Magnetic Resonance Imaging Techniques for Visualization of the Subthalamic Nucleus

J Neurosurg 115:971–984, 2011

Magnetic resonance imaging techniques for visualization of the subthalamic nucleus

A review

Ellen J. L. Brunenberg, M.Sc.,1 Bram Platel, Ph.D.,2 Paul A. M. Hofman, Ph.D., M.D.,3 Bart M. ter Haar Romeny, Ph.D.,1,4 and Veerle Visser-Vandewalle, Ph.D., M.D.5,6 1Department of Biomedical Engineering, Eindhoven University of Technology, Eindhoven; Departments of 3Radiology, 4Biomedical Engineering, and 5Neurosurgery, and 6Maastricht Institute for Neuromodulative Development, Maastricht University Medical Center, Maastricht, The Netherlands; and 2Fraunhofer MEVIS, Bremen, Germany

The authors reviewed 70 publications on MR imaging–based targeting techniques for identifying the subthalamic nucleus (STN) for deep brain stimulation in patients with Parkinson disease. Of these 70 publications, 33 presented quantitatively validated results. There is still no consensus on which targeting technique to use for surgery planning; methods vary greatly between centers. Some groups apply indirect methods involving anatomical landmarks, or atlases incorporating anatomical or functional data. Others perform direct visualization on MR imaging, using T2-weighted spin echo or inversion recovery protocols. The combined studies do not offer a straightforward conclusion on the best targeting protocol. Indirect methods are not patient specific, leading to varying results between cases. On the other hand, direct targeting on MR imaging suffers from lack of contrast within the subthalamic region, resulting in a poor delineation of the STN. These defi- ciencies result in a need for intraoperative adaptation of the original target based on test stimulation with or without microelectrode recording. It is expected that future advances in MR imaging technology will lead to improvements in direct targeting. The use of new MR imaging modalities such as diffusion MR imaging might even lead to the specific identification of the different functional parts of the STN, such as the dorsolateral sensorimotor part, the target for deep brain stimulation. (DOI: 10.3171/2011.6.JNS101571)

Key Words • deep brain stimulation • targeting • subthalamic nucleus • MR imaging • functional neurosurgery

eep brain stimulation of the STN was introduced terioration and behavioral changes98 and have been re- as a treatment for advanced Parkinson disease ported in approximately 40%–50% of patients.89 It is hy- in 1993.69 Patients who initially showed a good pothesized that these side effects are the result of direct Dresponse to levodopa but suffer from on-off fluctuations stimulation of or current spread to the nonmotor parts of after a prolonged medication regimen have been shown the STN.88 In addition to a large dorsolateral motor part, to benefit from STN DBS. The beneficial effects of STN the STN has 2 other functional subterritories, namely a DBS on the motor symptoms of Parkinson disease have small ventromedial limbic part and a cognitive part inter- been described in long-term follow-up studies,74,95,96 but mediate in position and size.41,88 unwanted stimulation-induced side effects can occur. Precise targeting of the dorsolateral (motor) part of These side effects have been described as cognitive de- the STN is of great importance for 2 reasons: first of all, to obtain the best possible effect on motor symptoms, and second, to minimize the undesirable side effects on cog- Abbreviations used in this paper: AC = anterior commissure; 35,92 DBS = deep brain stimulation; FSE = fast spin echo; IR = inversion nition and behavior. Yet, the operative technique of recovery; MCP = midcommissural point; MER = microelectrode recording; MPRAGE = magnetization prepared rapid acquisition This article contains some figures that are displayed in color gradient echo; PC = posterior commissure; STIR = short tau inver- online but in black and white in the print edition. sion recovery; STN = subthalamic nucleus; TSE = turbo spin echo.

J Neurosurg / Volume 115 / November 2011 971 E. J. L. Brunenberg et al.

STN DBS including the imaging method for primary targeting of the STN still varies greatly between centers.53 Magnetic resonance imaging is one of the most frequent- ly used modalities, alone or in combination with CT or ventriculography. 90 Most DBS procedures also involve a secondary targeting step using intraoperative assess- ments, such as MER or macrostimulation, to adapt the position of the electrodes to the clinical goals.17,30 The aim of this paper is to review the multitude of available methods and give a systematic overview of techniques for primary targeting of the dorsolateral part of the STN. This review discusses both indirect and direct targeting procedures, as well as comparative studies. The overall focus of the review is on MR imaging, because this modality is most widely used. The next section of this paper will summarize indirect and direct targeting methods using MR imaging. Subsequently, the results of Fig. 1. Diagram showing AC-PC line–based targeting in a sagittal the systematic review will be presented. Finally, the ad- plane. After the AC and PC are identified, the midcommissural point vantages and drawbacks of the targeting techniques will (MCP) is determined. From there, the STN position can be calculated be discussed. using fixed distances (often 12 mm lateral, 3 mm inferior, and 3 mm posterior) based on an atlas.

Summary of Targeting Methods An often used method is the T2-weighted FSE MR 81,83 Methods for targeting the dorsolateral STN can be imaging protocol, on which the STN shows as a hypointense area (see Fig. 5). Another popular technique for classified as either direct or indirect. Indirect methods 48 rely on contextual information, while direct methods fo- direct visualization is the IR protocol (see Fig. 6). Still cus on visualizing the STN itself on preoperative images. subject to more research are the use of relaxation time maps12 and susceptibility-weighted (T2*) imaging (see 33 Indirect Targeting Fig. 7), together with special postprocessing methods to reduce signal loss and enhance contrast. Indirect targeting can rely on 2 types of contextual information—patient-specific landmarks or generic atlas Systematic Review information adapted to the patient’s anatomy. The gold standard in indirect targeting using ana- In order to present a review on the different target- tomical landmarks involves the AC and PC.82 Once one ing methods, we performed a thorough search for papers has localized the AC, PC, and the midcommissural point, on STN targeting. The criteria that we used are briefly the stereotactic coordinates of the STN follow from fixed outlined below. distances based on classical atlases such as the Schalten- brand-Wahren brain atlas, as can be seen in Fig. 1.77 The Search Strategy second most widely used landmark is the red nucleus First, we searched PubMed and ScienceDirect for (Fig. 2).7,10 Less widespread methods include the use of papers published between January 1999 and January the nigrocapsular angle (as explained in Fig. 3),34 the 2011, using the terms “targeting,” “magnetic resonance supramammillary commissure,52 the postmammillary imaging,” “MRI,” or “visualization” in combination with commissure,91 and the line connecting the mammillary “subthalamic nucleus” or “STN.” Second, we explored bodies and the PC.73 the bibliographies of relevant publications until no fur- Furthermore, regions of interest can also be deter- ther additional studies were found. Papers were selected mined by mapping an anatomical and/or functional at- if they provided qualitative descriptions or quantitative las onto the patient’s MR imaging (see Fig. 4). Anatomi- results of direct or indirect MR imaging–based STN tar- cal atlases incorporate information about the position geting methods. Language other than English was an ex- of brain structures.62,76 The most popular ones are the clusion criterion. Talairach-Tournoux atlas,85 the Schaltenbrand-Wahren We included 70 publications, of which 33 stud- atlas,77 and the Montreal Neurological Institute atlas.22 ies quantitatively validated results of patient trials. The Functional atlases consist of point sets that are collected quantitative results are reported in Tables 1–4. The de- during MER, postoperative imaging or neurological as- tails extracted from the studies for this purpose included sessments.24,37,39,59,63 the following: number of subjects, targeting techniques used (landmark, MR imaging protocol, atlas, registration Direct Targeting method), validation method, outcome, and the main conclusion. Direct targeting involves the use of specific MR imaging protocols that enable direct visualization of the Indirect Targeting STN, avoiding the need to employ contextual information. Anatomical Landmarks. Publications on landmark-

972 J Neurosurg / Volume 115 / November 2011 Magnetic resonance imaging for STN visualization

Fig. 3. Diagram showing targeting based on the nigrocapsular angle, in a sagittal plane. The STN lies in the corner that is formed by the descending internal capsule (IC) and the substantia nigra (SN).

substantia nigra on MR imaging. Lee et al.52 proposed the supramammillary commissure, which lacked reliability. Toda et al.91 used the postmammillary commissure and selected the central electrode in 81% of cases. Finally, Rijkers et al.73 suggested the line connecting the mammillary bodies and PC, although this remains to be validated. Brain Atlas Registration. Table 2 summarizes the papers found on atlas-based STN targeting that presented quantitative results. Ortega et al.62 digitized the anatomi- Fig. 2. Diagram showing red nucleus–based targeting in an axial cal Talairach-Tournoux atlas, registered it nonlinearly, plane. The center of the STN lies on the same line as the anterior and compared the resulting targets with recorded micro- boundary of the red nucleus (RN). electrode positions, obtaining good results. However, No- winski et al.60 demonstrated inconsistency of the 3 planes based targeting presenting quantitative results can be for an anatomical atlas in the STN region, restricting the found in Table 1. After the ventriculography-based study use of the atlas in 3D. by Schuurman et al. in 1999,79 one of the first MR imag- Sánchez-Castro et al. collected MR imaging data ing–based studies was presented by Starr et al.82 Their with visible STNs to use as a brain atlas.18,76 Duay et al.31 AC-PC line–based targets differed 1–1.5 mm from di- increased the speed of the registration algorithm and rectly visualized STNs. Due to the variability in manual showed that the targeting variability is comparable to selection of the AC-PC line,65 the need for automated that of an expert, with a registration error similar to those methods became apparent. Ardekani and Bachman5 used reported by Sánchez-Castro et al.76 Bardinet et al.9 com- an atlas-based automatic analysis method, and the result- bined 2 kinds of anatomical information, namely histo- ing landmarks were very close (within 1 mm) to points logical and MR imaging data, into a human basal ganglia determined manually. A similar study by Pallavaram et atlas (see also Yelnik et al.100). After atlas registration, the al.64 yielded more accurate targets than achieved by man- resulting STN coordinates correlated well with MER and ual selection. postoperative MR imaging. The authors improved the The first studies on red nucleus–based targeting were registration step, enabling a better match with individual presented by Aziz et al.7 and Bejjani et al.10 The latter patient anatomy (error < 1 mm).8 reported selection of the central electrode in 19 of 24 With respect to functional atlases, Nowinski et al.61 cases. However, authors of other studies, based on best investigated the differences in spatial position between response23 and anatomical relationships,27 have stated that the anatomical and functional STN. The anatomical the red nucleus is not reliable enough. To solve this, Liu STN was derived from the Schaltenbrand-Wahren atlas,77 et al.56 used the red nucleus and the substantia nigra, but while the functional STN was constructed from ventricu- they did not report quantitative results. Pollo et al.70 ex- lography, MER, and radiographic data, and neurological ploited the AC-PC line, red nucleus, thalamus, internal assessment of 184 patients with Parkinson disease.59 The capsule, substantia nigra, and midline, resulting in a mean anatomical and functional STN correlated well. target in the inferior STN. D’Haese et al.24 developed a targeting method using Regarding less widespread methods, according to MER and optimal electrode contacts determined postop- Giller et al.,34 the use of the nigrocapsular angle was jus- eratively and showed that automatic target prediction is tified by the higher visibility of the internal capsule and feasible. Pallavaram et al.63 extended this, achieving a bet-

J Neurosurg / Volume 115 / November 2011 973 E. J. L. Brunenberg et al.

Fig. 4. Diagram illustrating the concept of indirect targeting based on atlas mapping. On the upper left, an atlas slice with anatomical information can be seen, containing labeled structures. At the bottom left, a functional atlas is represented, consist- ing of a cloud of target points that were used in previously performed DBS operations. On the right, the patient’s MR imaging data can be seen. The gray arrows represent the transformation that is necessary to map the atlases to the patient’s MR imaging data, resulting in overlaid information as shown in the center. ter correlation of the electrophysiological maps with the coordinates suffer from interpatient variability in STN underlying anatomy. Guo et al.37,39 compared 6 methods size72 and position.72,102 Littlechild et al.55 and Slavin et using functional data, including a brain atlas, MER data, al.80 both reported that the STN target varies considerably and a collection of previous targets. The combination of in relation to the midcommissural point, again suggest- all data performed best, followed by the previous targets ing that direct visualization on MR imaging enables more and the MER data. All 3 methods provided a more accu- accurate targeting than the atlas based method. Ashkan rate estimation than techniques solely dependent on anat- et al.6 confirmed this, finding a significant difference- be omy. The authors used these results to construct probabi- tween the atlas target and the STN as visualized on MR 38 listic maps for STN DBS trajectory planning. imaging. They found that the latter is on average 1.7 mm more medial, 0.7 mm more anterior, and 0.7 mm more Direct Targeting ventral than the atlas target. A substantial interindividual variability of STN size, T2-Weighted MR Imaging. As can be seen in Table 3, a multitude of papers have reported validated results orientation, and position and a significant difference between MR imaging– and atlas-based coordinates in 66 of T2-weighted MR imaging for direct STN identifica- 67 tion. Starr et al.83 showed that direct targeting is possible patients was also demonstrated by Patel et al. Davies and Daniluk and colleagues found similar variation, on T2-weighted FSE MR imaging, leading to consistent 25,28 electrode placement in the dorsolateral STN.81 Bejjani et based on series of 60 and 62 patients. In the latter se- al.10 selected the central track for implantation in 19 of ries, the lateral coordinate of the STN varied by as much 32 as 5.8 mm and the anteroposterior size of the STN varied their 24 cases. Egidi et al. reported even slightly better 26 results: their visualized target track was chosen in 85% by 8 mm, implying that the use of atlas-based targeting of cases. Dormont et al.29 examined postmortem tissue may result in failure to identify the STN during MER. 66 and showed that the hypointense signal reported by these Patel et al. validated direct targeting on axial and first studies correlates with the presence of iron and cor- coronal MR images. In a first series, they used macro- responds anatomically to the STN. They noted, however, stimulation for target adjustment; subsequently target that the most posterior part of the STN was not visible in verification was based on preoperative MR imaging only. any of the cases. The latter method yielded a mean error of only 0.3–0.4 In addition, research has been done on the correspon- mm. Together with a follow-up study on the improvement dence between MR imaging–based and landmark-based of activities of daily living and the motor score, this led to STN coordinates. Zhu et al.102 and Richter et al.72 con- the conclusion that targeting based on T2-weighted MR cluded that MR imaging is needed, because atlas-based imaging is safe and effective.68

974 J Neurosurg / Volume 115 / November 2011 Magnetic resonance imaging for STN visualization

Fig. 5. Coronal T2-weighted FSE MR image showing direct visualization of the STN. Fig. 6. Coronal IR MR image showing direct visualization of the STN.

Inversion Recovery MR Imaging. Researchers have T2*-weighted MR imaging. They compared the results also investigated IR methods for direct visualization of with STIR MR imaging and found that the posterior STN the basal ganglia. Starr81 identified the globus pallidus parts were not visible on the T2*-weighted images, lead- internus, while Benabid et al.11 depicted the globus pal- ing to the conclusion that a combination with STIR might lidus internus and the thalamus. The protocols they used be more useful to target the STN. were not suitable for visualizing the STN. Although many Bonny et al.12 generated multiple images with in- research groups have tried to improve upon these results, creased T2*-weighting to maximize the contrast between no quantitative studies on this subject are available yet. the STN and surrounding structures. Elolf et al.33 also Kitajima et al.48 proposed the STIR method and com- used multiple gradient echoes to add contrast to con- pared this to T2-weighted FSE MR imaging. The upper ventional T1-weighted MR imaging. Wu et al.99 used a and lateral STN margins were depicted more clearly by steady-state free precession MR imaging method, facili- FSE, while fast STIR was superior with respect to the vis- tating the visualization of midbrain nuclei with a heavy ibility of the lower margin and provided a better contrast T2*-weighting and relatively short echo time. However, between the STN and substantia nigra. In addition, Taoka none of these studies presented quantitative results on et al.86 developed 2 sets of guidelines that facilitate identi- STN targeting. fication of the STN on STIR images, the so-called “Suke- To enhance T2*-weighted images, Volz et al.97 pre- roku sign” and the “dent internal-capsule sign.” Ishimori sented a method to reduce signal loss, resulting in good et al.47 investigated phase-sensitive IR and showed that contrast. Another applicable technique is susceptibility- the STN position on IR MR imaging differs by at most 2 weighted imaging, a method that uses the phase data. mm from the coordinates found with AC-PC line–based Young and Chen101 exploited this technique to produce targeting—comparable to the results obtained with T2- T2* contrast additive to the T1 or T2 contrast intrinsic weighted images.6 to the imaging protocol, thus improving the contrast between the nuclei of interest. Rauscher et al.71 and Vertin- Susceptibility-Weighted MR Imaging. Apart from sky et al.94 also performed postprocessing with phase im- T2-weighted and IR MR imaging, it is also possible to ages, giving rise to a better visibility of the STN and even acquire T2*-weighted (susceptibility-weighted) MR im- of the subthalamic fasciculus. However, for these meth- ages, and T2* imaging is even more sensitive to local iron ods again, the STN position as identified on MR imaging deposits, as occur in the STN among other areas. Dor- has not been validated. mont et al.29 established the correlation of these iron deposits and the hypointensities on T2*-weighted MR im- Relaxation-Time Maps Derived From MR Imaging. aging, while Slavin et al.80 applied this to 3T data. Taoka Apart from T1-, T2- or T2*-weighted acquisitions, it is et al.87 also visualized the iron content of the STN using also possible to map the true relaxation time for each

J Neurosurg / Volume 115 / November 2011 975 E. J. L. Brunenberg et al.

Comparative Studies Much research has been done on the comparison of direct and indirect targeting. Table 4 summarizes the comparative papers that presented quantitative results. Studies That Favor Indirect Targeting. Zonenshayn et al.103 investigated STN targeting using 4 different methods, namely: 1) coronal MR imaging, 2) the STN center on a Schaltenbrand-Wahren atlas, 3) targeting based on the AC-PC line, and 4) a composite target based on all 3 methods. The results were compared with the final target (found with help of MER). The combination of 3 methods appeared to be best, while the use of MR imaging as a sole method gave the worst result. Cuny et al.23 also compared 3 methods. The first technique was direct identification on T2-weighted MR imaging (see Table 1). The second and third methods both involved indirect targeting based on the AC-PC line, determined by ventriculography or MR imaging, respectively. The most effective contact was taken as the gold standard. The authors concluded that indirect targeting based on MR imaging worked best, while direct targeting gave the worst outcome. Andrade-Souza et al.3 investigated direct targeting using coronal MR imaging, indirect targeting using the AC-PC line, and a technique using the red nucleus as a landmark. The implantation was optimized using MER, while the most effective contact was identified using post- Fig. 7. Axial susceptibility-weighted MR image showing direct visualization of the STN. operative MR imaging. After comparing the mean distances between the targets and optimal contact, the authors concluded that the red nucleus is a reliable marker voxel by performing multiple MR acquisitions with dif- for STN targeting. This is contradictory to the results of ferent parameters. Bonny et al.12 already reconstructed Danish et al.27 (see Table 1). Breit et al.13 also evaluated T2* maps from their measurements in 2001. Helms et coronal T2-weighted MR imaging, this time in compari- al.45 used T2* maps for improved delineation of iron-rich son with AC-PC line–based targeting using ventriculog- structures, in particular the STN and substantia nigra. raphy. The implantation was refined using MER, and the Lebel et al.51 investigated T2* mapping of the basal gan- actual target was chosen to be the most effective contact. glia at 4.7 T, and obtained increased spatial resolution and Their findings showed that targeting the STN using the sensitivity to iron content. AC-PC line from ventriculography is more accurate than Guo et al.36 were the first to introduce relaxation- direct targeting on MR imaging. Castro et al.18 compared time maps for T1 and T2 into their DBS targeting ap- coronal T2-weighted (IR) MR imaging with indirect tar- plication.37,39 They compared the centers of basal ganglia geting using a registered atlas based on MR imaging data nuclei based on the relaxation-time maps with the coor- with visible STNs. Their results showed an average error dinates derived from the Schaltenbrand-Wahren atlas and of 1.72 mm, which they considered small enough to jus- the actual surgical targets in 15 patients who underwent tify the use of the atlas. surgery. As the mean displacement was 3.21 ± 0.80 mm, In further research, Andrade-Souza et al.4 compared these results indicated the potential of the relaxation-time 2D T2-weighted axial and 3D reconstruction MR imag- maps for DBS targeting. ing. They found that indirect and direct targets based on 3D reconstruction more closely approximate the optimal High Field Strength MR Imaging. Following the ex- 51 contact than targets chosen using 2D MR imaging. How- ample of Lebel et al., more studies have recently been ever, both indirect targets were better than the direct tar- published on direct visualization of the STN and other gets. basal ganglia using MR imaging scanners with high field strengths. Abosch et al.1 acquired susceptibility-weighted Studies That Favor Direct Targeting on MR Imaging. MR imaging data at 7.0 T and showed that the superior Starr et al.82 performed MR imaging and MER to tar- resolution and contrast at this field strength dramatically get the STN. They used a combination of indirect AC-PC improved delineation of the STN. Cho et al.19 also visual- line–based targeting and direct targeting on T2-weighted ized the STN using T2*-weighted scans obtained with a MR imaging. The authors managed to visualize the STN 7-T MR imaging unit, and in addition imaged the sub- directly for 92% of procedures, which supports the choice stantia nigra in 9 healthy controls and 8 patients with Par- for direct imaging instead of AC-PC line–based methods kinson disease, revealing distinct morphological changes for STN targeting. The reliability of indirect methods as due to this disorder.20 compared with direct targeting was also investigated by

976 J Neurosurg / Volume 115 / November 2011 Magnetic resonance imaging for STN visualization

TABLE 1: Summary of indirect studies based on anatomical landmarks visualized on MR images*

No. of Authors & Year Pts Landmarks MRI Protocol Validation Method Outcome Conclusions Schuurman et 15 AC-PC 3D GRE, 1.5 T intraop ventriculog difference in coordinates: Correspondence btwn al., 1999 1.29 mm for difft tar- ventriculog & MRI justi- gets fies use of MRI. Starr et al., 2002 180 AC-PC 3D GRE, 1.5 T MRI: coronal FSE, 1.5 T direct/indirect concordant MRI-based targeting to w/in 1–1.5 mm should be combined w/ MER & postop MRI. Pallavaram et 2 AC-PC T1-weighted, 1.5 T comp of 38 neurosur- mean distance of 1 mm w/ Automatic AC-PC selec- al., 200865 geons same SD tion or direct targeting should be used. Ardekani & 42 AC-PC T1- & T2-weighted, comp of manual & au- avg Euclidean distance of The authors’ model-based Bachman, 1.5 & 3 T tomatic method 1.1 mm algorithm is better than 2009 earlier automatic AC- PC–based methods. Bejjani et al., 12 RN coronal spin echo, MER central track selected 19 MRI targeting is useful for 2000 1.5 T of 24 times STN stimulation but should be combined w/ MER. Cuny et al., 14 RN coronal double echo AC-PC based using distances to optimum tar- Targeting of STN based on 2002 spin echo, 1.5 T MRI/ventriculog; best get: 2.61 mm for AC- RN is less effective than clinical response PC MRI, 3.92 mm for AC-PC–based meth- RN method ods. Danish et al., 26 AC-PC, RN sagittal T1-weighted, comp of distances: SD of MCP-RN & STN- RN is not a reliable marker 2006 axial FSE, 1.5 T MCP-STN, MCP-RN, RN larger than SD of for STN position. STN-RN MCP-STN Pollo et al., 31 AC-PC, RN, IC, 3D MPRAGE & co- Schaltenbrand-Wahren mean target coordinates Combination of different 2003 thalamus, SN ronal IR, 1.5 T atlas located in inf STN landmarks yields a visual procedure to find the STN. Giller et al., 29 nigrocapsular sagittal FSE IR, 1.5 T no quant comp done IC detected in 97% of Indirect sagittal method 2005 angle (IC, slices, SN in 95%, STN using the nigrocapsular SN) in 73% angle can refine STN localization. Lee et al., 2006 6 SMC coronal MPRAGE, RN-based targeting RN based matches atlas The method of this paper FSE, 1.5 T (12, −3, −4 mm); SMC- is sensitive to ventricu- based errors in x lar enlargement, but the (14.25) & z (−5.0) top of the SMC is a reliable landmark. Toda et al., 2009 26 PMC, RN, axial & coronal TSE, RN-based & AC-PC– central track selected in Composite targeting mammillo- 1.5 & 3 T based targeting, MER 81% of 3-T cases, 31% method on 3 T offers thalamic tract of 1.5-T cases reliable STN targeting.

* AC-PC = AC-PC line; avg = average; comp = comparison; difft = different; FSE = fast spin echo; GRE = gradient echo; IC = internal capsule; inf = inferior; IR = inversion recovery; MCP = midcommissural point; MPRAGE = magnetization prepared rapid acquisition gradient echo; PMC = postmammillary commissure; Pts = patients; quant = quantitative; RN = red nucleus; SMC = supramammillary commissure; SN = substantia nigra; TSE = turbo spin echo; ventriculog = ventriculography.

Schlaier et al.78 They determined STN targets in 5 differ- Although the results of all studies described showed ent ways: 1) direct targeting using axial T2-weighted MR superior reliability for direct targeting, the finding was imaging, 2) direct targeting on coronal T2-weighted MR not validated with intra- or postoperative information. imaging, 3) indirect targeting using an axial atlas slice, Koike et al.49 did use the identified STN thickness dur- 4) indirect targeting on a coronal atlas slice, and 5) in- ing MER and clinical parameters, namely the effect of direct targeting using AC-PC references. Direct target- DBS on the disease and on the medication dose, as evalu- ing seemed more reliable than atlas-based targeting, due ation parameters. They compared direct targeting on T2- to large interpatient variability in the STN coordinates weighted FSE MR imaging to the conventional indirect as derived from MR imaging. Ashkan et al.6 confirmed AC-PC line–based method. The results showed a signifi- these results on variability in STN position. Their direct cantly larger mean STN thickness (indicating a longer target differed on average 0.7–1.7 mm in all 3 directions. electrode track through the STN and thus better target-

J Neurosurg / Volume 115 / November 2011 977 E. J. L. Brunenberg et al.

TABLE 2: Summary of indirect studies based on brain atlases*

No. of Validation Authors & Year Pts Atlas Info Registration Method Method Outcome Conclusions Ortega et al., 2008 10 digitized Talairach- nonlinear registra- final microelec- difference in coordinates: System matching the atlas Tournoux atlas tion trode position 1.89 ± 1.00 mm to T1-weighted MRI offers good targeting results. Sánchez Castro et 8 MRI of pt w/ best nonrigid (affine & direct STN target- smallest difference in coor- Automatic STN location us- al., 2005 visible STN nonlinear) ing by expert dinates: 1.80 ± 0.62 mm ing nonrigid atlas registration is possible & accurate. Bardinet et al., 16 postmortem MRI & rigid & affine regis- electrophys re- 81.5% of recordings & 70% Atlas/MRI coregistration 2005 histology data tration cordings (6) of electrodes w/in STN could be used as stan- & final elec- as defined by atlas dard method for targeting trode position deep brain structures. (10) Nowinski et al., 184 ventriculog, MER, linear scaling along STN in Schalten- functional STN (p ≥ 0.5) Functional STN can be 2007 neurol assess- AC-PC brand-Wahren completely inside ana- used for identification of ment atlas tomical STN STN on images. D’Haese et al., 21 MER, stimulation nonlinear registra- preop AC-PC & mean distance: 3.66 mm Automatic prediction of STN 2005 parameters, final tion retrospective for AC-PC–based tar- target from previous tar- electrode posi- automatic targeting, 2.71 mm for auto- gets & electrophys data tions geting matic targeting (rt STN) is an achievable goal. Guo et al., 200739 28 atlas, MER, optimal nonrigid registration actual surgical mean distance smallest Method w/ both anatomical contacts, MRI- target location for combination of all at- & functional data pro- based methods las data: 1.7 ± 0.7 mm vides most reliable STN target.

* Electrophys = electrophysiological; Info = information; neurol = neurological; sig = significant. ing) in the direct targeting group, and clinical parameters ting of images, and the need for preparatory T1-weighted also displayed larger improvements in the direct targeting sequences used to plan the acquisition of T2-weighted group. sequences.10,82 Although all studies mentioned previously in this The comparative studies presented do not provide us section used 1.5-T MR imaging, Acar et al.2 investigated with a straightforward conclusion on the best STN target- the benefit of direct targeting on 3.0-T MR images over ing protocol. Earlier publications tend to favor indirect traditional AC-PC line–based targeting using 1.5-T MR methods, mainly based on the AC-PC line, while authors images. They calculated Euclidean distances between the of more recent studies are more inclined to prefer direct directly and indirectly determined coordinates in 3 di- visualization. Most publications reported the use of 1.5-T mensions, resulting in mean differences between the two MR imaging, and the majority of hospitals still use this locations of 0.45 mm, 0.72 mm, and 0.98 mm in the x, y, field strength, so the recent preference for direct targeting and z axes, respectively. According to the authors, MR is not likely to be due to 3-T MR imaging. The phenom- imaging has advanced to the extent that direct targeting enon might be caused by advances in MR imaging at 1.5 of the STN is no longer imprecise. T, with respect to contrast, noise, and distortion, resulting in a targeting method that is more specific than AC-PC line–based targeting. Due to continuous progress in MR Discussion imaging technology, improvements in direct targeting are still anticipated. In this paper we reviewed indirect and direct STN Although it seems that the future of STN target- targeting methods based on MR imaging. The most com- ing will be focused on patient-specific direct methods, mon indirect methods use either landmarks such as the it is nevertheless important to keep in mind the limita- AC-PC line6,64,65,82 and the red nucleus7,10,23,27 or atlases tions of STN visualization on anatomical MR imaging. built from anatomical8,9,18,31,76 and/or functional24,37,39,59,63 Often only the anterior STN is visible as a hypointense information. These methods are applicable in all cases region.29 In addition, the contrast between the STN and but are not very patient specific. Direct methods already surrounding structures is not optimal, hindering identi- in use clinically comprise T2-weighted FSE10,29,32,83 and fication of the STN boundaries.33,45,48 Furthermore, the IR MR imaging.47,48,86 These techniques can account for question remains whether the visualized STN coincides interpatient variability6,26,55,67,80 but have drawbacks of with the functional target for DBS (often determined by their own, including low contrast and technical issues,44 MER).40,43,55 Promising new methods such as relaxation- such as a long acquisition time,10,66,102 required reformat- time maps and susceptibility-weighted imaging, as well

978 J Neurosurg / Volume 115 / November 2011 Magnetic resonance imaging for STN visualization

TABLE 3: Summary of direct studies using T2-weighted MR imaging and relaxation-time maps*

No. of Authors & Year Pts MRI Protocol Validation Method Outcome Conclusions Starr et al., 1999 6 coronal T2-weighted FSE MER coordinates close to AC-PC– Direct visualization facilitates MRI, 1.5 T based values; initial MER reliable targeting of the track passed through the STN STN, but MER mapping in every case remains necessary. Starr, 2002 44 coronal T2-weighted FSE MER & postop MRI STN directly visualized on MRI Combined approach leads to MRI, 1.5 T in 92% of cases; MER used to consistent electrode place- adapt target intraoperatively ment in the dorsolateral STN. Bejjani et al., 2000 12 coronal T2-weighted MER & optimal contact central track selected for 19 of MRI targeting is useful for spin-echo MRI, 1.5 T position 24 electrodes STN stimulation but should be combined w/ MER. Egidi et al., 2002 13 axial T2-weighted spin- MER & postop MRI targeted track chosen for 22 of 1.5-T MR images are highly echo MRI, 1.5 T 26 electrodes effective for STN targeting. Littlechild et al., 2003 25 axial T2-weighted MRI, MER & optimal contact sig variation in target position (4– The STN varies in position & 1.5 T position 5 mm); mean distance btwn can be accurately targeted target & electrode 1.7 mm from MRI alone. Slavin et al., 2006 13 T2-weighted FSE MRI in MER variation in target position ~2 3 T MRI enables accurate di- 3 directions, 3.0 T mm; all tracks pass through rect visualization of the STN STN. Patel et al., 2002 26, 19† axial & coronal T2- macrostimulation, in- mean target error of 0.3 mm me- MRI-directed targeting of the weighted MRI, 1.5 T traop MRI diolat & 0.4 mm AP STN through guide tubes is accurate & allows direct verification & corrections as necessary. Guo et al., 2005 15 T1 & T2 maps, 1.5 T atlas coordinates, ac- mean displacement 3.21 ± 0.80 The results indicate the po- tual surgical targets mm tential capability of this system to accurately identify the STN.

* AP = anteroposterior. † The authors performed a first series of 26 cases using intraoperative macrostimulation as the control method and a subsequent series of 19 patients with MR imaging–based verification. as imaging at 7 T, have not been validated yet in large ing significant differences in all 3 dimensions. Instead clinical studies. Besides, technical issues strongly influ- of drawing conclusions about which targeting method is ence the MR imaging procedure and quality. Examples more suitable, the authors stated that apparently it would are the MR compatibility of the stereotactic frame, im- be better to conclude that the indirect AC-PC line–based age artifacts due to patient movement, and magnetic field target does not coincide with the center of the STN as distortion, although distortion can be controlled reason- visualized on anatomical MR imaging. Moreover, vari- ably for the midbrain.26,57 Because of these issues, further ous follow-up42,50,54,75 and postmortem84 studies state that research into STN localization methods, both direct and the best clinical target is located in the dorsolateral part indirect, seems necessary. of the STN and the area just superior to the STN (zona The comparative studies that have validated the tar- incerta, field of Forel). geting with the most effective contact location also have The results from several papers show that stimulating their limitations. The conclusions of Cuny et al.,23 An- the dorsolateral motor part of the STN is more effective drade-Souza et al.3,4 and Breit et al.,13 that indirect target- than stimulating the center.17,42,50,54,75,84 As conventional ing is more relevant for electrode placement, should thus direct targeting on anatomical MR imaging cannot dis- be considered with care. Because the primary targeting in tinguish the dorsolateral STN due to lack of contrast, it these studies was indirect, the final contact, even after in- is useful to further investigate the use of other MR imag- traoperative adjustment, was close to the original indirect ing–based methods to facilitate specific targeting of this target. In addition, analyzing indirect versus direct tar- area of the STN. Examples of techniques that have not geting according to the optimal contact only emphasizes yet proven their clinical utility but might aid in STN lo- differences between the final position and the primary calization are the use of landmarks such as the nigrocap- targets. This bias would of course also exist if primary sular angle,34 the mammillary commissures,52,91 and the targeting were to be performed using a direct method. line connecting the mammillary bodies.73 In addition, a Caire et al.17 also compared STN localization methods fast-developing line of research focuses on susceptibility- based on the AC-PC line and 1.5-T MR imaging, reveal- weighted MR imaging and T2* mapping to improve the

J Neurosurg / Volume 115 / November 2011 979 E. J. L. Brunenberg et al.

TABLE 4: Summary of studies comparing indirect and direct methods*

No. of Authors & Year Pts Methods Validation Method Outcome Conclusions Zonenshayn et al., 15 coronal MRI (T2-weighted/ final target based on distance btwn MRI/final target: Combination of 3 targeting 2000 IR), atlas, AC-PC, com- MER 2.6 ± 1.3 mm; btwn composite/ methods offers best correla- bination final target: 1.3 ± 1.1 mm tion w/ the final target. Cuny et al., 2002 14 T2-weighted MRI, AC-PC most effective contact distance btwn AC-PC (MRI)/ Indirect targeting based on MRI w/ ventriculog, AC-PC best contact: 2.6 ± 1.2 mm; works best; direct targeting w/ MRI btwn direct/best contact: 3.9 gives worst outcome. ± 1.9 mm Andrade-Souza et 14 coronal MRI, AC-PC, RN- optimal contact (MER distances to optimal contact: RN is a reliable marker to ap- al., 20053 based targeting & postop MRI) MRI: 4.7 mm; AC-PC: 3.4 mm; proximate the optimal contact RN: 3.2 mm position. Breit et al., 2006 30 coronal T2-weighted most effective contact mean targeting error 4.1 ± 1.7 AC-PC using ventriculog is MRI, AC-PC based on mm for MRI & 2.4 ± 1.1 mm more accurate than direct ventriculog for AC-PC targeting on MRI. Castro et al., 2006 39 coronal T2-weighted IR comp in MRI coordi- best atlas registration method (B- Automatic STN targeting is pos- MRI, anatomical atlas nate system splines): avg error 1.72 ± 0.48 sible & as accurate as current based on direct targets mm expert methods. Andrade-Souza et 14 axial T2-weighted MRI, optimal contact (MER distances to optimal contact: 2D 3D reconstr leads to a better ap- al., 20054 3D reconstr (both direct & postop MRI) MRI direct: 4.7 mm; 3D re- proximation of the optimal & indirect) constr direct: 3.5 mm; 2D MRI contact; indirect methods are indirect: 3.4 mm; 3D reconstr better than direct targets. indirect: 2.6 mm Starr, 2002 44 coronal T2-weighted FSE comp in atlas coordi- in 34% the STN lies >1 mm more Direct targeting accounts better MRI, AC-PC nate system lateral than 12 mm & in 40% for interpatient variability than the STN lies >1 mm lower than indirect targeting. 4 mm Schlaier et al., 2005 14 axial/coronal T2-weighted comp in atlas coordi- all STN coordinates displayed a Direct targeting is more reliable MRI & atlas, AC-PC nate system range of 4–5 mm due to large interpatient variability. Ashkan et al., 2007 29 axial T2-weighted MRI, comp in atlas coordi- direct: 1.7 mm more medial, 0.7 Direct targeting is more ac- AC-PC nate system mm more anterior, 0.7 mm curate due to variability in more ventral STN position. Koike et al., 2008 44 T2-weighted FSE MRI MER & clinical param- direct group: longer track through Direct targeting w/ single-track (using RN), AC-PC eters STN & larger clinical improve- recording can be standard ment for DBS. Acar et al., 2007 20 axial T2-weighted FSE comp in atlas coordi- mean distances were 0.45 mm, Direct targeting is no longer im- MRI (3T), AC-PC (1.5T) nate system 0.72 mm & 0.98 mm in x, y, z practical and/or imprecise. directions Caire et al., 2009 22 coronal T2-weighted TSE comp of STN center sig differences in coordinates in MRI- & AC-PC–based targets MRI, AC-PC coordinates all 3 directions do not coincide.

* Reconstr = reconstruction. contrast in the STN region.45,51,94,97,99 It is unlikely, how- shown that different parts of the STN in a rat brain could ever, that this improved contrast would give rise to identi- be separated visually and automatically.14,16 Recently, fication of the dorsolateral STN specifically. Coenen et al.21 also published a case study on diffusion In addition to trying to identify the dorsolateral STN, tensor imaging–based fiber anatomy in the STN region, possibilities of dividing the STN functionally and identi- to help identification of the tremor suppression target. fying the motor part should also be investigated. Modali- Apart from the primary targeting based on (MR) im- ties such as functional or diffusion-weighted MR imaging aging, there are other parts of the surgical “pipeline” that could yield features that facilitate separation of the motor influence the clinical outcome of a DBS procedure and part of the STN. Diffusion-weighted MR imaging sen- should not be ignored. The most important are the type sitizes the MR acquisition to water diffusion in specific of stereotactic frame, the selection of the trajectory (this directions. The diffusion profile in a voxel can be fitted can be done manually or automatically15), registration of as an ellipse (diffusion tensor imaging58) or a higher or- different imaging modalities, intraoperative brain shift,46 der shape (high angular resolution diffusion imaging93). intraoperative electrode adjustment, and postoperative By analyzing these profiles in the STN itself, it has been parameter estimation. In striving for the best possible re-

980 J Neurosurg / Volume 115 / November 2011 Magnetic resonance imaging for STN visualization sults of DBS, all these factors should be given appropriate atlas of the human basal ganglia. II. Atlas deformation strat- attention. egy and evaluation in deep brain stimulation for Parkinson disease. Clinical article. J Neurosurg 110:208–219, 2009 9. Bardinet E, Dormont D, Malandain G, Bhattacharjee M, Conclusions Pidoux B, Saleh C, et al: Retrospective cross-evaluation of a histological and deformable 3D atlas of the basal ganglia on This review gave a systematic overview of studies series of Parkinsonian patients treated by deep brain stimula- on noninvasive STN targeting. The studies either pre- tion. Med Image Comput Comput Assist Interv 8(Pt 2): sented a single indirect or direct targeting method, or 385–393, 2005 compared different techniques. Because each method 10. Bejjani BP, Dormont D, Pidoux B, Yelnik J, Damier P, Arnulf has been shown to have its own benefits but also specific I, et al: Bilateral subthalamic stimulation for Parkinson’s disease by using three-dimensional stereotactic magnetic reso- drawbacks, it is not possible to reach a straightforward nance imaging and electrophysiological guidance. J Neuro- conclusion on which targeting technique should be used. surg 92:615–625, 2000 Therefore, identification of the dorsolateral and motor 11. Benabid AL, Koudsie A, Benazzouz A, Le Bas JF, Pollak P: part of the STN remains an open question, although it Imaging of subthalamic nucleus and ventralis intermedius of is expected that future advances in MRI technology will the thalamus. Mov Disord 17 (3 Suppl 3):S123–S129, 2002 help to find an answer to this question. 12. Bonny J, Durif F, Bazin JE, Touraille E, Yelnik J, Renou JP: Contrast optimization of Macaca mulatta basal ganglia in magnetic resonance images at 4.7 Tesla. J Neurosci Methods Disclosure 107:25–30, 2001 13. Breit S, LeBas JF, Koudsie A, Schulz J, Benazzouz A, Pol- The authors report no conflict of interest concerning the mate- lak P, et al: Pretargeting for the implantation of stimulation rials or methods used in this study or the findings specified in this electrodes into the subthalamic nucleus: a comparative study paper. Ellen Brunenberg’s research is supported by the Netherlands of magnetic resonance imaging and ventriculography. Neuro- Organization for Scientific Research (NWO TopTalent grant). surgery 58 (1 Suppl):ONS83–ONS95, 2006 Author contributions to the study and manuscript preparation 14. Brunenberg E, Pelgrim E, ter Haar Romeny B, Platel B: k- include the following. Conception and design: Brunenberg, Platel, means and graph cuts clustering of diffusion MRI in rat STN. Visser-Vandewalle. Acquisition of data: Brunenberg. Analysis Proc Intl Soc Mag Reson Med 18:4045, 2010 (Abstract) and interpretation of data: Brunenberg, Platel. Drafting the article: 15. Brunenberg EJL, Vilanova A, Visser-Vandewalle V, Temel Y, Brunenberg. Critically revising the article: all authors. Reviewed Ackermans L, Platel B, et al: Automatic trajectory planning submitted version of manuscript: all authors. Approved the final ver- for deep brain stimulation: a feasibility study. Med Image sion of the manuscript on behalf of all authors: Brunenberg. Study Comput Comput Assist Interv 10(Pt 1):584–592, 2007 supervision: ter Haar Romeny, Visser-Vandevalle. 16. Brunenberg EJL, Prckovska V, Platel B, Strijkers GJ, ter Haar Romeny BM: Untangling a fiber bundle knot: preliminary results on STN connectivity using DTI and HARDI on rat Proc Intl Soc Mag Reson Med 17: References brains. 741, 2009 (Ab- stract) 1. Abosch A, Yacoub E, Ugurbil K, Harel N: An assessment of 17. Caire F, Ouchchane L, Coste J, Gabrillargues J, Derost P, Ulla current brain targets for deep brain stimulation surgery with M, et al: Subthalamic nucleus location: relationships between susceptibility-weighted imaging at 7 tesla. Neurosurgery 67: stereotactic AC-PC-based diagrams and MRI anatomy-based 1745–1756, 2010 contours. Stereotact Funct Neurosurg 87:337–347, 2009 2. Acar F, Miller JP, Berk MC, Anderson G, Burchiel KJ: Safety 18. Castro FJ, Pollo C, Meuli R, Maeder P, Cuisenaire O, Cuadra of anterior commissure-posterior commissure-based target MB, et al: A cross validation study of deep brain stimulation calculation of the subthalamic nucleus in functional stereo- targeting: from experts to atlas-based, segmentation-based tactic procedures. Stereotact Funct Neurosurg 85:287–291, and automatic registration algorithms. IEEE Trans Med Im- 2007 aging 25:1440–1450, 2006 3. Andrade-Souza YM, Schwalb JM, Hamani C, Eltahawy H, 19. Cho ZH, Min HK, Oh SH, Han JY, Park CW, Chi JG, et al: Hoque T, Saint-Cyr J, et al: Comparison of three methods of Direct visualization of deep brain stimulation targets in Par- targeting the subthalamic nucleus for chronic stimulation in kinson disease with the use of 7-tesla magnetic resonance im- Parkinson’s disease. Neurosurgery 56 (2 Suppl):360–368, aging. Clinical article. J Neurosurg 113:639– 647, 2010 2005 20. Cho ZH, Oh SH, Kim JM, Park SY, Kwon DH, Jeong HJ, et al: 4. Andrade-Souza YM, Schwalb JM, Hamani C, Hoque T, Saint- Direct visualization of Parkinson’s disease by in vivo human Cyr J, Lozano AM: Comparison of 2-dimensional magnetic brain imaging using 7.0T magnetic resonance imaging. Mov resonance imaging and 3-planar reconstruction methods for Disord 26:713–718, 2011 targeting the subthalamic nucleus in Parkinson disease. Surg 21. Coenen VA, Mädler B, Schiffbauer H, Urbach H, Allert N: Neurol 63:357–363, 2005 Individual fiber anatomy of the subthalamic region revealed 5. Ardekani BA, Bachman AH: Model-based automatic detec- with DTI—a concept to identify the DBS target for tremor tion of the anterior and posterior commissures on MRI scans. suppression: erratum. Neurosurgery 68:1780–1781, 2011 Neuroimage 46:677–682, 2009 22. Collins DL, Zijdenbos AP, Kollokian V, Sled JG, Kabani NJ, 6. Ashkan K, Blomstedt P, Zrinzo L, Tisch S, Yousry T, Limou- Holmes CJ, et al: Design and construction of a realistic digi- sin-Dowsey P, et al: Variability of the subthalamic nucleus: tal brain phantom. IEEE Trans Med Imaging 17:463–468, the case for direct MRI guided targeting. Br J Neurosurg 1998 21:197–200, 2007 23. Cuny E, Guehl D, Burbaud P, Gross C, Dousset V, Rougier A: 7. Aziz TZ, Nandi D, Parkin S, Liu X, Giladi N, Bain P, et al: Lack of agreement between direct magnetic resonance imag- Targeting the subthalamic nucleus. Stereotact Funct Neuro- ing and statistical determination of a subthalamic target: the surg 77:87–90, 2001 role of electrophysiological guidance. J Neurosurg 97:591– 8. Bardinet E, Bhattacharjee M, Dormont D, Pidoux B, Malan- 597, 2002 dain G, Schüpbach M, et al: A three-dimensional histological 24. D’Haese PF, Cetinkaya E, Konrad PE, Kao C, Dawant BM:

J Neurosurg / Volume 115 / November 2011 981 E. J. L. Brunenberg et al.

Computer-aided placement of deep brain stimulators: from 42. Hamel W, Fietzek U, Morsnowski A, Schrader B, Herzog J, planning to intraoperative guidance. IEEE Trans Med Im- Weinert D, et al: Deep brain stimulation of the subthalamic aging 24:1469–1478, 2005 nucleus in Parkinson’s disease: evaluation of active electrode 25. Daniluk S, Davies K: Optimal stimulation site. J Neurosurg contacts. J Neurol Neurosurg Psychiatry 74:1036–1046, 108:425–429, 2008 (Letter) 2003 26. Daniluk S, Davies KG, Ellias SA, Novak P, Nazzaro JM: As- 43. Hamid NA, Mitchell RD, Mocroft P, Westby GWM, Milner sessment of the variability in the anatomical position and size J, Pall H: Targeting the subthalamic nucleus for deep brain of the subthalamic nucleus among patients with advanced stimulation: technical approach and fusion of pre- and post- Parkinson’s disease using magnetic resonance imaging. Acta operative MR images to define accuracy of lead placement. J Neurochir (Wien) 152:201–210, 2010 Neurol Neurosurg Psychiatry 76:409–414, 2005 27. Danish SF, Jaggi JL, Moyer JT, Finkel L, Baltuch GH: Con- 44. Hariz MI, Krack P, Melvill R, Jorgensen JV, Hamel W, Hira- ventional MRI is inadequate to delineate the relationship be- bayashi H, et al: A quick and universal method for stereotac- tween the red nucleus and subthalamic nucleus in Parkinson’s tic visualization of the subthalamic nucleus before and after disease. Stereotact Funct Neurosurg 84:12–18, 2006 implantation of deep brain stimulation electrodes. Stereotact 28. Davies KG, Daniluk S: Stereotactic targeting of the subtha- Funct Neurosurg 80:96–101, 2003 lamic nucleus: relevance of magnetic resonance-based evalu- 45. Helms G, Gringel T, Knauth M, Dechent P, Elolf E: Delin- ation of interindividual variation in diencephalic anatomy. eation of the subthalamic nucleus (STN) on high-resolution Stereotact Funct Neurosurg 86:330–331, 2008 maps of R2*. Proc Intl Soc Mag Reson Med 16:880, 2008 29. Dormont D, Ricciardi KG, Tandé D, Parain K, Menuel C, (Abstract) Galanaud D, et al: Is the subthalamic nucleus hypointense on 46. Hunsche S, Sauner D, Maarouf M, Poggenborg J, Lackner K, T2-weighted images? A correlation study using MR imaging Sturm V, et al: Intraoperative X-ray detection and MRI-based and stereotactic atlas data. AJNR Am J Neuroradiol 25: quantification of brain shift effects subsequent to implanta- 1516–1523, 2004 tion of the first electrode in bilateral implantation of deep 30. Dormont D, Seidenwurm D, Galanaud D, Cornu P, Yelnik J, brain stimulation electrodes. Stereotact Funct Neurosurg Bardinet E: Neuroimaging and deep brain stimulation. AJNR 87:322–329, 2009 Am J Neuroradiol 31:15–23, 2010 47. Ishimori T, Nakano S, Mori Y, Seo R, Togami T, Masada T, 31. Duay V, Bresson X, Castro JS, Pollo C, Cuadra MB, Thiran et al: Preoperative identification of subthalamic nucleus for JP: An active contour-based atlas registration model applied deep brain stimulation using three-dimensional phase sensi- to automatic subthalamic nucleus targeting on MRI: method tive inversion recovery technique. Magn Reson Med Sci 6: and validation. Med Image Comput Comput Assist Interv 225–229, 2007 11(Pt 2):980–988, 2008 48. Kitajima M, Kakeda S, Moriya J, Ohnari N, Sato T, Hayashida 32. Egidi M, Rampini P, Locatelli M, Farabola M, Priori A, Y, et al: Direct visualization of normal subthalamic nucleus Pesenti A, et al: Visualisation of the subthalamic nucleus: a with high resolution MR imaging at 3.0 T: comparison be- multiple sequential image fusion (MuSIF) technique for di- tween FSE T2-WI and fast STIR image. Proc Intl Soc Mag rect stereotaxic localisation and postoperative control. Neurol Reson Med 15:780, 2007 (Abstract) Sci 23 (2 Suppl 2):S71–S72, 2002 49. Koike Y, Shima F, Nakamizo A, Miyagi Y: Direct localization 33. Elolf E, Bockermann V, Gringel T, Knauth M, Dechent P, of subthalamic nucleus supplemented by single-track elec- Helms G: Improved visibility of the subthalamic nucleus on trophysiological guidance in deep brain stimulation lead im- high-resolution stereotactic MR imaging by added suscepti- plantation: techniques and clinical results. Stereotact Funct bility (T2*) contrast using multiple gradient echoes. AJNR Neurosurg 86:173–178, 2008 Am J Neuroradiol 28:1093–1094, 2007 50. Lanotte MM, Rizzone M, Bergamasco B, Faccani G, Mel- 34. Giller CA, Babcock EE, Mendelsohn DB: Use of sagittal im- carne A, Lopiano L: Deep brain stimulation of the subtha- ages for localization of the subthalamic nucleus. Technical lamic nucleus: anatomical, neurophysiological, and outcome note. J Neurosurg 102:571–575, 2005 correlations with the effects of stimulation. J Neurol Neuro- 35. Guehl D, Edwards R, Cuny E, Burbaud P, Rougier A, Modolo surg Psychiatry 72:53–58, 2002 J, et al: Statistical determination of the optimal subthalamic 51. Lebel RM, Eissa A, Gee M, Wieler M, Martin WR, Wilman nucleus stimulation site in patients with Parkinson disease. J AH: MRI methods at 4.7 T for imaging Parkinson’s disease. Neurosurg 106:101–110, 2007 Proc Intl Soc Mag Reson Med 16:2283, 2008 (Abstract) 36. Guo T, Deoni SCL, Finnis KW, Parrent AG, Peters TM: Ap- 52. Lee C, Young B, Sanders MF: The role of the supramammilla- plication of T1 and T2 maps for stereotactic deep-brain neu- ry commissure in MR localization of the subthalamic nucleus. rosurgery planning. Conf Proc IEEE Eng Med Biol Soc 5: Stereotact Funct Neurosurg 84:193–204, 2006 5416–5419, 2005 53. Lemaire JJ, Coste J, Ouchchane L, Caire F, Nuti C, Derost P, 37. Guo T, Finnis KW, Deoni SC, Parrent AG, Peters TM: Com- et al: Brain mapping in stereotactic surgery: a brief overview parison of different targeting methods for subthalamic nucle- from the probabilistic targeting to the patient-based anatomic us deep brain stimulation. Med Image Comput Comput As- mapping. Neuroimage 37 (1 Suppl 1):S109–S115, 2007 sist Interv 9(Pt 1):768–775, 2006 54. Lemaire JJ, Coste J, Ouchchane L, Hemm S, Derost P, Ulla 38. Guo T, Parrent AG, Peters TM: Automatic target and trajec- M, et al: MRI anatomical mapping and direct stereotactic tar- tory identification for deep brain stimulation (DBS) proce- geting in the subthalamic region: functional and anatomical dures. Med Image Comput Comput Assist Interv 10(Pt 1): correspondence in Parkinson’s disease. Int J CARS 2:75–85, 483–490, 2007 2007 39. Guo T, Parrent AG, Peters TM: Surgical targeting accuracy 55. Littlechild P, Varma TRK, Eldridge PR, Fox S, Forster A, analysis of six methods for subthalamic nucleus deep brain Fletcher N, et al: Variability in position of the subthalamic stimulation. Comput Aided Surg 12:325–334, 2007 nucleus targeted by magnetic resonance imaging and micro- 40. Hamani C, Richter EO, Andrade-Souza Y, Hutchison W, electrode recordings as compared to atlas co-ordinates. Ste- Saint-Cyr JA, Lozano AM: Correspondence of microelec- reotact Funct Neurosurg 80:82–87, 2003 trode mapping with magnetic resonance imaging for subtha- 56. Liu X, Rowe J, Nandi D, Hayward G, Parkin S, Stein J, et lamic nucleus procedures. Surg Neurol 63:249–253, 2005 al: Localisation of the subthalamic nucleus using Radionics 41. Hamani C, Saint-Cyr JA, Fraser J, Kaplitt M, Lozano AM: Image Fusion and Stereoplan combined with field potential The subthalamic nucleus in the context of movement disor- recording. A technical note. Stereotact Funct Neurosurg 76: ders. Brain 127:4–20, 2004 63–73, 2001

982 J Neurosurg / Volume 115 / November 2011 Magnetic resonance imaging for STN visualization

57. Menuel C, Garnero L, Bardinet E, Poupon F, Phalippou D, Vandersteen M, Adriaensen P, et al: The microanatomical en- Dormont D: Characterization and correction of distortions vironment of the subthalamic nucleus. Technical note. J Neu- in stereotactic magnetic resonance imaging for bilateral sub- rosurg 107:198–201, 2007 thalamic stimulation in Parkinson disease. J Neurosurg 103: 74. Rodriguez-Oroz MC, Zamarbide I, Guridi J, Palmero MR, 256–266, 2005 Obeso JA: Efficacy of deep brain stimulation of the subtha- 58. Mori S, Zhang J: Principles of diffusion tensor imaging and its lamic nucleus in Parkinson’s disease 4 years after surgery: applications to basic neuroscience research. Neuron 51:527– double blind and open label evaluation. J Neurol Neurosurg 539, 2006 Psychiatry 75:1382–1385, 2004 59. Nowinski WL, Belov D, Pollak P, Benabid AL: A probabilistic 75. Saint-Cyr JA, Hoque T, Pereira LCM, Dostrovsky JO, Hutchi- functional atlas of the human subthalamic nucleus. Neuroin- son WD, Mikulis DJ, et al: Localization of clinically effec- formatics 2:381–398, 2004 tive stimulating electrodes in the human subthalamic nucleus 60. Nowinski WL, Liu J, Thirunavuukarasuu A: Quantification on magnetic resonance imaging. J Neurosurg 97:1152–1166, and visualization of the three-dimensional inconsistency of 2002 the subthalamic nucleus in the Schaltenbrand-Wahren brain 76. Sánchez Castro FJ, Pollo C, Villemure JG, Thiran JP: Auto- atlas. Stereotact Funct Neurosurg 84:46–55, 2006 matic subthalamic nucleus targeting for deep brain stimula- 61. Nowinski WL, Thirunavuukarasuu A, Liu J, Benabid AL: tion. A validation study, in: International Congress Series Correlation between the anatomical and functional human (CARS 2005: Computer Assisted Radiology and Surgery). subthalamic nucleus. Stereotact Funct Neurosurg 85:88–93, Amsterdam: Elsevier, 2005, Vol 1281, pp 804–809 2007 77. Schaltenbrand G, Wahren W: Atlas for Stereotaxy of the Hu- 62. Ortega M, Juan MC, Alcañiz M, Gil JA, Monserrat C: De- man Brain, ed 2. Stuttgart: Thieme, 1977 formable brain atlas validation of the location of subthalamic 78. Schlaier J, Schoedel P, Lange M, Winkler J, Warnat J, Doren- nucleus using T1-weighted MR images of patients operated on beck U, et al: Reliability of atlas-derived coordinates in deep for Parkinson’s. Comput Med Imaging Graph 32:367–378, brain stimulation. Acta Neurochir (Wien) 147:1175–1180, 2008 2005 63. Pallavaram S, D’Haese PF, Kao C, Yu H, Remple M, Neimat J, 79. Schuurman PR, de Bie RM, Majoie CB, Speelman JD, Bosch et al: A new method for creating electrophysiological maps for DA: A prospective comparison between three-dimensional DBS surgery and their application to surgical guidance. Med magnetic resonance imaging and ventriculography for target- Image Comput Comput Assist Inter 11(Pt 1):670–677, coordinate determination in frame-based functional stereo- 2008 tactic neurosurgery. J Neurosurg 91:911–914, 1999 64. Pallavaram S, Dawant BM, Koyama T, Yu H, Neimat J, Kon- 80. Slavin KV, Thulborn KR, Wess C, Nersesyan H: Direct vi- rad PE, et al: Validation of a fully automatic method for the sualization of the human subthalamic nucleus with 3T MR AJNR Am J Neuroradiol 27: routine selection of the anterior and posterior commissures imaging. 80–84, 2006 in magnetic resonance images. Stereotact Funct Neurosurg 81. Starr PA: Placement of deep brain stimulators into the subthalamic nucleus or globus pallidus internus: technical approach. 87:148–154, 2009 Stereotact Funct Neurosurg 79:118–145, 2002 65. Pallavaram S, Yu H, Spooner J, D’Haese PF, Bodenheimer B, 82. Starr PA, Christine CW, Theodosopoulos PV, Lindsey N, Konrad PE, et al: Intersurgeon variability in the selection of Byrd D, Mosley A, et al: Implantation of deep brain stimu- anterior and posterior commissures and its potential effects lators into the subthalamic nucleus: technical approach and on target localization. Stereotact Funct Neurosurg 86:113– magnetic resonance imaging-verified lead locations.J Neuro- 119, 2008 surg 97:370–387, 2002 66. Patel NK, Heywood P, O’Sullivan K, Love S, Gill SS: MRI- 83. Starr PA, Vitek JL, DeLong M, Bakay RA: Magnetic reso- directed subthalamic nucleus surgery for Parkinson’s disease. nance imaging-based stereotactic localization of the globus Stereotact Funct Neurosurg 78:132–145, 2002 pallidus and subthalamic nucleus. Neurosurgery 44:303– 67. Patel NK, Khan S, Gill SS: Comparison of atlas- and mag- 314, 1999 netic-resonance-imaging-based stereotactic targeting of the 84. Sun DA, Yu H, Spooner J, Tatsas AD, Davis T, Abel TW, et subthalamic nucleus in the surgical treatment of Parkinson’s al: Postmortem analysis following 71 months of deep brain disease. Stereotact Funct Neurosurg 86:153–161, 2008 stimulation of the subthalamic nucleus for Parkinson disease. 68. Patel NK, Plaha P, O’Sullivan K, McCarter R, Heywood P, Case report. J Neurosurg 109:325–329, 2008 Gill SS: MRI directed bilateral stimulation of the subthalamic 85. Talairach J, Tournoux P: Co-planar Stereotaxic Atlas of the nucleus in patients with Parkinson’s disease. J Neurol Neuro- Human Brain. New York: Thieme, 1988 surg Psychiatry 74:1631–1637, 2003 86. Taoka T, Hirabayashi H, Nakagawa H, Sakamoto M, Kitano 69. Pollak P, Benabid AL, Gross C, Gao DM, Laurent A, Benaz- S, Takahama J, et al: “Sukeroku sign” and “dent internal-cap- zouz A, et al: [Effects of the stimulation of the subthalamic sule sign”—identification guide for targeting the subthalamic nucleus in Parkinson disease.] Rev Neurol (Paris) 149:175– nucleus for placement of deep brain stimulation electrodes. 176, 1993 (Fr) Neuroradiology 51:11–16, 2009 70. Pollo C, Meuli R, Maeder P, Vingerhoets F, Ghika J, Ville- 87. Taoka T, Kubota Y, Hirabayashi H, Nakagawa H, Sakamoto mure JG: Subthalamic nucleus deep brain stimulation for Par- M, Iwasaki S, et al: Signal and shape of subthalamic nucleus kinson’s disease: magnetic resonance imaging targeting using on susceptibility-weighted image (SWI): comparison with visible anatomical landmarks. Stereotact Funct Neurosurg STIR image. Proc Intl Soc Mag Reson Med 15:2175, 2007 80:76–81, 2003 (Abstract) 71. Rauscher A, Witoszynskyj S, Kolind S, Coenen V, Li D: Vi- 88. Temel Y, Blokland A, Steinbusch HWM, Visser-Vandewalle sualization of the subthalamic nuclei at high spatial resolution V: The functional role of the subthalamic nucleus in cognitive and high contrast with susceptibility weighted phase imaging. and limbic circuits. Prog Neurobiol 76:393–413, 2005 Proc Intl Soc Mag Reson Med 16:879, 2008 (Abstract) 89. Temel Y, Kessels A, Tan S, Topdag A, Boon P, Visser-Vande- 72. Richter EO, Hoque T, Halliday W, Lozano AM, Saint-Cyr JA: walle V: Behavioural changes after bilateral subthalamic Determining the position and size of the subthalamic nucle- stimulation in advanced Parkinson disease: a systematic re- us based on magnetic resonance imaging results in patients view. Parkinsonism Relat Disord 12:265–272, 2006 with advanced Parkinson disease. J Neurosurg 100:541–546, 90. Temel Y, Prinsenberg T, Visser-Vandewalle V: Imaging of the 2004 subthalamic nucleus for deep brain stimulation: a systematic 73. Rijkers K, Temel Y, Visser-Vandewalle V, Vanormelingen L, review. Neuromodulation 11:8–12, 2008

J Neurosurg / Volume 115 / November 2011 983 E. J. L. Brunenberg et al.

91. Toda H, Sawamoto N, Hanakawa T, Saiki H, Matsumoto S, 99. Wu ML, Young GS, Chen NK: Midbrain nuclei visualization Okumura R, et al: A novel composite targeting method using improved by susceptibility-enhanced 3D multi-echo SSFP for high-field magnetic resonance imaging for subthalamic nucleus deep brain stimulation guidance. Proc Intl Soc Mag Reson deep brain stimulation. Clinical article. J Neurosurg 111:737– Med 18:701, 2010 (Abstract) 745, 2009 100. Yelnik J, Bardinet E, Dormont D, Malandain G, Ourselin S, 92. Tsai ST, Lin SH, Lin SZ, Chen JY, Lee CW, Chen SY: Tandé D, et al: A three-dimensional, histological and deformable Neuropsychological effects after chronic subthalamic stimula- atlas of the human basal ganglia. I. Atlas construction based on tion and the topography of the nucleus in Parkinson’s disease. immunohistochemical and MRI data. Neuroimage 34:618–638, Neurosurgery 61:E1024–E1030, 2007 2007 93. Tuch DS, Reese TG, Wiegell MR, Makris N, Belliveau JW, 101. Young GS, Chen NK: High contrast susceptibility weighted Wedeen VJ: High angular resolution diffusion imaging reveals imaging: Reliable Unwrapping Susceptibility Technique (RUST intravoxel white matter fiber heterogeneity. Magn Reson Med SWI) improved visualization of midbrain nuclei for deep brain 48:577–582, 2002 stimulation. Proc Intl Soc Mag Reson Med 15: 937, 2007 94. Vertinsky AT, Coenen VA, Lang DJ, Kolind S, Honey CR, Li (Abstract) D, et al: Localization of the subthalamic nucleus: optimization 102. Zhu XL, Hamel W, Schrader B, Weinert D, Hedderich J, Herzog with susceptibility-weighted phase MR imaging. AJNR Am J J, et al: Magnetic resonance imaging-based morphometry and Neuroradiol 30:1717–1724, 2009 landmark correlation of basal ganglia nuclei. Acta Neurochir 95. Visser-Vandewalle V, van der Linden C, Temel Y, Celik H, (Wien) 144:959–969, 2002 Ackermans L, Spincemaille G, et al: Long-term effects of bilat- 103. Zonenshayn M, Rezai AR, Mogilner AY, Beric A, Sterio D, eral subthalamic nucleus stimulation in advanced Parkinson dis- Kelly PJ: Comparison of anatomic and neurophysiological ease: a four year follow-up study. Parkinsonism Relat Disord methods for subthalamic nucleus targeting. Neurosurgery 47: 11:157–165, 2005 282–294, 2000 96. Volkmann J, Allert N, Voges J, Weiss PH, Freund HJ, Sturm V: Safety and efficacy of pallidal or subthalamic nucleus stimulation in advanced PD. Neurology 56:548–551, 2001 Manuscript submitted September 9, 2010. 97. Volz S, Hattingen E, Preibisch C, Gasser T, Deichmann R: Accepted June 8, 2011. Reduction of susceptibility-induced signal losses in multi-gra- Please include this information when citing this paper: pub- dient-echo images: application to improved visualization of the lished online July 29, 2011; DOI: 10.3171/2011.6.JNS101571. subthalamic nucleus. Neuroimage 45:1135–1143, 2009 Address correspondence to: Ellen J. L. Brunenberg, Eindhoven 98. Voon V, Kubu C, Krack P, Houeto JL, Tröster AI: Deep brain University of Technology, Department of Biomedical Engineering, stimulation: neuropsychological and neuropsychiatric issues. WH 2.106, PO Box 513, 5600 MB Eindhoven, The Netherlands. Mov Disord 21 (Suppl 14):S305–S327, 2006 email: [email protected].

984 J Neurosurg / Volume 115 / November 2011