(1-2):149-54.
Hamalainen M, Hari R, Ilmoniemi R, Knuutila J, Lounasmaa O (1993). Magnetoencephalography - theory, instrumentation, and applications to noninvasive studies of the working human brain. Reviews of Modern Physics 65(2):413 - 497.
Han ZS, Zhang ET, Craig AD (1998). Nociceptive and thermoreceptive lamina I neurons are anatomically distinct. Nat Neurosci 1(3):218-25.
Hauck M, Lorenz J, Engel AK (2007). Attention to painful stimulation enhances gamma-band activity and synchronization in human sensorimotor cortex. J Neurosci 27(35):9270-7.
Hiltunen K, Schmidt-Kaunisaho K, Nevalainen J, Narhi T, Ainamo A (1995). Prevalence of signs of temporomandibular disorders among elderly inhabitants of Helsinki, Finland. Acta Odontol Scand 53(1):20-3.
Hofbauer RK, Rainville P, Duncan GH, Bushnell MC (2001). Cortical representation of the sensory dimension of pain. J Neurophysiol 86(1):402-11.
Hollins M, Delemos KA, Goble AK (1991). Vibrotactile adaptation on the face. Percept. Psychophys. 49(21-30.
120
Hoshiyama M, Kakigi R, Koyama S, Kitamura Y, Shimoio M, Watanabe S (1995). Somatosensory evoked magnetic fields after mechanical stimulation of the scalp in humans. Neurosci Lett 195(1):29-32.
Hsiao FJ, Chen WT, Liao KK, Wu ZA, Ho LT, Lin YY (2008). Oscillatory characteristics of nociceptive responses in the SII cortex. Can J Neurol Sci 35(5):630-7.
Hyvarinen J (1982). The parietal cortex of monkey and man. Berlin: Springer.
Ioannides AA (2006). Magnetoencephalography as a research tool in neuroscience: state of the art. Neuroscientist 12(6):524-44.
Jantsch HH, Kemppainen P, Ringler R, Handwerker HO, Forster C (2005). Cortical representation of experimental tooth pain in humans. Pain 118(3):390-9.
Jaros U, Hilgenfeld B, Lau S, Curio G, Haueisen J (2008). Nonlinear interactions of high-frequency oscillations in the human somatosensory system. Clin Neurophysiol 119(11):2647-57.
Jiang T, Li J, Jin Z, Wang YW, Feng HL, Ishikawa T (2006). [Comparison of atypical orofacial pain and temporomandibular disorders synovitis pain processing in the human brain using functional magnetic resonance imaging]. Zhonghua Kou Qiang Yi Xue Za Zhi 41(11):670-3.
Johansson RS, Trulsson M, Olsson KA, Westberg KG (1988). Mechanoreceptor activity from the human face and oral mucosa. Exp Brain Res 72(1):204-8.
Johnson KO (2001). The roles and functions of cutaneous mechanoreceptors. Curr Opin Neurobiol 11(4):455-61.
Jones AK, Brown WD, Friston KJ, Qi LY, Frackowiak RS (1991). Cortical and subcortical localization of response to pain in man using positron emission tomography. Proc Biol Sci 244(1309):39-44.
Jones EG, Powell TP (1970). An anatomical study of converging sensory pathways within the cerebral cortex of the monkey. Brain 93(4):793-820.
Jones EG (2000). Cortical and subcortical contributions to activity-dependent plasticity in primate somatosensory cortex. Annu Rev Neurosci 23(1-37.
Jousmaki V (2000). Tracking functions of cortical networks on a millisecond timescale. Neural Netw 13(8-9):883-9.
121
Junge D (1998). Oral Sensorimotor Function Pacific: Medico Dental Media International Inc.
Karhu J, Hari R, Lu ST, Paetau R, Rif J (1991). Cerebral magnetic fields to lingual stimulation. Electroencephalogr Clin Neurophysiol 80(459-468.
Kass J (1990). Somatosensory system London: Academic Press.
Krubitzer L, Clarey J, Tweedale R, Elston G, Calford M (1995). A redefinition of somatosensory areas in the lateral sulcus of macaque monkeys. J Neurosci 15(5 Pt 2):3821-39.
Kupers R, Kehlet H (2006). Brain imaging of clinical pain states: a critical review and strategies for future studies. Lancet Neurol 5(12):1033-44.
Kupers RC, Gybels JM, Gjedde A (2000). Positron emission tomography study of a chronic pain patient successfully treated with somatosensory thalamic stimulation. Pain 87(3):295-302.
Kupers RC, Svensson P, Jensen TS (2004). Central representation of muscle pain and mechanical hyperesthesia in the orofacial region: a positron emission tomography study. Pain 108(3):284-93.
Lang E, Kaltenhauser M, Seidler S, Mattenklodt P, Neundorfer B (2005). Persistent idiopathic facial pain exists independent of somatosensory input from the painful region: findings from quantitative sensory functions and somatotopy of the primary somatosensory cortex. Pain 118(1-2):80-91.
Langheim FJ, Merkle AN, Leuthold AC, Lewis SM, Georgopoulos AP (2006). Dipole analysis of magnetoencephalographic data during continuous shape copying. Exp Brain Res 170(4):513-21.
Layzer RB (1986). Muscle pain, cramps, and fatigue. In: Myology: Basic and Clinical. AG Engel and BQ Banker editors. New York: McGraw-Hill, pp. 1907 - 1923.
Lee YH, Yu KK, Kwon H, Kim JM, Kim K, Park YK, et al. (2009). A whole-head magnetoencephalography system with compact axial gradiometer structure. Supercond. Sci. and Technol. 22(1 - 6.
Leuthold AC (2003). Subtraction of heart artifact from MEG data: the matched filter revisited. . Society for Neuroscience.
122
Lewine JD, Orrison JWW (1995b). Magnetoencephalography and magnetic source imaging. In: Functional Brain Imaging. WW Orrison, JD Lewine, JA Sanders and MF Hartshorne editors. St. Louis: Mosby-year Book, pp. 369-417.
Lipton JA, Ship JA, Larach-Robinson D (1993). Estimated prevalence and distribution of reported orofacial pain in the United States. J Am Dent Assoc 124(10):115-21.
Locker D, Slade G (1988). Prevalence of symptoms associated with temporomandibular disorders in a Canadian population. Community Dent Oral Epidemiol 16(5):310-3.
Lorente de No R (1947). Analysis of the distribution of action currents of nerve in a volume conductor. In: Stud Rockefeller Inst Med Res, pp. 384-477.
Lorente de No R (1949). Cerebral cortex: Architecture, intracortical connections, motor projections. In: Physiology of the Nervous System. J Fulton editor: Oxford University Press, pp. 667.
Lumpkin EA, Caterina MJ (2007). Mechanisms of sensory transduction in the skin. Nature 445(7130):858-65.
McCarthy G, Allison T, Spencer DD (1993). Localization of the face area of human sensorimotor cortex by intracranial recording of somatosensory evoked potentials. J Neurosurg 79(6):874-84.
Mense S (1990). Physiology of nociception in muscle. In: Advances in Pain Research and Therapy. JR Fricton and E Awad editors. New York: Raven.
Mense S (1991). Considerations concerning the neurobiological basis of muscle pain. Can J Physiol Pharmacol 69(610 - 616.
Mense S (1993). Nociception from skeletal muscle in relation to clinical muscle pain. Pain 54(241 - 290.
Merchant H, Battaglia-Mayer A, Georgopoulos AP (2003). Functional organization of parietal neuronal responses to optic-flow stimuli. J Neurophysiol 90(2):675-82.
Morelli N, Pesaresi I, Cafforio G, Maluccio MR, Gori S, Di Salle F, et al. (2008). Functional magnetic resonance imaging in episodic cluster headache. J Headache Pain.
123
Nagamatsu K, Nakasato N, Hatanaka K, Kanno A, Iwasaki M, Yoshimoto T (2001). Neuromagnetic localization of N15, the initial cortical response to lip stimulus. Neuroreport 12(1):1-5.
Nakahara H, Nakasato N, Kanno A, Murayama S, Hatanaka K, Itoh H, et al. (2004). Somatosensory-evoked fields for gingiva, lip, and tongue. J Dent Res 83(4):307-11.
Nakamura A, Yamada T, Goto A, Kato T, Ito K, Abe Y, et al. (1998). Somatosensory homunculus as drawn by MEG. Neuroimage 7(4 Pt 1):377-86.
Nakasato N, Itoh H, Hatanaka K, Nakahara H, Kanno A, Yoshimoto T (2001). Movement-related magnetic fields to tongue protrusion. Neuroimage 14(4):924- 35.
Nash P, Henderson L, Macefield V, Klineberg I, Murray G (2007). Brain representation of noxious stimulation of masseter: an fMRI study. IADR/AADR/CADR 85th General Session and Exhibition.
Neuroimaging D Biomagnetism A Primer, pp. 11.
Nevalainen P, Ramstad R, Isotalo E, Haapanen ML, Lauronen L (2006). Trigeminal somatosensory evoked magnetic fields to tactile stimulation. Clin Neurophysiol 117(9):2007-15.
Nguyen BT, Tran TD, Hoshiyama M, Inui K, Kakigi R (2004). Face representation in the human primary somatosensory cortex. Neurosci Res 50(2):227-32.
Nguyen BT, Inui K, Hoshiyama M, Nakata H, Kakigi R (2005). Face representation in the human secondary somatosensory cortex. Clin Neurophysiol 116(6):1247-53.
Nikouline VV, Wikstrom H, Linkenkaer-Hansen K, Kesaniemi M, Ilmoniemi RJ, Huttunen J (2000). Somatosensory evoked magnetic fields: relation to pre- stimulus mu rhythm. Clin Neurophysiol 111(7):1227-33.
Nolte J (2002). Cerebral cortex. In: The Human Brain An Introduction to its Functional Anatomy. St. Louis: Mosby, Inc.
Nourallah H, Johansson A (1995). Prevalence of signs and symptoms of temporomandibular disorders in a young male Saudi population. J Oral Rehabil 22(5):343-7.
124
Okeson JP (1996). Differential Diagnosis and Management Considerations of the Temporomandibular Disorders. In: Orofacial Pain Guidelines for Assessment, Diagnosis, and Management. Il: Quintessence Publishing Co, Inc,, pp. 113-84.
Okeson JP (2003). Causes of Functional Disturbances in the Masticatory System. In: Management of Temporomandibular Disorder and Occlusion.: 5th ed. Mosby, pp. 149-89.
Pantev C (1995). Evoked and induced gamma-band activity of the human cortex. Brain Topogr 7(4):321-30.
Penfield W, Rasmussen T (1950). The cerebral cortex of man: a clinical study of localization of function. New York.: Macmillan.
Picard C, Olivier A (1983). Sensory cortical tongue representation in man. J Neurosurg 59(5):781-9.
Pippin B, Moore, T., McCulloch, A., Moore, s. (1940). A method of repositioning the mandible in the treatment of lesions of the temporomandibular joint. The Washington University Dental Journal. 6(4):107-20.
Rainville P, Bushnell MC, Duncan GH (2000). PET studies of the subjective experience of pain. In: Pain imaging. KL Casey and MC Bushnell editors. Seattle, WA: IASP Press, pp. 123 - 56.
Ray S, Niebur E, Hsiao SS, Sinai A, Crone NE (2008). High-frequency gamma activity (80-150Hz) is increased in human cortex during selective attention. Clin Neurophysiol 119(1):116-33.
Riggs L, Moses SN, Bardouille T, Herdman AT, Ross B, Ryan JD (2008). A complementary analytic approach to examining medial temporal lobe sources using magnetoencephalography. Neuroimage.
Robinson CJ, Burton H (1980). Somatic submodality distribution within the second somatosensory (SII), 7b, retroinsular, postauditory, and granular insular cortical areas of M. fascicularis. J Comp Neurol 192(1):93-108.
Ross B, Herdman AT, Pantev C (2005). Stimulus induced desynchronization of human auditory 40-Hz steady-state responses. J Neurophysiol 94(6):4082-93.
Sahlin K, Edstrom L, Sjoholm H, Hultman E (1981). Effects of lactic acid accumulation and ATP decrease on muscle tension and relaxation. Am J Physiol 240(3):C121-6.
125
Sakamoto K, Nakata H, Kakigi R (2008). Somatotopic representation of the tongue in human secondary somatosensory cortex. Clin Neurophysiol 119(9):2125-34.
Sarnat BG (1951). The Temporomandibular Joint Springfield, Il: 1st ed. Charles C Thomas.
Schiffman A (1984). Myofascial pain associated with unilateral masseteric hypertrophy in a condylectomy patient. J Craniomand Pract 2(373 - 376.
Schiffman EL, Fricton JR, Haley DP, Shapiro BL (1990). The prevalence and treatment needs of subjects with temporomandibular disorders. J Am Dent Assoc 120(3):295-303.
Schwartz L (1959). Disorders of the Temporomandibular Joint. Philadelphia: WB Saunders.
Sessle BJ (2000). Acute and chronic craniofacial pain: brainstem mechanisms of nociceptive transmission and neuroplasticity, and their clinical correlates. Crit Rev Oral Biol Med 11(1):57-91.
Shore NA (1959). Occlusal equilibration and temporomandibular joint dysfunction. Philadelphia: JB Lippincott Co.
Simone DA, Kajander KC (1996). Excitation of rat cutaneous nociceptors by noxious cold. Neurosci Lett 213(1):53-6.
Snedecor GW, Cochran WG (1989). Statistical Methods 8th ed. Ames, IA: Iowa State University Press.
Solberg WK, Woo MW, Houston JB (1979). Prevalence of mandibular dysfunction in young adults. J Am Dent Assoc 98(1):25-34.
Solberg WK (1986). Temporomandibular disorders: masticatory myalgia and its management. Br Dent J 160(10):351-6.
Steeghs P, Baraniuk R, Odegard J (2002). Applications in Time-Frequency Signal Processing Boca Raton, FL: CRC.
Stegenga B, de Bont LG, Boering G (1989). Osteoarthrosis as the cause of craniomandibular pain and dysfunction: a unifying concept. J Oral Maxillofac Surg 47(3):249-56.
126
Stegenga B, de Bont LG, Boering G, van Willigen JD (1991). Tissue responses to degenerative changes in the temporomandibular joint: a review. J Oral Maxillofac Surg 49(10):1079-88.
Stern J, Jeanmonod D, Sarnthein J (2006). Persistent EEG overactivation in the cortical pain matrix of neurogenic pain patients. Neuroimage 31(2):721-31.
Tagaris GA, Kim SG, Andersen P, Ugurbil K, Georgopoulos AP (1997). Mental rotation studied by functional magnetic resonance imaging at high field (4 Tesla): Performance and cortical activation. J. Cogn. Neurosci. 9(419 - 432.
Tagaris GA, Richter W, Kim SG, Pellizzer G, Andersen P, Ugurbil K, et al. (1998). Functional magnetic resonance imaging of mental rotation and memory scanning: a multidimensional scaling analysis of brain activation patterns. Brain Res Brain Res Rev 26(2-3):106-12.
Tailarach J, Tournoux P (1988). Co-planar stereotaxic atlas of the human brain New York: Thieme Medical Publishers.
Talbot JD, Marrett S, Evans AC, Meyer E, Bushnell MC, Duncan GH (1991). Multiple representations of pain in human cerebral cortex. Science 251(4999):1355-8.
Tierra-Criollo CJ, Infantosi AF (2006). Low-frequency oscillations in human tibial somatosensory evoked potentials. Arq Neuropsiquiatr 64(2B):402-6.
Tronnier VM, Rasche D, Hamer J, Kienle AL, Kunze S (2001). Treatment of idiopathic trigeminal neuralgia: comparison of long-term outcome after radiofrequency rhizotomy and microvascular decompression. Neurosurgery 48(6):1261-7; discussion 1267-8.
Tzagarakis C, Jerde TA, Lewis SM, Ugurbil K, Georgopoulos AP (2009). Cerebral cortical mechanisms of copying geometrical shapes: a multidimensional scaling analysis of fMRI patterns of activation. Exp Brain Res 194(3):369-80.
Uutela K, Hamalainen M, Somersalo E (1999). Visualization of magnetoencephalographic data using minimum current estimates. Neuroimage 10(2):173-80.
Van Buren JM, Borke RC (1972). Variations and connections of the human thalamus Berlim: Springer Verlag.
Vartiainen N, Kirveskari E, Kallio-Laine K, Kalso E, Forss N (2009). Cortical reorganization in primary somatosensory cortex in patients with unilateral chronic pain. J Pain 10(8):854-9.
127
Von Korff M, Dworkin SF, Le Resche L, Kruger A (1988). An epidemiologic comparison of pain complaints. Pain 32(2):173-83.
Vrba J, Robinson SE (2001). Signal processing in magnetoencephalography. Methods 25(2):249-71.
Whang KC, Crowe DA, Georgopoulos AP (1999). Multidimensional scaling analyses of two construction-related tasks. Exp Brain Res 125(3):231-8.
Wolters CH, Beckmann RF, Rienacker A, Buchner H (1999). Comparing regularized and non-regularized nonlinear dipole fit methods: a study in a simulated sulcus structure. Brain Topogr 12(1):3-18.
Wood PB, Glabus MF, Simpson R, Patterson JC, 2nd (2009). Changes in gray matter density in fibromyalgia: correlation with dopamine metabolism. J Pain 10(6):609-18.
Woodbury CJ, Koerber HR (2007). Central and peripheral anatomy of slowly adapting type I low-threshold mechanoreceptors innervating trunk skin of neonatal mice. J Comp Neurol 505(5):547-61.
Yang TT, Gallen CC, Schwartz BJ, Bloom FE (1993). Noninvasive somatosensory homunculus mapping in humans by using a large-array biomagnetometer. Proc Natl Acad Sci U S A 90(7):3098-102.
Zaslansky R, Sprecher E, Katz Y, Rozenberg B, Hemli JA, Yarnitsky D (1996). Pain-evoked potentials: what do they really measure? Electroencephalogr Clin Neurophysiol 100(5):384-91.
Zhuo M (2008). Cortical excitation and chronic pain. Trends Neurosci 31(4):199- 207.
128