History of Doppler Ultrasound

The origins of Doppler ultrasound can be traced back to the description of the Doppler effect by the Austrian physicist Christian Doppler. First presented to the Royal Bohemian Society of Sciences in May 1842, his work was titled “On the colored light of the double stars and certain other stars of the heavens.” This presentation described the Doppler effect as it subsequently became known, which explains the shift in the frequency of a wave when either the transmitter of the wave or the receiver of the wave is moving with respect to the wave-propagating medium. This effect applies to sound, and sound emanating from or reflected by an object moving toward an observer is characterized by a higher frequency in proportion to the speed of the moving object. Conversely, if sound emanates from an object moving away from an observer, the sound has a lower frequency in proportion to the speed of the moving object. According to this principle, ultrasound can be used to measure the velocity of flowing blood. Ultrasound is generated by a small crystal, emitted from a probe, and then reflected off the moving blood cells, and the reflected signal is also received by the probe. The shift in frequency between the emitted and the reflected ultrasound varies proportionally with the velocity of the flowing blood. Blood flowing toward the probe reflects the ultrasound at a higher frequency and produces a positive Doppler shift, whereas ultrasound reflected from blood flowing away from the probe is reduced in frequency or exhibits a negative Doppler shift. The signals are processed, and a spectral analyzer displays multiple spectral dots for the many reflections that are recorded.

The use of Doppler ultrasound to measure blood flow was initially reported by Satomora in 1959.¹ Satomora and Kaneko were originally interested in measuring cerebral blood flow; however, they concluded that the skull was an insurmountable barrier to the passage of ultrasound and focused on the extracranial carotid arteries in their initial investigations. After refinements in the equipment and advances in the technology, Doppler ultrasound was introduced as a clinical tool to examine blood flow velocity in the extracranial and peripheral arteries. Refinements in signal processing and the introduction of B-mode imaging, as well as duplex imaging (simultaneous display of B-mode tissue imaging and vascular flow velocity measurements), followed by the development of color flow ultrasound, have continued to improve its diagnostic capabilities for extracranial and intracranial vascular pathology. It has recently been demonstrated that sonothrombolysis (the use of ultrasound to help dissolve blood clots) can be used as a treatment of stroke and potential treatment of intracerebral hemorrhage. These observations have extended the applications of Doppler ultrasound from diagnostics to therapeutics in the brain.

B-Mode Ultrasonography

The basic principle underlying real-time brightness-mode (B-mode) imaging is the variable acoustic impedance that different body tissues naturally possess. When the technique was first developed to use ultrasound to image tissue, the intensity and time of flight of the reflected ultrasound signal (echo) were represented as a unidimensional tracing, with the amplitude of the wave representing intensity and the distance along the x-axis representing the reflector depth or distance from the transducer (known as amplitude modulation mode, or A-mode). The standard display now allows visualization of tissue interfaces when sound waves are spatially represented by picture elements (pixels) on a video display, with the brightness of each pixel correlating with the amplitude or intensity of the echo signal. Scanner screens use 256 shades of gray to represent the tissues’ different ability to absorb or reflect sound, which is displayed as a picture.

When the piezoelectric crystal in the probe emits a pulse, this focused beam travels along a single plane. Any sound reflected back along this plane is received by the probe, recorded, and presented on the display screen. By insonating adjacent segments of tissue with multiple crystals located within a sonographic head or with a rotating head, a two-dimensional representation of the object being insonated can be shown on the display screen. This B-mode display has become a standard mode of displaying the anatomy and differential densities of the insonated tissue. The rapid processing and display of these data allow “real-time” imaging in ultrasonography such that pulsations of vessels and cerebrospinal fluid spaces can be identified.

Duplex Ultrasonography

The combination of improved beam focusing and pulsed Doppler blood flow velocity measurements improved the ability to localize the source of the reflected signal within tissue. Doppler spectral analysis of this signal, through the use of fast Fourier transformations, allowed the complex reflected signal of each single-frequency component to be displayed visually. This resultant “power spectrum” for the Doppler signal could be used to distinguish the various flow characteristics that exist in normal and diseased vessels.

These vascular diagnostic modalities were later integrated with B-mode imaging to produce duplex ultrasonography.¹⁰ Modern duplex scanning uses one transducer (5 to 7 MHz) to produce simultaneous B-mode images and pulsed Doppler waveform analysis. The screen of the display module provides real-time anatomic B-mode data and a graphic representation of the Fourier-transformed pulsed spectral analysis. An additional advantage of this diagnostic tool is that it has the ability to correct for variable angles of insonation, which if not kept between 55 and 65 degrees, could significantly alter the recorded frequencies (Fig. 347-1).

FIGURE 347-1 Effect of the insonation angle between the ultrasound beam and the artery on velocity values. The percentage of the actual velocity measured changes as a function of the cosine of the insonation angle (phi) according to the following formula: FV = v/cos(phi), where the true flow velocity (FV) equals the measured velocity (v) divided by the cosign (cos) of the angle of insonation (phi). The use of duplex imaging can determine this angle and correct for it, thereby increasing the accuracy of the velocity measurements.

One of the major limitations of duplex scanning is that just a small region of the artery can be studied at any one time. This technology is also only a two-dimensional representation of a three-dimensional dynamic process. Color flow imaging, which combines real-time, B-mode, gray-scale imaging with color encoding of multigated Doppler flow information, begins to address these issues by sampling the mean Doppler frequency shift at various depths over the entire scan area. The color assigned to the frequency data depends on the magnitude and direction of flow, with hue (red versus blue) depicting the direction of flow, the hue’s saturation denoting the magnitude of the frequency shift, and the brightness (luminance) demonstrating the variance in mean flow (i.e., the turbulence), which is superimposed on a gray-scale B-mode image depicting the surrounding anatomy. Color Doppler allows more rapid identification of the component vessels than duplex alone does, and it also provides rapid identification of laminar flow patterns for placement of a single pulsed Doppler gate for acquisition of spectral waveform data. Today, the combined use of B-mode two-dimensional imaging and the real-time, color-enhanced spectral analysis of pulsed wave Doppler have enabled experienced operators to recognize carotid stenosis with a sensitivity approaching 100% (Fig. 347-2).

FIGURE 347-2 Top, Color flow duplex scan of a normal right internal carotid artery (RICA) bifurcation showing the normal anatomy in gray scale, as well as the red color indicating blood flow at a peak systolic blood flow velocity of 59 cm/sec, indicative of no significant stenosis. Bottom, Color flow duplex scan of a significant (70% to 79%) stenosis of the RICA. The gray-scale picture indicates excess tissue in the intravascular space at the carotid bifurcation with a peak systolic blood flow velocity of 334 cm/sec and significant spectral broadening (increased gray color within the spectrum).

Transcranial Doppler

Aaslid and colleagues first reported the ability to record blood flow velocity in the intracranial arteries with Doppler ultrasound in 1982 and introduced TCD ultrasonography.¹¹ TCD ultrasonography used an optimized 2-MHz frequency with a pulsed Doppler range-gated design. The lower 2-MHz frequency allowed penetration through the cranium in the thin portions of the bone. The introduction of TCD ultrasound permitted examination of the intracranial vasculature, which has improved the diagnosis of intracranial disease. The technique of TCD ultrasound was initially used in the Department of Neurosurgery in Bern, Switzerland, for the diagnosis of vasospasm after subarachnoid hemorrhage.¹² Many subsequent uses have been described for this technology and are reviewed in this chapter.

A complete TCD examination normally includes examination through three transcranial windows: the transtemporal, transorbital, and transoccipital windows.¹³ Through these three windows, most of the basal intracranial arteries can be examined (Fig. 347-3). The transtemporal window is used to examine the middle cerebral artery, (MCA), anterior cerebral artery, intracranial internal carotid artery (ICA), and proximal posterior cerebral artery. The transorbital window is normally used to examine the ophthalmic artery and the intracavernous and supraclinoid ICA. The transoccipital window is used for examination of the posterior circulation, specifically the two vertebral arteries and the basilar artery. The origin of the posterior inferior cerebellar arteries can be examined in many patients.

FIGURE 347-3 The three windows for transcranial Doppler examination of the basal cerebral vessels. On the lateral portion of the skull, the transtemporal region (A and C) is used to examine the middle cerebral artery, proximal anterior cerebral artery, distal internal carotid artery, and proximal posterior cerebral artery. The spectral waveform of the middle cerebral artery (B) has a characteristic low-resistance profile. Anteriorly, the transorbital window is used to examine the carotid siphon, as well as the ophthalmic artery (C). The third window is the transforaminal window, which can be used to examine the posterior circulation vessels. The two vertebral arteries and the basilar artery are examined easily by this method. The posterior inferior cerebellar arteries can often be examined.

The original technique for examination was a handheld method in which manual manipulation of the probe with recordings of sample volumes at preselected depths was used to examine specific sites in the intracranial vasculature. Normally, as the distal intracranial arteries become vertically oriented in the sylvian and intrahemispheric fissures, they are considered beyond the scope of the normal TCD examination.¹⁴ The distal intracranial arteries can occasionally be examined in unusual circumstances, however, such as after craniotomy, when bone is removed and examination windows are created.

The Doppler signal reflected from intracranial arteries is used to measure ultrasound frequency shifts, which are converted into blood flow velocity in centimeters per second. To determine the actual blood flow in milliliters per minute, the total average velocity, vessel diameter, and angle of insonation need to be known precisely,¹⁵ and these parameters are not usually measured during routine TCD examinations. The angle of insonation (angle between the ultrasound beam and the vessel being recorded from) is important to consider when measuring TCD velocity. The true velocity and observed velocity are equal when the angle of insonation is zero, as often happens when examining the MCA trunk. As the angle of insonation increases, there is a reduction in the observed frequency as a function of the cosine of the angle (see Fig. 347-1). For example, if the angle between the ultrasound beam and the flow reflector is 30 degrees, 97% of the true velocity can be observed by the recording device. As the angle of insonation increases, the proportion of the true velocity observed decreases. If the angle of insonation is 60 degrees, 50% of the true velocity is observed by the recording equipment. TCD examination techniques incorporate this strategy to reduce the insonation angle to as low as possible when recording from different intracranial arteries. The introduction of color flow ultrasound adds the ability to determine the angle of insonation to some extent during the examination. Figure 347-3 illustrates a typical velocity recording from the MCA.

Since the introduction of TCD, the technology has undergone significant improvements as a result of technical advances in equipment and signal processing that allowed an increased capability of detecting abnormalities in intracranial vascular structure and function. TCD allows three-dimensional vessel maps to be constructed, and the recording site can be documented easily in most cases. Advantages of this technology include an easier learning curve for beginning examiners and positive identification of abnormal vascular anatomy, which can be due to anatomic variations, as well as pathologic changes from occlusion or displacement.

Advances in computerized recording devices and the miniaturization of computerized digital processing have allowed easier recording of Doppler signals for analyses. This development has permitted continuous recording of spectral outline and full spectral signals, which enables more detailed analysis of physiologic events.¹⁶ Digital recording with the capability of displaying continuous data over time has allowed calculation of changes in velocity secondary to alterations in blood flow evoked by physiologic stimuli, as well as analysis of cerebrovascular control mechanisms, such as CO₂ autoregulation and changes in flow caused by cortical activation. Two-channel TCD ultrasound was introduced to resolve physiologic questions regarding control of the cerebral circulation and to document simultaneous changes in venous and arterial flow.¹⁵ Subsequently, multichannel TCD has been used to register signals simultaneously from both cerebral hemispheres recorded from both MCAs.¹⁷ Multichannel TCD has allowed detection of emboli by recording from multiple sample volumes simultaneously.¹⁸ Reflected power recordings have enabled calculation of relative flow volume, which can be used for scientific purposes to calculate changes in cerebral blood flow.¹⁵ Power Doppler can be performed transcranially to provide further detail in imaging intracranial vascular structures, such as aneurysms and arteriovenous malformations. Miniaturization has allowed TCD units to become more portable, and in the future, battery-operated devices may serve as a neurovascular stethoscope to permit rapid determination of vessel patency and flow characteristics.

Intracranial Stenosis

TCD ultrasonography has extended the ability of Doppler ultrasound to diagnose intracranial arterial stenosis reliably, which may be caused by a variety of pathologic conditions other than vasospasm. A common cause of intracranial stenosis in patients with symptoms related to cerebrovascular disease is atherosclerosis involving the basal intracranial vessels. Other conditions can also cause intracranial arterial stenosis, including moyamoya disease, intracranial arterial dissection, vasculitis, preeclampsia, sickle cell disease, and arterial-to-arterial emboli undergoing various stages of recanalization.

Atherosclerotic disease of the intracranial vessels has received much less attention than similar lesions of the extracranial cerebral vessels.²⁴ Although intracranial arterial stenosis secondary to atherosclerosis is less common than extracranial stenosis in Western populations, intracranial atherosclerotic lesions can be a significant cause of stroke and transient ischemic attacks. Bogousslavsky and coworkers analyzed the natural history of patients with MCA stenosis or occlusion entered into the Extracranial-Intracranial Bypass Study Group.²⁵ During a follow-up period of 42 months, 11.7% of the patients per year experienced recurrent cerebrovascular events (transient ischemic attack or stroke).

Several early studies reviewed the accuracy of TCD for the diagnosis of intracranial stenosis secondary to atherosclerosis.^26–28 Spencer and Whisler compared carotid siphon stenosis assessed by angiography and TCD.²⁷ In a group of 33 carotid siphons visualized angiographically, 11 showed stenosis ranging from 30% to 75%. Comparison of TCD with angiography revealed a sensitivity of 73% and specificity of 95%. Ley-Pozo and Ringelstein compared intra-arterial digital subtraction angiography with TCD in detecting occlusive disease of the carotid siphon and MCA.²⁸ Sixteen of 17 cases of carotid siphon stenosis were identified correctly with TCD. Rorick and colleagues compared TCD with angiography for the diagnosis of intracranial stenosis.²⁹ These investigators found that the presence of coexisting extracranial stenosis can be a confounding variable affecting the accuracy of TCD in detecting intracranial lesions. It has been useful to examine the intracranial arteries with TCD in patients with cerebrovascular symptoms who are not found to have significant pathology in the extracranial arteries.

Intracranial Hemodynamics

Evaluation of intracranial hemodynamics with TCD has allowed assessment of vascular control of the cerebral circulation, as well as identification of patients with cerebrovascular occlusive disease who are at higher risk for stroke. In patients with occlusive disease, the two most useful capabilities of TCD with regard to hemodynamic parameters are evaluation of collateral patterns and cross-flow and direct assessment of hemodynamic changes in the distal vascular territories caused by proximal occlusive lesions.

Maintenance of adequate cerebral blood flow depends on sufficient cerebral perfusion pressure through the inflow vessels along with adequate blood pressure, intact anatomy of the extracranial and intracranial vasculature, and compensatory mechanisms when the vasculature or blood pressure becomes compromised. The normally configured circle of Willis functions as a manifold to normalize perfusion pressure to the distal cerebral vessels if one or more extracranial vessels becomes occluded or hemodynamically compromised secondary to stenosis. The circle of Willis normally allows nearly complete compensation in the event of carotid occlusion.³⁰

Compensation for carotid occlusion usually occurs via (1) crossover through the anterior communicating artery and reversed flow in the proximal anterior cerebral artery (A₁) ipsilateral to the occlusion, (2) forward flow in the posterior communicating artery ipsilateral to the occlusion, or (3) reversed flow in the ipsilateral ophthalmic artery.³¹ Major differences in the functional capacity of the circle of Willis are found in the general population, however, and some patients cannot recruit sufficient flow to maintain adequate cerebral perfusion pressure.

The use of TCD to evaluate vasomotor reserve (VMR) is based on the principle that changes in velocity are proportional to changes in flow through a vessel if the vessel diameter is constant. Changes in blood flow velocity through the MCA can be used to reflect relative changes in blood flow in that artery as a result of alterations in CO₂ concentration or acetazolamide or moderate changes in arterial blood pressure (autoregulation). By using these principles, the functional capacity of the distal regulating vessels in the cerebral circulation can be assessed with TCD.

When cerebral perfusion pressure decreases as a result of proximal arterial stenosis or occlusion, the distal cerebral regulatory vessels become dilated maximally and exceed their functional capacity to autoregulate and respond to CO₂. Acetazolamide has been used to induce distal cerebral vasodilation and probably works by a similar pH-dependent mechanism as inhalation of CO₂. Lack of responsiveness to CO₂ or to acetazolamide can be used to indirectly detect reduced cerebral perfusion pressure in the MCAs as a result of the effects of proximal occlusive disease.

Testing of VMR with TCD can provide valuable physiologic information regarding prognosis in patients with certain vascular lesions. Kleiser and Widder reported the results of a natural history study of patients with unilateral carotid occlusion after VMR testing.³⁰ In a group of 86 patients with unilateral carotid occlusion, 11 had exhausted their VMR and had an increased ipsilateral stroke rate (17% per year for 3 years as compared with an ipsilateral stroke rate of 3% per year for the entire group, which is comparable to previously published series). Similar findings of increased stroke risk in patients with carotid occlusion and impaired vascular reserve, measured with other methods, including xenon-enhanced computed tomography (CT) with acetazolamide challenge and positron emission tomography (PET), have been reported.^32,33 Patients with impaired vascular reserve and recent symptoms distal to ICA occlusion are the subject of an ongoing randomized clinical trial comparing extracranial-to-intracranial bypass surgery with medical therapy for prevention of stroke (Carotid Occlusion Surgery Study [COSS]).

The prognosis of patients with carotid stenosis may be influenced by VMR. Gur and associates reported that patients with asymptomatic carotid stenosis and impaired VMR had a worse prognosis than did those with intact VMR.³⁴ With the use of VMR testing it appears possible to identify patients with poor hemodynamic reserve and carotid stenosis or occlusion who have a high risk for stroke and may benefit from medical or surgical therapy to improve cerebral perfusion.

Cerebral Autoregulation

Cerebral autoregulation is the ability of the brain to maintain constant cerebral blood flow despite changes in cerebral perfusion pressure. The methodology for determining cerebral autoregulation in the past was cumbersome and invasive and required radioisotopes for cerebral blood flow measurement and vasoactive medication to change the blood pressure. TCD can be used to determine autoregulation noninvasively in the MCA perfusion territories.^35,36 Preliminary clinical testing in patients with cerebrovascular occlusive disease indicates that cerebral autoregulation is absent in patients with severely impaired CO₂ reactivity.^36,37 Noninvasive testing of cerebral autoregulation with TCD may prove useful in the complete hemodynamic evaluation of patients with cerebrovascular occlusive disease and may also be helpful in managing patients after head injury.

Positional Vertebral Artery Obstruction

Cerebrovascular insufficiency can sometimes occur in the posterior circulation secondary to positional obstruction of one or both vertebral arteries in the setting of impaired collateral pathways from the anterior circulation. Several reports have found TCD monitoring of the posterior cerebral arteries bilaterally during various head positions to be useful in detecting transient hemodynamic insufficiency in this condition.^38,39 The most common cause of positional vertebral artery obstruction is cervical spondylosis, which can be treated surgically by anterolateral removal.³⁹ The essential diagnostic findings on TCD are a transient drop in posterior cerebral artery velocity signals with head turning and rebound hyperemia on return to a neutral position. Examination with TCD is helpful in differentiating true positional ischemia from positional vertigo and can identify patients who require angiography to define the location and nature of the obstruction.

Intracranial Emboli

Interest has developed in the ability of TCD to directly detect intracranial microemboli in the basal cerebral vessels and their relationship to stroke and ischemic symptoms in patients with occlusive vascular disease. Doppler ultrasound was previously used clinically to detect intravascular air microemboli audibly during cardiac surgery⁴⁰ and during neurosurgical procedures in the sitting position.⁴¹ The ability of TCD to detect intracranial microemboli was first recognized by monitoring MCA velocity during carotid endarterectomy (CEA) and cardiac surgery.^42,43

Clinical correlation of detection of intracranial and extracranial microemboli has been described.⁴⁴ Intracranial microemboli have been detected in patients with atrial fibrillation, prosthetic heart valves,⁴⁵ carotid stenosis,^46,47 fibromuscular dysplasia, arterial dissection,^48,49 and intracranial stenosis, as well as during invasive procedures such as angiography,⁵⁰ angioplasty,⁵¹ and vascular and heart surgery⁵² and after aneurysm treatment.⁵³ Monitoring for intracranial air microemboli after venous injection has been helpful in identifying patients with cardiac and pulmonary defects causing right-to-left circulation shunts.⁵⁴ Most microemboli do not cause overt symptoms; however, multiple microemboli have been associated with impaired neuropsychological function after cardiac surgery.⁵⁵ The clinical utility of intracranial monitoring of emboli is still being established, but much has been learned with TCD about microembolism to the brain. Monitoring of emboli with TCD has played a useful role in identifying the site of active embolization in the arterial system in patients with transient ischemic attack or recent stroke, distinguishing embolic versus hemodynamic causes of stroke and transient ischemic attack, identifying high-risk stages of neurovascular and surgical procedures, and detecting patients with vascular lesions at higher risk for ischemic events.

Effect of Vessel Narrowing from Subarachnoid Hemorrhage on Blood Flow Velocity

Normal values for cerebral blood flow velocity have been published by multiple investigators. Under normal conditions, the average flow velocity in the MCA is 62 cm/sec. Velocities greater than 120 cm/sec correlate with vasospasm, as evidenced by vessel narrowing on angiography. Flow velocities greater than 200 cm/sec often correlate with severe vasospasm seen angiographically and are associated with clinical episodes of ischemia and infarction.^12,58,65,66

Early studies performed by Aaslid and Seiler and their colleagues in Bern, Switzerland, established the correlation between elevated velocities and cerebral infarction from vasospasm.^58,67,68 Seiler and colleagues observed a group of 39 patients with subarachnoid hemorrhage monitored by TCD ultrasonography. In no patient did a cerebral infarct develop if blood flow velocities had not exceeded 140 cm/sec. Blood flow velocities greater than 200 cm/sec were associated with ischemic episodes and infarctions, but some patients remained asymptomatic.^58,67,68 These observations are in keeping with previous angiographic studies of vasospasm indicating that there is not an absolute concordance between the degree of vessel narrowing and the development of delayed ischemia. Most patients in whom major episodes of delayed ischemia develop have significant vessel narrowing; however, other patients progress to severe vessel narrowing but preserve cerebral blood flow and DINDs do not develop. Occasionally, diffuse distal spasm unaccompanied by proximal vessel spasm may develop in patients with subarachnoid clots located in the distal cerebral fissures.¹⁴ Frequently, careful examination of the CT scan for the presence of distally located blood clots can help predict patients at risk for distal vessel vasospasm. Cerebral blood flow measurements and angiography are required to confirm the diagnosis.

Subsequent studies have correlated the degree of angiographic vasospasm with TCD velocities and found the best correlations in the MCA and distal ICA.^69,70 Vasospasm in the distal anterior cerebral artery is not normally accessible with TCD, and the proximal anterior cerebral arteries can sometimes be problematic because of congenital atresia and other factors that lead to poor correlation between vessel diameter and increased velocity. Multiple physiologic factors have been shown to affect blood flow velocity in patients with vasospasm, including ICP, hematocrit, increased blood pressure from induced hypertension, hyperdynamic therapy, and hyperemia, which can result from aggressive hemodynamic support.^71,72 Aaslid and colleagues initially described a mathematical ratio of flow velocities,⁶⁷ which was analyzed further by Lindegaard and associates,⁷⁰ between the MCA and the extracranial ICA (V_MCA/V_ICA) to correct for flow changes induced by vasospasm. Because velocity can change from alterations in either flow or vessel diameter, the extracranial-to-intracranial ratio can be helpful in distinguishing hyperemia from elevated velocities caused by spasm. If vasospasm progresses to such a severe degree that cerebral blood flow is significantly reduced, intracranial velocities may begin to decrease as the vessels become severely attenuated and thus lead to a decrease in blood flow velocity in the extracranial carotid artery as well, which decreases intracranial velocity but results in a persistently elevated V_MCA/V_ICA ratio.

Predictive Value of Transcranial Doppler Ultrasound in Vasospasm

The time course of vasospasm can vary according to the severity of the subarachnoid hemorrhage and the extent of the spasm. The peak incidence of vasospasm occurs in a delayed fashion after subarachnoid hemorrhage. Weir and coworkers carefully analyzed angiograms from 293 patients with aneurysms.⁷³ These investigators found that angiographically detected vasospasm initially occurred 3 days after subarachnoid hemorrhage and appeared to be maximal between days 6 and 8 and was much reduced by day 12. Subsequent studies using TCD confirmed the delayed onset and time course of vasospasm.^67,74,75 Multiple investigators have observed that patients who have early and more severe increases in flow velocity are at higher risk for the development of DINDs. The use of TCD has addressed several controversies in the past regarding the timing of vasospasm. Romner and associates were unable to find significant vasospasm within 12 hours after subarachnoid hemorrhage in a group of patients harboring aneurysms.⁷⁶ It was previously thought that an early and late phase of vasospasm existed, thus leading to management decisions based on these preconceptions. These authors also showed that progression of vasospasm was more likely to occur in patients operated on 49 to 96 hours after aneurysmal rupture than in patients operated on within 48 hours of rupture, thereby supporting the notion that early aneurysm surgery may be less likely to aggravate vasospasm than surgery during the time when vasospasm begins to occur to a significant extent.⁷⁷

Grosset and colleagues showed that rapid increases in velocity are predictive of patients in whom DINDs will develop.⁷⁸ These investigators repeatedly examined 121 consecutive patients after acute subarachnoid hemorrhage frequently for 14 days. The average highest velocity attained was greater in the 47 patients in whom DINDs developed (186 cm/sec) than in the 74 patients in whom DINDs did not develop (149 cm/sec). There was no significant difference, however, between symptomatic and asymptomatic patients in the velocities recorded before onset of the neurological deficits, but the rate of increase in TCD velocities was greater in patients in whom DINDs subsequently developed. In these patients, there was a maximum increase in velocity of 65 ± 5 cm/sec per 24-hour period versus increases of 47 ± 3 cm/sec per 24-hour period in patients in whom DINDs did not develop. An increase greater than 50 cm/sec per 24-hour period may help predict patients in whom DINDs will eventually develop; information gathered from sequential recording may be useful in managing these patients.

Several treatments are available for cerebral vasospasm after subarachnoid hemorrhage, and it is useful to consider the effects of these treatments on vessel diameter and TCD values. Triple-H therapy (hypertension, hypervolemia, and hemodilution) may have unpredictable effects on basal vessel velocity recordings with TCD. Reports have indicated that in some cases increases in velocity after this treatment can be due to increases in blood flow and not further reduction in basal vessel diameter.^79,80 The state of cerebral autoregulation may play a role in determining the changes in velocity after induction of triple-H therapy.⁸⁰ These reports underscore the need for combined assessment with cerebral blood flow measurements and the use of intracranial-to-extracranial ratios to correct for blood flow changes when using TCD to monitor patients undergoing triple-H therapy for vasospasm.

Calcium channel blockers have been shown by several prospective randomized trials to be useful for preventing vasospasm and reducing morbidity and mortality after subarachnoid hemorrhage.⁸¹ It has been shown that cerebroselective calcium channel blockers cause reductions in subarachnoid hemorrhage–induced elevated intracranial blood flow velocities measured with TCD.⁸²

Endovascular therapy has been useful in reversing vessel narrowing from vasospasm and improving cerebral blood flow. TCD ultrasonography and cerebral blood flow studies have shown improvement in blood flow and a reduction in elevated blood flow velocities secondary to spasm after treatment.^82,83 Interventional treatment with papaverine has been found to decrease cerebral blood flow velocities and improve cerebral blood flow, although the effect appears to be more transient.

Arteriovenous Malformations

Hemodynamic changes in arteriovenous malformations have been well characterized by blood flow studies, direct pressure measurements, and TCD ultrasonography.⁸⁴ Classic arteriovenous malformations show high velocity in feeding arteries corresponding to elevated flow volume and have a low pulsatility index on TCD, which indicates decreased vascular resistance. It has been shown by color flow Doppler that arteriovenous malformations can be imaged transcranially.⁸⁵ Characteristic features on color flow imaging and the hemodynamic changes on TCD can lead to the diagnosis of arteriovenous malformation in patients examined for neurological symptoms such as headaches, seizures, or other focal findings that may lead to routine examination. For decision making regarding treatment planning, other imaging modalities such as CT, magnetic resonance imaging (MRI), and angiography are essential.

Detection of Intracranial Aneurysms

Standard TCD has not generally been useful as a screening test to detect cerebral aneurysms; however, with the introduction of color flow, two-dimensional transcranial color-coded sonography (TCCS) has shown increased sensitivity for the detection of aneurysms.⁸⁶ The addition of intravascular contrast agents has been useful to improve the signal-to-noise ratio for examination. Klotzsch and colleagues reported the use of TCCS in 30 patients with known intracranial aneurysms.⁸⁷ Twenty-nine angiographically proven aneurysms with diameters ranging from 3 to 16 mm and a mean of 7.2 ± 3.6 mm were detected by TCCS. In another study, Baumgartner and associates evaluated 29 patients with 30 radiographically proven cerebral aneurysms.⁸⁶ Twenty-three of 27 (85%) nonthrombosed aneurysms with diameter 6 to 25 mm were identified.

Monitoring during Carotid Endarterectomy

Monitoring probes can be placed unilaterally or bilaterally over the temporal windows for continuous recording of mean V_MCA without interfering with the surgical field during CEA. One distinct advantage of TCD over other types of monitoring is that TCD provides information relevant to all the major causes of perioperative cerebrovascular morbidity, including (1) hypoperfusion during cross-clamping, (2) intraoperative and postoperative emboli, (3) intraoperative or postoperative thrombosis, and (4) postoperative hyperperfusion syndrome.

Hypoperfusion during cross-clamping is a major concern during CEA. Controversy exists regarding the need for placement of an intraluminal shunt during CEA. Options include placement of a shunt in all cases, no shunt placement in any case, or selective shunting in patients at high risk for cross-clamp hypoperfusion. The relative change in blood flow can be estimated by expressing the V_MCA determined 10 to 20 seconds after cross-clamping as a percentage of the V_MCA immediately before clamping.

Halsey published results from a collaborative study of 1495 CEAs in an attempt to define the relationship between intraoperative measurements of V_MCA and the need for shunt placement.⁸⁸ Patients were categorized according to their risk for ischemia (cross-clamp hypoperfusion) based on the reduction in V_MCA values during carotid artery clamping. The study indicated that patients at high risk for ischemia (85% reduction in V_MCA) who did not undergo shunting were at high risk for perioperative stroke. In more than 75% of patients with a low risk for ischemia (V_MCA fell by <60%), there was a better outcome in those in whom shunting was not performed. In the group at intermediate risk for ischemia (V_MCA fell 60% to 85%), there was no difference in outcome between patients with shunting and those without. Based on these data, if V_MCA stabilizes to at least 40% of the preclamp value, it is unlikely the patient would benefit from a shunt. Comparison studies have been conducted between TCD and other methods of identifying patients at risk for cross-clamp ischemia who may benefit from an intraluminal shunt. There appears to be excellent correlation between TCD and electroencephalographic (EEG) criteria for ischemia.^89,90

Another advantage of TCD monitoring is the ability to obtain immediate feedback regarding the adequacy of shunt function. When a shunt is used, TCD confirms restoration of MCA flow and provides a continuous monitor of shunt function. An obstructed shunt causes an immediate reduction in flow velocity that precedes any change in cerebral function or EEG parameters. The major disadvantage of TCD monitoring is failure to obtain adequate signals because of hyperostosis in some patients. It is useful to have an alternative form of monitoring available if this circumstance arises. The authors use a combination of intraoperative TCD and EEG recording to detect ischemia during cross-clamping and use selective shunting if hypoperfusion is detected.

Cerebral emboli are a major cause of perioperative stroke from CEA.^91,92 Microemboli detected by TCD are associated with stroke and ischemic events during and after CEA. Intra-arterial microemboli can be recognized easily with TCD monitoring by the characteristic chirping, clicking, or whistling sounds on the audio output and spectral display. The TCD features of particulate and gas emboli have been characterized and software has been developed for automatic detection and counting of emboli.^93,94 The prevalence and timing of perioperative cerebral emboli associated with CEA must be considered when using TCD monitoring. Some cerebral microemboli are detected during most CEAs (97% in one report⁹⁵); however, most microemboli do not result in stroke. Most emboli are detected on restoration of ICA perfusion, but these are predominately air emboli, which are believed to be less harmful than particulate emboli and not associated with ischemic events. Emboli noted during dissection of the carotid artery⁹⁶ or frequent emboli detected immediately postoperatively are associated with postoperative stroke and ischemic events.^92,97–99 Dextran or other antiplatelet drugs can be effective in reducing postoperative microemboli.^100,101 Information gained from TCD monitoring may allow the surgeon to alter surgical technique or institute treatment postoperatively to reduce ischemic events.

Intraoperative or postoperative occlusion may be due to technical problems at the site of endarterectomy, such as thrombosis or an intimal flap, and may be affected by coagulability factors. Findings on TCD that suggest impending or completed carotid occlusion include frequent and increasing microemboli and a decrease in V_MCA to postclamp values.⁹² Neurological deterioration in the recovery room should prompt immediate TCD evaluation for impending or completed postoperative occlusion, and if ipsilateral flow is poor, rapid re-exploration may be indicated to prevent stroke.

Postoperative hyperperfusion syndrome occurs in about 1% of patients after CEA, and when cerebral hemorrhage occurs, the prognosis is poor.^91,102 The syndrome is thought to occur when abnormally high blood flow develops in vascular beds that have been habituated to low perfusion pressure and are suddenly exposed to normal arterial pressure after restoration of normal perfusion pressure during CEA.^103,104 Hyperperfusion can be associated with defective cerebral autoregulation at normal blood pressure¹⁰⁵ and can be diagnosed by TCD before clinical signs develop. Elevations in ipsilateral V_MCA ranging from 30% to 230% are found in 10% to 20% of patients after CEA, but only in some of these patients do headaches or more serious sequelae develop.¹⁰² Early diagnosis with TCD allows strict control of blood pressure to restore flow velocities to the normal range and avoid serious complications.

Posttraumatic Vasospasm

Cerebral vasospasm can occur after head injury and is associated with traumatic subarachnoid hemorrhage. Posttraumatic vasospasm has been well documented and has been shown to be responsible for clinical deterioration and infarction in some patients with closed head injury.^56,106–108 TCD is useful in monitoring head-injured patients for this complication. Initial investigations by Weber and colleagues reported a 40% incidence of vasospasm, by TCD criteria, in a group of severely head-injured patients.¹⁰⁹ Martin and coworkers reported a 27% incidence of vasospasm with the use of TCD in a group of head-injured patients.¹⁰⁷ Detection of vasospasm in head-injured patients with TCD is made more difficult by the concurrent existence of cerebral hyperemia after head injury, which may result in false-positive values for vasospasm if the criteria for vasospasm in patients with subarachnoid hemorrhage are used. Blood flow velocity recordings should be obtained from the most distal portion of the extracranial ICA in the neck to allow calculation of V_MCA/V_ICA ratios. The V_MCA/V_ICA ratio should be used to correct for hyperemia when using TCD to detect posttraumatic vasospasm. Measurement of cerebral blood flow with quantitative techniques is helpful in identifying patients at risk for cerebral ischemia.¹⁰⁶

Vascular Dissection

Carotid artery dissection extracranially and at the skull base can be a complication after head injury that can sometimes produce devastating delayed cerebral infarction. One of the diagnostic clues to extracranial dissection can be the presence of intracranial arterial emboli. We determined the incidence of intracranial emboli in the MCA on the side of ICA dissection in 10 patients with ICA dissection secondary to trauma and in 7 with spontaneous ICA dissection.⁴⁸ The diagnosis was confirmed by carotid angiography and studied by TCD from the time of diagnosis through initiation of therapy.

Emboli were detected in the MCA distal to the dissection in 10 of the 17 patients (59%). Patients with microemboli detected by TCD experienced a stroke (70%) much more frequently than did patients without emboli (14%; P = .0498). It may be helpful to use TCD as an adjunctive tool to manage patients with suspected carotid dissection, and TCD may prove useful in evaluating the efficacy of treatment in reducing microemboli and subsequent stroke.

Transcranial Doppler Findings with Increased Intracranial Pressure

Changes in V_MCA recordings have been described with increasing ICP. With early rises in ICP, the pulsatility index increases with a progressive reduction in diastolic velocity and initially no change in mean velocity. With further increases in ICP, there is a simultaneous reduction in mean velocity and a further increase in the pulsatility index.¹¹⁰ It is likely that these changes occur as a result of progressive increases in resistance of the cerebrovascular bed. At high ICP values that produce arrest of the cerebral circulation, the V_MCA waveform progresses through a predictable sequence. This sequence begins with a progressive reduction in diastolic velocity initially to zero and then commencement of an alternating flow pattern with a reversed diastolic component (Fig. 347-5). Further obstruction to flow causes progression to alternating flow, which is antegrade in systole and retrograde in diastole with equal components. Further progression leads to small systolic peaks and then an absence of the waveform entirely. These characteristic TCD changes have correlated with previously described angiographic findings of nonfilling of the cerebral arterial tree in patients undergoing cerebral angiography.¹¹⁰ It has been documented by cerebral angiography and radionuclide isotope scanning that an alternating flow pattern on TCD corresponds to arrest of the cerebral circulation (Fig. 347-5).^110,111

FIGURE 347-5 Tracing indicating progression of the velocity waveform from a normal pattern to one seen with cerebral circulatory arrest. With decreasing cerebral perfusion pressure (CPP) or increased resistance to flow, there is a progressive decrease in diastolic velocity evolving to oscillating flow and finally to small systolic peaks and to zero flow. The lower panel indicates transcranial Doppler findings in cerebral circulatory arrest.

(Adapted from Hassler W, Steinmetz H, Pirschel J. Transcranial Doppler study of intracranial circulatory arrest. J Neurosurg. 1989;71:195.)

Changes in the pulsatility index and mean velocity alone are not specific for increased ICP. A high resistance in the cerebral vascular bed that produces increased pulsatility and decreased velocity can be seen with intense vasoconstriction. Intense vasoconstriction can occur as a result of vigorous hyperventilation, low cerebral metabolism, and cerebral vasoconstricting drugs such as barbiturates. For these reasons, mean velocity and the pulsatility index, when taken in isolation, cannot predict ICP reliably. The pulsatility index can be affected markedly by cardiac factors.¹¹² Progressive increases in the pulsatility index can indicate a progressive rise in ICP in patients who are observed carefully and known to not have concomitant cardiac changes and alterations in CO₂ or other factors that produce vasoconstriction.

Cerebrovascular tone pressure, as well as the critical closing pressure of the cerebral circulation, can be determined by analysis of the V_MCA waveform and the arterial pressure waveform. It may be possible in the future to predict ICP more reliably from these parameters with TCD and arterial blood pressure recordings.^113,114

Transcranial Doppler as a Confirmatory Test for Brain Death

TCD sonography is useful as a confirmatory test in the diagnosis of brain death. Arrest of the intracranial circulation can be detected reliably with TCD, which is often but not always present when brain death occurs. The true diagnosis of brain death is a clinical diagnosis, and blood flow studies such as cerebral angiography, radioisotope scanning, and TCD are confirmatory tests. According to the President’s Commission on guidelines for determination of brain death in the United States, confirmatory tests can be used to shorten the observation period in patients who fulfill the clinical criteria for brain death before organ retrieval or discontinuation of mechanical ventilation can proceed.¹¹¹

Many studies have examined the role of TCD in the determination of brain death by comparing clinical findings and the results of blood flow studies with sonographic waveforms in patients progressing to brain death.^{111,115–118} Most of the studies comparing TCD with clinical criteria for brain death have indicated a sensitivity and specificity of TCD approaching 100%.

Several caveats apply to TCD and any other cerebral blood flow methods as confirmatory tests for brain death. False-negative and false-positive results can occur in rare circumstances. False-positive TCD recordings have been reported in several situations. Cerebral circulatory arrest can be transient and reverse after a period of minutes in patients with subarachnoid hemorrhage who have a marked increase in ICP that results in temporary arrest of the cerebral circulation.^119,120 For this reason, when using TCD to confirm arrest of the cerebral circulation, it should be repeated or be documented that cerebral circulatory arrest is present for a sufficient period (i.e., 30 minutes under normothermic conditions) to be incompatible with survival. Another possible cause of false-positive TCD findings can be found in patients who have sustained cerebral circulatory arrest but have residual circulation to the brainstem, which may sustain respiratory effort or minimal cranial nerve function for a short time. False-positive TCD tracings showing alternating flow velocity patterns have been seen in unusual situations in which abnormally low diastolic pressure is present, such as in patients with intra-aortic balloon pumps.

False-negative TCD results (in which TCD indicates preserved flow and patients are clinically brain-dead) have been reported in patients with clinical brain death. When massive destruction of the brainstem occurs as a result of destructive lesions or posterior fossa lesions, brainstem function may be absent, and the patient fulfills the clinical criteria for brain death, but with preserved supratentorial blood flow. EEG activity may persist, and TCD waveforms can indicate preserved blood flow.¹²¹ These examples of false-positive and false-negative studies are not an indication of failure of the TCD technique itself, but they emphasize why confirmatory tests do not always concur 100% with the clinical diagnosis of brain death. Despite these shortcomings, TCD can be especially useful in patients in whom the clinical diagnosis of brain death is made difficult because of extensive trauma and cranial nerve damage and in patients after trauma who have been given paralytic agents, which makes clinical evaluation difficult. TCD examination may be helpful in avoiding unnecessary surgery in this situation.

Sonothrombolysis

The majority of investigations of the utility of neurosonology have been directed toward diagnostic applications of this technology. Experimental and clinical work in ultrasound has revealed that ultrasound has potential therapeutic uses, including treatments used in dentistry and for musculoskeletal disorders, hemostasis, treatment of fibroid tumors, lipolysis, and other applications. One property of ultrasound with potential applications in the nervous system that has recently been discovered and characterized is the ability to accelerate the dissolution of blood clots, a process referred to as sonothrombolysis. It has also been demonstrated that ultrasound can potentate the effect of tissue plasminogen activator (t-PA) on the dissolution of blood clots. Intravascular catheters with small ultrasound-emitting probes have been used in combination with t-PA for the dissolution of intravenous and intra-arterial blood clots in patients with peripheral vascular disease. More recently, intracranial intra-arterial thrombectomy via endovascular sonothrombolysis has been evaluated in clinical trials. Independent of this research, it was serendipitously discovered as part of a clinical routine in which t-PA–treated stroke patients are monitored with TCD for MCA recanalization that some patients can recanalize earlier than expected and experience dramatic clinical recovery.¹²² Subsequently, the feasibility and safety of sonothrombolysis of MCA clots in stroke patients at the bedside with TCD have been established.^122,123 These results have been confirmed by other groups worldwide that monitor t-PA infusion with diagnostic ultrasound.^124,125

Because intravenous t-PA by itself works by induction of mostly partial recanalization of large thrombi,^122,123,126 early augmentation of fibrinolysis to improve arterial recanalization is desirable. Fibrin strands are cross-linked and held by activated platelets to form a net-like structure that captures red blood cells (Fig. 347-6). To capture red blood cells, the net should have openings of less than 6 µm (note that the average size of erythrocytes is 6 to 8 µm). Thrombus size and structure will further create an area of high resistance to the antegrade movement of red blood cells, and once the embolus reaches the lumen of a vessel comparable to its size, it acts like a plug precluding flow if arterial pressure is insufficient to further distend the vessel. Delivery of intravenous t-PA to the thrombus is probably dependent on residual blood flow around the thrombus and plasma flow through the thrombus. Mechanical agitation with ultrasound can expose layers of thrombi to circulating t-PA and facilitate streaming of plasma through the thrombus, thus bringing more t-PA to binding sites. Ultrasound is a pressure wave that can travel through tissues and deliver this mechanical energy to stagnant flow areas and clot interfaces. Numerous in vitro experiments have confirmed the ability of ultrasound to enhance fibrinolysis through disaggregation of non–cross-linked fibrin strands and thereby further thin them and promote plasma flow and delivery of lytic drug to the clot.^127–129 Kilohertz frequencies are believed to induce more mechanical stretching of fibrin, whereas megahertz ultrasound promotes fibrinolysis more through enzymatic mechanisms.^130,131

FIGURE 347-6 Thrombus appearance on magnetic resonance angiography (top right image), digital subtraction angiography (middle right), high-resolution microscopy revealing net-like structure produced by acute fibrin clot with fibrin, platelets, and red blood cells that are trapped (top left), and computed tomographic angiography (bottom image). Power motion Doppler (insert with millimeter scale) shows detectable flow around the thrombus and patent branches.

A multicenter collaborative group was formed to confirm the effect of TCD in accelerating t-PA–induced thrombolysis in stroke patients in a clinical trial. In the Combined Lysis of Thrombus in Brain Ischemia Using Transcranial Ultrasound and Systemic TPA (CLOTBUST) trial,¹²³ 73% of patients achieved any degree of recanalization (46% complete, 27% partial) with t-PA plus TCD versus 50% (17% complete, 33% partial) with t-PA alone within 2 hours of treatment (P < .001). The rate of sustained complete recanalization at 2 hours was 38% versus 13%, respectively (P = .03). The symptomatic intracerebral hemorrhage rate was 3.8% in both groups (not significant).¹²³ This trial was the first properly powered clinical trial that confirmed the effect of ultrasound-enhanced thrombolysis in human subjects and demonstrated a positive biologic effect of diagnostic low-power ultrasound.

A recent meta-analysis of six randomized and three nonrandomized clinical studies of sonothrombolysis suggested that any diagnostic ultrasound monitoring can at least double the chance of early complete arterial recanalization at no increase in the risk for symptomatic intracerebral hemorrhage.¹³²

Precision ultrasound delivery to the thrombus can be achieved with refinement of operator-independent ultrasound targeting, imaging technologies, contrast microspheres tagged to recognize and adhere to the thrombus, crossed beams under MRI guidance, and direct embedding in the clot. Some of these technologies are undergoing evaluation and some are already in clinical trials.

Because the application of frequencies below diagnostic range (i.e., kilohertz) in humans resulted in increased symptomatic bleeding rates¹³³ whereas 2 MHz showed a strong signal of efficacy and safety,¹²³ mechanisms by which kilohertz and megahertz frequencies interact with the clot–residual flow interface, endothelium, and brain tissues are currently under investigation. Catheter-based ultrasound delivery to arterial thrombi and intraventricular clots is the subject of ongoing clinical trials.

Molina and colleagues pioneered the use of gaseous microspheres in combination with CLOTBUST monitoring methods and reported safety and recanalization rates with the first-generation air-filled microspheres (Levovist), as well as with the newer diagnostic microspheres.¹³⁴ These microspheres were originally designed as ultrasound contrast agents to improve the quality of ultrasound images. After intravenous injection, the microspheres circulate in the bloodstream and cross lung capillaries into the arterial circulation. When contacted by an ultrasound beam, the microspheres undergo expansion in size followed by transient oscillation or complete breakup (Fig. 347-7). The likelihood of oscillation or cavitation with microsphere destruction may depend not only on the resonant frequency but also on the power of the ultrasound delivered to the target tissues.

FIGURE 347-7 The upper images show microphotography of destruction of a single microsphere by an ultrasound pulse. The time axis is in microseconds. The lower images reveal microspheres permeating around or through thromboembolic material in the middle cerebral artery to areas of no flow before treatment (middle bottom image, “initial permeation”). It was possible to detect this phenomenon in a clinical feasibility study because of initial dilution of the microspheres and the ability of transcranial Doppler (TCD) to detect a single microsphere in the intracranial circulation. The bottom right image shows partial recanalization detected by TCD at 7 minutes of treatment with tissue plasminogen activator, microspheres, and TCD. This finding was confirmed by repeat computed tomographic angiography (image insert above the “thrombus”).

Observations of in vitro experiments with a temporal bone/MCA flow model have indicated that even low-power TCD with conventional TCD equipment can destroy some of the intra-arterial microspheres and thereby lead to a mechanical effect. When microspheres oscillate or burst, they expand like microscopic balloons,¹³⁵ move in three dimensions, and transmit mechanical energy momentum from passing ultrasound waves to surrounding fluid and structures. If an ultrasound beam intercepts microspheres at the clot–residual flow interface, fibrinolysis is facilitated with or even without t-PA.

Therapeutic use of the newest generation of microspheres, such as perflutren-lipid microspheres, which are more stable in solution (they tend to “bubble up” less than their predecessors) and are therefore suitable for continuous infusion, has been further explored. This in turn enables continuous replenishment of microspheres at the clot–residual flow interface during the entire t-PA infusion. A multicenter feasibility study showed that perflutren-lipid microspheres reached intracranial thrombi in all patients, and in 75% they immediately permeated through or around thrombi and reached areas with no detectable flow (Fig. 347-7).¹³⁶ The first dose of these microspheres was administered safely during t-PA infusion and did not cause any symptomatic intracerebral hemorrhage. Perflutren-lipid microspheres, TCD, and t-PA lysed 50% of proximal MCA occlusions.¹³⁶ An additional multicenter microsphere dose escalation study called TUCSON (Transcranial Ultrasound in Clinical SONothrombolysis, NCT00504842) was recently completed and the results are pending. Transcranial ultrasound delivery in an operator-independent and dose-controlled manner is now being tested in a separate clinical trial.