Waveform Recognition

Andrei V. Alexandrov, MD Semmes-Murphey Professor and Chairman Department of Neurology The University of Tennessee Health Science Center Memphis, TN The Journal for Vascular Ultrasound 36(2):1–10, 2012 Extra- and Intracranial Waveform Analysis Algorithm, Descriptions, Classiﬁ cations, and Differential Diagnosis

Andrei V. Alexandrov, MD, RVT

ABSTRACT Waveform recognition is an essential step in the interpretation of vascular ultrasound studies. Waveforms can vary widely as the result of systemic and focal hemodynamic changes in the precerebral and particularly intracranial vessels. Waveform pattern recognition can aid or refi ne the application of strict velocity criteria, and this review will focus on typical normal and abnormal waveforms. Algorithms for waveform analysis and interpretation of extra- and intracranial waveforms are Thediscussed Journal for Vascular alongUltrasound 36(2):1–10,with 2012 correlative imaging and clinical fi ndings. Wave- Extra- andform Intracranial descriptions andWaveform classifi cations Analysis based on prediction of vessel patency, systemic versus local Algorithm,circulatory Descriptions, conditions, Classifiand resistance cations, to fl ow are provided for specifi c clinical conditions. Inter- pretation of waveforms rests on our ability to integrate systemic and cerebral hemodynamics acrossand Difa wideferential spectrum Diagnosis of conditions. Intracranial vasculature poses unique challenges yet may yield newAndrei knowledge V. Alexandrov, MD, if RourVT technologies are able to quantify blood fl ow beyond just velocities in the future. ABSTRACT Waveform recognition is an essential step in the interpretation of vascular ultrasound studies. Waveforms can vary widely as the result of systemic and focal hemodynamic changes in the precerebral and particularly intracranial vessels. Waveform pattern recognition can aid or refi ne the application of strict velocity criteria, and this review will focus on typical normal and abnormal waveforms. Algorithms for waveform analysis and interpretation of extra- and intracranial waveforms are discussed along with correlative imaging and clinical fi ndings. Wave- form descriptions and classifi cations based on prediction of vessel patency, systemic versus local circulatory conditions, and resistance to fl ow are provided for specifi c clinical conditions. Inter- “Dopplerpretation of waveforms waveform rests on our never ability to lies.integrate It systemic is our and owncerebral inabilityhemodynamics Algorithm for Waveform Analysis across a wide spectrum of conditions. Intracranial vasculature poses unique challenges yet may toyield newunderstand knowledge if our technologies its language are able to quantify is blood the fl ow beyond pr oblem.”just velocities in —Merrillthe future. P. Spencer, MD For its normal function, brain needs continuous blood fl ow supply throughout the cardiac cycle, and its deliv- 2 “Doppler waveform never lies. It is our own inability Algorithm for Waveform Analysis ery is governed by cerebral autoregulation. The follow- to understand its language is the problem.” —Merrill P. Spencer, MD For its normal function, brain needs continuous blood fl ow supply throughout the cardiac cycle, and its deliv- ing specifi c steps help to structure the analysis of ery is governed by cerebral autoregulation.2 The follow- Introductioning specifi c steps help to structure the analysis of waveforms and provide data for interpretation: Introduction waveforms and provide data for interpretation:

The lowT-rheseist alnocew nat-urree osf ithset baraninc veas cnulaatutrue re Stepof 1.t Identifyhe b thera beginningin va sandcu thela endtu ofr ae single sets the stage for a wide variety of waveforms to ap- cardiac cycle (Figure 1, upper left insert). Step 1. Identify the beginning and the end of a single pear setsfrom interactions the stage between forsystemic a andwide local he- variety of waveforms to ap- modynamic factors. Although our ability to directly cardiac cycle (Figure 1, upper left insert). visuapearlize ves sfrels omis imp interactionsroving, the spectral a nbetweenalysis Step 2.systemic Determine the followingand local aspects: he- provides critical information to apply diagnostic criteria.m Thisod informationynam icis quantififact oedr sth.r uA vellotchityo u•g sharpnessh our of atheb systolicility fl owto acceleration; directly numbers, which in turn come from optimized spec- • the end-diastolic fl ow consistent with the expected Step 2. Determine the following aspects: tral wvaivsefuoramsl.i Szoeme vwaevsefsoremlss a rei sdi aginmostpic rbyo vinresistanceg, th ine thes parterialect systemral asuppliednaly bys theis sam- themselves as no single velocity number could de- pled vessel; scribpre theovidesm. Sometim ecriticals one has to i ninformationvoke or develop • waveformto apply shape transmission diagnostic from the prcri-oximal to a hemodynamic model to explain waveforms.1 The the distal part of the vessel; • sharpness of the systolic fl ow acceleration; aim oteria.f this rev iewThis is to de scinformationribe a range of wavefor miss •quantifi symmetry with ed the contra-lateralthru v ehomologouslocity seg- that could be found in the precerebral and intracra- ment; and • the end-diastolic fl ow consistent with the expected nial numbers,vessels under nor mwhichal and abn orinma l cturnirculato rycome • the frpreomsence ooptimizedf any cardiac, sy stspec-emic or focal conditions. circulatory condition that could explain the wave- resistance in the arterial system supplied by the sam- tral waveforms. Some wavefform.orms are diagnostic by pled vessel; themselves as no single velocityStep 3. Synthesize number the information could and de-explain the waveform appearance as attributable to: • waveform shape transmission from the proximal to Froms cther iComprbe ehensivethem Str.oke S Centerom eandt iNeurmeovasculars on e has to invoke or develop Ultrasound Laboratory, University of Alabama at Birmingham • technical error or an artifact; 1 the distal part of the vessel; Hospital,a Birmingham,hemodynamic Alabama. model to explain waveforms. The Address correspondence to: Andrei V. Alexandrov, MD, RVT, • systemic hemodynamic conditions; RWUH M226, 619 19th St. South, Birmingham, Alabama 35249-3280. • the presence of a focal lesion; and • symmetry with the contra-lateral homologous seg- E-mail:a avalexandrim [email protected] this review is to descr•i bincreeased, a r normal,ange or odecrf easedwa intracranialveform resists ance. that could be found in the precerebral and intracra- ment; and nial vessels under normal and abnormal circulatory • the presence of any cardiac, systemic or focal conditions. circulatory condition that could explain the waveform.

Step 3. Synthesize the information and explain the waveform appearance as attributable to: From the Comprehensive Stroke Center and Neurovascular Ultrasound Laboratory, University of Alabama at Birmingham • technical error or an artifact; Hospital, Birmingham, Alabama. Address correspondence to: Andrei V. Alexandrov, MD, RVT, • systemic hemodynamic conditions; RWUH M226, 619 19th St. South, Birmingham, Alabama 35249-3280. • the presence of a focal lesion; and E-mail: [email protected] • increased, normal, or decreased intracranial resistance. The Journal for Vascular Ultrasound 36(2):1–10, 2012 Extra- and Intracranial Waveform Analysis Algorithm, Descriptions, Classiﬁ cations, and Differential Diagnosis

Andrei V. Alexandrov, MD, RVT

ABSTRACT Waveform recognition is an essential step in the interpretation of vascular ultrasound studies. Waveforms can vary widely asThe the Journal forresult Vascular Ultrasound of 36(2):1–10,systemic 2012 and focal hemodynamic changes in the precerebralExtra- and and Intracranial particularly W intracranialaveform Analysis vessels. Waveform pattern recognition can aid or refi ne the applicationAlgorithm, of strict Descriptions, velocity criteria, Classifi andcations, this review will focus on typical normal and abnormal waveforms. andAlgorithms Differential for waveform Diagnosis analysis and interpretation of extra- and intracranial waveforms are discussedAndrei Valong. Alexandrov with, MD, RVT correlative imaging and clinical fi ndings. Wave- form descriptions and classifi cations based on prediction of vessel patency, systemic versus local ABSTRACT Waveform recognition is an essential step in the interpretation of vascular ultra- circulatory conditions,sound studies. and Waveforms resistance can vary widely to as thefl resultow of are systemic provided and focal hemodynamic for specifi c clinical conditions. Inter- changes in the precerebral and particularly intracranial vessels. Waveform pattern recognition can aid or refi ne the application of strict velocity criteria, and this review will focus on typical normal pretation of waveformsand abnormal waveforms.rests onAlgorithms our for waveformability analysis to and interpretationintegrate of extra- systemic and in- and cerebral hemodynamics tracranial waveforms are discussed along with correlative imaging and clinical fi ndings. Wave- across a wide spectrumform descriptions of andconditions. classifi cations based onIntracranial prediction of vessel patency vasculature, systemic versus local poses unique challenges yet may circulatory conditions, and resistance to fl ow are provided for specifi c clinical conditions. Inter- yield new knowledgepretation if of ourwaveforms technologies rests on our ability to integrate are ablesystemic andto cerebral quantify hemodynamics blood fl ow beyond just velocities in across a wide spectrum of conditions. Intracranial vasculature poses unique challenges yet may yield new knowledge if our technologies are able to quantify blood fl ow beyond just velocities in the future. the future.

“Doppler waveform never lies. It is our own inability Algorithm for Waveform Analysis to understand its language is the problem.” —Merrill P. Spencer, MD For its normal function, brain needs continuous blood fl ow supply throughout the cardiac cycle, and its deliv- ery is governed by cerebral autoregulation.2 The follow- “Doppler waveform never lies. It is our own inabilitying specifi c steps Algorithmhelp to structure th efor ana lyWsis oaveformf Analysis to understand its language Introductionis the problem.”waveforms and provide data for interpretation: The low-resistance nature of the brain vasculature Step 1. Identify the beginning and the end of a single sets the stage for a wide variety of waveforms to ap- cardiac cycle (Figure 1, upperFor left i insert).ts normal function, brain needs continuous blood —Merrill P. Spencer,pear MD from interactions between systemic and local hemodynamic factors. Although our ability to directly visualize vessels is improving, the spectral analysis Step 2. Determinefl the o followingw su aspects:pply throughout the cardiac cycle, and its deliv- provides critical information to apply diagnostic cri- 2 teria. This information is quantifi ed thru velocity • sharpness of the systolicery flis ow governed acceleration; by cerHeebralart Rhyth autorm and egulation.MCA Flow The follow- numbers, which in turn come from optimized spec- • the end-diastolic fl ow consistent with the expected tral waveforms. Some waveforms are diagnostic by resistance in the arterialing system specifi supplied byc thes sam-teps heDilpac rottioc stPSVru ct ure the analysis of themselves as no single velocity number could de- pled vessel; notch EDV scribe them. Sometimes one has to invoke or develop • waveform shapewaveforms transmission from the pr andoximal to pr ovide data for interpretation: Introductiona hemodynamic model to explain waveforms.1 The the distal part of the vessel; aim of this review is to describe a range of waveforms • symmetry with the contra-lateral homologous seg- that could be found in the precerebral and intracra- ment; and nial vessels under normal and abnormal circulatory • the presence of any cardiac, systemic or focal The low-resistance nature of the brain vasculatciruculatoryre condition that could explain the wave- Aortic R conditions. Valve closure form. Step 1. Identify the beginning andP T the end of a single Q sets the stage for a wide variety of waveforms to ap- S Step 3. Synthesizecar the informationdiac cycle and explain (Figur the e 1, upper left insert). pear from interactions between systemic and localwaveform he- appearance as attributable to: systole diastole From the Comprehensive Stroke Center and Neurovascular Ultrasound Laboratory, University of Alabama at Birmingham • technical error or an artifact; Alexandrov AV. Neurovascular Ultrasound Examination. Wiley: 2009 (in press). modynamic factors. AlHospital,tho Birmingham,ugh Alabama.our ability to directly Address correspondence to: Andrei V. Alexandrov, MD, RVT, • systemic hemodynamic conditions; RWUH M226, 619 19th St. South, Birmingham, Alabama 35249-3280. • the presence of a focalStep lesion; and 2. Determine the following aspects: visualize vessels is imE-mail:pro [email protected], the spectral anal• yincrseased,is normal, or decreased intracranial resistance. provides critical information to apply diagnostic cri-J Vasc Ultrasound 2012;36;103-112. teria. This information is quantifi ed thru velocity • sharpness of the systolic fl ow acceleration; numbers, which in turn come from optimized spec- • the end-diastolic fl ow consistent with the expected tral waveforms. Some waveforms are diagnostic by resistance in the arterial system supplied by the sam- themselves as no single velocity number could de- pled vessel; scribe them. Sometimes one has to invoke or develop • waveform shape transmission from the proximal to a hemodynamic model to explain waveforms.1 The the distal part of the vessel; aim of this review is to describe a range of waveforms • symmetry with the contra-lateral homologous seg- that could be found in the precerebral and intracra- ment; and nial vessels under normal and abnormal circulatory • the presence of any cardiac, systemic or focal conditions. circulatory condition that could explain the waveform.

ICA

Bulb ECA

CCA

Aorta

J Vasc Ultrasound 2012;36;103-112. J Vasc Ultrasound 2012;36;103-112. Waveform Recognition

An asymptomatic 32 year old man with arterial blood pressure 130/80.

Interpretation: this waveform shows a sharp systolic flow acceleration and stepwise deceleration with positive end-diastolic flow. The end- diastolic velocity, that is between 20% - 50% of peak systolic values, indicates a low resistance to arterial flow. Alexandrov AV. Cerebrovascular Ultrasound in Stroke Treatment and Prevention. Blackwell: 2004. Waveform Recognition

An asymptomatic 32 year old man with arterial blood pressure 130/80.

Interpretation: this recording shows a bi-directional signal with simultaneous sharp systolic up-strokes and similar stepwise deceleration in both flow directions. Both waveforms show low resistance flow patterns. Alexandrov AV. Cerebrovascular Ultrasound in Stroke Treatment and Prevention. Blackwell: 2004. Waveform Recognition A 65 year old man with a new onset aphasia and chronic hypertension.

Envelope is a line that follows the waveform The waveform above baseline has a rapid systolic up-stroke and a rounded peak systolic complex followed by a stepwise flow deceleration. The end-diastolic velocities below 30% of peak systolic values indicate relative increase in flow resistance. Weak signal below baseline is not optimized and measurements are erroneous. Alexandrov AV. Cerebrovascular Ultrasound in Stroke Treatment and Prevention. Blackwell: 2004. Waveform Recognition A 37 year old man with closed traumatic brain injury, ICP 52 mm Hg.

Above baseline: sharp systolic up-strokes are followed by sharp deceleration indicating an increased resistance to flow. Below baseline: a low resistance flow in a vein. A loud thump-like early systolic sound (circled) due to vessel wall motion causes envelope spikes (*) and measurement errors below baseline.

Alexandrov AV. Cerebrovascular Ultrasound in Stroke Treatment and Prevention. Blackwell: 2004. Waveform Recognition A 35 year old man with subarachnoid hemorrhage (Day 2), liver failure.

Both waveforms have sharp systolic up-strokes and an abrupt flow deceleration. These pulsatile waveforms with the end- diastolic velocities within 20% - 25% of peak systolic values indicate high resistance to arterial flow due to increased cardiac output and autoregulatory response.

Alexandrov AV. Cerebrovascular Ultrasound in Stroke Treatment and Prevention. Blackwell: 2004. Waveform Recognition A 42 year old woman with closed traumatic brain injury. *

The waveforms above baseline (sharp systolic up-strokes, stepwise deceleration, low resistance) have variable velocities with regular heart rate. Velocity fluctuations can spontaneously occur every 4 cardiac cycles due to breathing. A cycle with the highest velocities (*) can be used for manual calculations.

Alexandrov AV. Cerebrovascular Ultrasound in Stroke Treatment and Prevention. Blackwell: 2004. Waveform Recognition A 54 year old man with an acute small cortical stroke and LVH.

Both waveforms have sharp up-strokes, late arrival of maximum systolic velocites, and stepwise flow deceleration. The end- diastolic velocities fall below 30% of peak systoli due to irregular heart rate: this also affects estimation of flow resistance using single cardiac cycle or only 2-5 cycles-averaged values. Alexandrov AV. Cerebrovascular Ultrasound in Stroke Treatment and Prevention. Blackwell: 2004. Waveform Recognition A 60 year old woman with recent TIA and atrial fibrillation. *

Waveforms towards the probe have irregular arrival of cardiac cycles with sharp up-strokes and variable velocities. As a practical rule, a cycle with the highest velocities (*) can be used for manual calculations. However, estimation of flow resistance and representative mean velocity is difficult since the pulse rate and cardiac output are affected. Alexandrov AV. Cerebrovascular Ultrasound in Stroke Treatment and Prevention. Blackwell: 2004. Signal Optimization

Alexandrov AV. Cerebrovascular Ultrasound in Stroke Treatment and Prevention. Blackwell: 2004. AB

E F

J Vasc Ultrasound 2012;36;103-112. Waveform Recognition

A 67 year old man with resolving MCA stroke and carotid occlusion.

The waveform above baseline shows a delayed systolic flow acceleration, flattened systolic complex, and slow diastolic deceleration. End-diastolic velocities above 50% of peak systoli indicate very low flow resistance. In cerebrovascular studies, this waveform is called a “blunted” flow signal. Alexandrov AV. Cerebrovascular Ultrasound in Stroke Treatment and Prevention. Blackwell: 2004. Waveform Recognition

A 73 year old male with MCA stroke and carotid occlusion.

The waveform above baseline has an upward systolic up-stroke. This waveform has to be compared to a non-affected vessel in order to decide if only a slight delay in systolic acceleration is present. Regardless, this is NOT a blunted signal since a clear systolic complex is visualized. Alexandrov AV. Cerebrovascular Ultrasound in Stroke Treatment and Prevention. Blackwell: 2004. Waveform Recognition

A 62 year old woman with an M1-MCA occlusion.

This is a minimal bi-directional signal with no end-diastolic flow. This waveform can be representative of a residual flow signal around MCA clot if collaborated by additional findings indicating occlusion at this location. Bruits and vessel intercept at nearly 90 degree angle should be considered as possible explanation. Alexandrov AV. Cerebrovascular Ultrasound in Stroke Treatment and Prevention. Blackwell: 2004. Waveform Recognition

A 41 year old woman with TBI and clinical progression to brain death.

Reverberating or oscillating flow signals. These waveforms represent an extremely high resistance to flow: (above baseline) sharp spikes with short flow reversal during closure of the aortic valve and no end diastolic flow; (below) reversed flow during entire diastoli.

Alexandrov AV. Cerebrovascular Ultrasound in Stroke Treatment and Prevention. Blackwell: 2004. Brain

Arm

J Vasc Ultrasound 2012;36;103-112. Thrombolysis in Brain Ischemia (TIBI) waveforms Normal Stenotic Dampened Blunted Minimal Absent V IV III II I 0

3 2 0-1 Thrombolysis in Myocardial Infarction (TIMI) flow grades J Vasc Ultrasound 2012;36;103-112. 2012 EXTRA- AND INTRACRANIAL WAVEFORM ANALYSIS 9

Table 1 A Summary of Waveform Descriptions and Classiﬁ cations

Condition Waveform Description or Classifi cation Comments Carotid artery stenosis Waveforms helpful to suspect or confi rm the In addition, waveform analysis helps signifi cance of the stenosis: spectral to identify cardiac infl uences on the narrowing, tardus parvus, and suppressed fl ow profi le and velocity fi ndings (high resistance) waveform in the (e.g., atrial fi brillation, valvular pre-stenotic segment. insuffi ciency or stenosis). Subclavian, innominate, Steal waveforms (stages fr om latent to present Diagnostic criteria for the steal and vertebral artery at rest) are well described and established. 21 phenomenon are supported by waveform disease recognition and documentation. Carotid artery occlusion Drum-like waveforms; early systolic and The presence of these waveforms may end-systolic vessel wall movements increase the confi dence in diagnosing complete ICA occlusion. Traumatic brain injury Hemodynamic phases after injury Although the utility of ultrasound for this (hypoperfusion, hyperemia and vasospasm) condition still needs to be established, were defi ned.22 Waveforms specifi c underrecognized vasospasm and for each phase as well as for assessment of diastolic fl ow versus ICP the mass effect need to be further and CPP are promising venues for determined. further development of criteria for diagnostic waveform patterns. Subarachnoid hemorrhage Waveforms consistent with vasospasm, Low- and high resistance waveform (SAH) hyperemia and increased ICP. patterns should be used to help r efi ne velocity and ratio changes in patients with SAH. Cerebral circulatory arrest Waveforms across the spectrum of ICP changes TCD is a confi rmatory test to determine or are well defi ned (incomplete or complete rule out cerebral circulatory arrest based cerebral circulatory arrest with oscillating or on assessment of the diastolic component reverberating fl ow patterns)18,19 and detection of specifi c waveforms indicating cessation of meaningful antegrade fl ow. Extra- and intracranial Waveform patterns of fl ow diversion, Waveform changes particularly with lesions arterial occlusions and compensatory vasodilation and fl ow not directly accessible with ultrasound high-grade stenoses collateralization are defi ned as well as other are very helpful (e.g., distal ICA stenosis, waveforms pointing to either pr oximal ICA dissection, siphon lesions, M2 MCA (blunted/ delayed fl ow acceleration) lesions, etc). or distal (increased resistance in a branch vessel) lesions 4 Acute ischemic stroke Thrombolysis in Brain Ischemia (TIBI) Waveform analysis helps to quickly classifi cation of the residual fl ow using determine the presence and persistence 6 types of waveforms (absent, minimal, of an arterial occlusion and completeness blunted, dampened, stenotic, and normal) 20 of focal recanalization/ tissue reperfusion at bedside.

J Vascvelocity Ultrasound is the key component 2012;36;103 that allows prediction-112. of patient when the distal M1 MCA embolus started to local recanalization as well as tissue reperfusion from recanalize during treatment with intravenous tissue TIBI waveforms. Persistent occlusion and no-refl ow plasminogen activator and moved to the proximal phenomena are associated with absent or poor end- M2 MCA. As a result, a dampened TIBI III waveform diastolic fl ow such as seen with TIBI 0–1 and 2–3 appeared as compared with the non-affected side waveforms. Normal or elevated EDV with TIBI 4–5 (data not shown). However, it still has the presence of waveforms are predictive of complete recanalization turbulence and now shows a high-intensity transient with or without residual stenosis, and TIMI 3 tissue embolic signal. These were indirect signs that resid- reperfusion. ual embolic material could still be present in the M1 Not all residual fl ow waveforms are easy to clas- MCA or an additional and more proximal embolus sify at first glance. Figure 12 depicts the presence of could exist. turbulence with an embolus that causes a high-grade stenosis in the distal M1 segment of the MCA. This is Conclusions the lesion on the other side of the Spencer’s curve as evident from a disturbed waveform, lower than ex- A summary of waveform descriptions and classifi ca- pected elevation of the systolic velocity and low end- tions (if applicable) is provided in Table 1. Waveforms diastolic fl ow. This is a TIBI grade IV stenotic signal. provide insights into real time hemodynamics of nor- The waveform below was obtained in the same mal and diseased vessels. Interpretation of waveforms