The first and foremost limitation of TCD is that it measures blood flow-velocity, not flow. Because velocity is influenced by vessel diameter, blood viscosity, acid–base balance and temperature, an abnormally high velocity may indicate either hyper- or hypoperfusion. For example, hypercapnia may increase middle cerebral artery (MCA) diameter by more than 20 %, which markedly alters the relationship between flow and flow-velocity.

Second, the quality of TCD information is user-dependent. Correct vessel identification and accurate velocity measurement rely on the training, skill, experience, and practice of the sonographer [6]. Interobserver agreement is high among qualified practitioners who regularly make TCD measurements, but declines substantially among occasional users [6].

Third, effective transtemporal insonation of intracranial vessels is not possible in all patients. Both cranial hyperostosis and the presence of intracranial vascular disease may prevent monitoring through this cranial site in a significant fraction of patients. For example, successful TCD measurement through all intracranial ultrasonic windows was possible in 78 % of a large series of healthy elderly patients [7]. Alternative insonation sites may be utilized to partially overcome this limitation. Commercially available submandibular probe fixation devices enable continuous simultaneous insonation of the extracranial internal carotid artery (EICA) and internal jugular vein [8]. This approach permits monitoring of blood flow-velocity both toward and away from the brain as well as detection of gaseous and particulate emboli. In addition, intermittent brief transorbital insonation of the carotid siphon with a handheld probe [9] may be helpful in documenting the establishment and maintenance of selective antegrade or retrograde cerebral perfusion during systemic circulatory arrest for aortic arch reconstruction [10].

Fourth, TCD provides no direct information on the cause of observed velocity changes. Sudden signal loss may be caused by flow cessation or unintentional probe movement. Correct causal determination requires input from other monitoring modalities.

Transcranial Doppler is most widely used to quantify cerebral hemodynamic function. Measurements include (1) large vessel blood flow direction, and maximum, mean (Vm), and minimum velocity and pulsatility; (2) cerebral autoregulation; (3) vasomotor reactivity (VMR) to ΔCO₂; (4) neurovascular coupling (NVC) and cerebral emboli detection [11].

For the first application, in diagnostic settings, momentary intra- and extracranial velocity measurements obtained from the patient are compared with vessel-specific age-corrected normative values. During perioperative and critical care hemodynamic monitoring , the potentially profound influences of anesthetic and surgical management on cerebral hemodynamics preclude this approach. Instead, abnormality is inferred from trended changes in flow-velocity referenced to a preprocedure baseline [12]. Because the MCA typically carries approximately 40 % of the hemispheric blood flow, it is generally the preferred monitoring site [13]. The issue of correct vessel identification may be minimized by using a recording depth of less than 50 mm. With an appropriate probe angle in the adult cranium, at this “shallow” insonation depth, the only arterial signals arise from the MCA or its branches.

When MCA diameter and blood viscosity remain constant, the concept of noteworthy velocity change as an indicator of altered cerebral perfusion has been based on its relationship to the concomitant appearance of clinical signs. In conscious subjects, hypoperfusion-derived syncope appeared with a greater than 60 % decrease in mean flow-velocity or a total signal absence at end-diastole [14]. During general anesthesia, a greater than 60 % flow-velocity decrease coincided with a cerebral blood flow fall to less than 20 mL/100 g/min and pathologic EEG suppression [15]. Flow-velocity declines of greater than 80 % have been associated with a significantly increased stroke risk [16]. However, the clinical correlation with MCA flow-velocity has been less than perfect. In the setting of carotid endarterectomy with regional anesthesia, McCarthy et al. [17] found that clamp-related neurologic dysfunction occurred in just one-third of cases in which MCA flow-velocity declined greater than 60 %.

The rapid detection of a cerebral blood flow obstruction by TCD may be brain- or life-saving. Anesthesia providers often lack data on the functional status of a patient’s intracranial arteries and veins. This paucity of knowledge increases the risk for a generally avoidable and potentially injurious cerebral blood flow obstruction. Despite the limitations described previously [17], TCD can rapidly identify an MCA flow obstruction during carotid endarterectomy or carotid artery stenting associated with head positioning, excessive vascular traction, carotid occlusion [16], dissection [18], or hematoma [19]. The optimal approach to intravascular shunt use remains controversial. Nevertheless, a meta-analysis of 32 studies with pre- and postcarotid endarterectomy brain imaging noted a 38 % reduction in brain infarct incidence in groups utilizing neuromonitoring-based selective shunting [20].

In the repair of an acute aortic dissection , initiation of total cardiopulmonary bypass through a femoral artery may redirect blood flow through the false lumen. Cerebral inflow obstruction may then result in a potentially lethal malperfusion syndrome. TCD can instantly detect its development and guide successful surgical correction [8].

TCD-detectable inflow obstruction may be seen during attempted retrograde cerebral perfusion even with the presence of venous effluent from open carotid arteries. With an internal jugular functional valve [21], retrograde flow of oxygenated blood from the superior vena cava will be directed upward through extracranial veins and downward through the extensive azygous system. Without a functional jugular valve, momentary complete systemic circulatory arrest may lead to cerebral venous collapse. Initiation of effective retrograde flow may require a brief pressure of more than the oft-recommended 25 mmHg to restore flow in these collapsed vessels [22]. Currently, TCD offers the only direct method to verify the establishment of retrograde cerebral perfusion . Using this approach, Estrera et al. [23] demonstrated the effectiveness of TCD in the management of retrograde cerebral perfusion. Interestingly, the majority of clinical studies describing experience with this perfusion technique used neither TCD nor cerebral oximetry to document successful initiation or maintenance of bihemispheric retrograde flow.

TCD is also invaluable for the documentation of selective antegrade cerebral perfusion during systemic circulatory arrest [8]. Bilateral TCD monitoring can assure the surgeon that both MCAs are appropriately perfused via a single right axillary, innominate, or carotid cannula. Further, Neri et al. [24] showed that selective antegrade perfusion could prevent the cerebral dysautoregulation that typically follows a period of hypothermic total circulatory arrest.

Before or immediately after cardiopulmonary bypass , TCD can detect and guide correction of a cerebral outflow obstruction occurring in response to a malpositioned venous perfusion cannula. The resultant cerebral vascular resistance increase is manifested by a decline or absence of flow-velocity during end-diastole [25].

There are both technical and physiologic explanations for the observed large MCA flow-velocity declines without clinical correlates. With insonation depths of greater than 55 mm, inadvertent insonation of the distal ICA may occur near the bifurcation into the middle and anterior cerebral arteries. Carotid clamping will markedly lower ICA velocity, even with adequate collateral flow through the circle of Willis. Alternatively, vigorous collateral flow through leptomeningeal collaterals may maintain hemispheric function despite MCA hypoperfusion [26].

Currently, TCD provides the only continuous, direct measure of cerebral hyperperfusion . The perioperative monitoring of cerebral perfusion is far from routine. Thus, the incidence as well as the clinical and socioeconomic significance of this all-too-common problem is under-appreciated. In fact, a surprisingly high (10 %) incidence of postoperative symptomatic hyperperfusion has been reported to occur after carotid endarterectomy [27] or carotid angioplasty and stenting [28]. It should be appreciated that this is at least five times higher than the oft-reported incidence of perioperative hypoperfusion toward which most carotid surgery neuromonitoring resources are directed [29].

Cerebral hyperperfusion is defined as a blood flow increase well in excess of metabolic demand, with flow-velocity increases of higher than 100 % above baseline [30]. Clinical manifestations include severe focal headache, face and eye pain, seizures, focal neurologic deficit, and cognitive disturbances including delirium [27]. Because structural damage and widespread edema are typically absent, diagnostic imaging studies are often uninformative. Without the rare complication of an intracerebral hemorrhage [29], most clinical signs of the hyperperfusion syndrome are transient and self-limiting, albeit expensive for the healthcare delivery system [30]. However, there may be a persistent impairment of cognitive or neuropsychologic function leading to a decreased quality of life [31, 32]. The full economic impact of postoperative or critical care hyperperfusion is incompletely understood (e.g., hospital readmission, rehabilitation attrition, impaired independence, slow return to work) [33], and specific syndrome components are associated with longer and more costly hospital stays [34].

The relationship between cerebral perfusion and systemic perfusion pressure is unpredictable. For example, during adult cardiopulmonary bypass, cerebral autoregulation remains intact in only half of patients [35] and the lower mean arterial pressure limit range is approximately 50 mmHg [36].

Optimal patient management requires continuous information on the adequacy of cerebral perfusion. An initially intact autoregulation may be transiently disrupted as a consequence of anesthetic technique, i.e., spinal anesthesia [37], or surgical necessity, i.e., deep hypothermic circulatory arrest [8]. Alternatively, it may be preserved through the adoption of improved perfusion techniques, i.e., supplemental antegrade or retrograde cerebral perfusion [38].

Transcranial Doppler and cerebral oximetry are two approaches to achieving this goal. Information provided by the two techniques may sometimes be complementary since TCD measures large vessel flow-velocity while cerebral oximetry measures microcirculatory hemoglobin oxygen saturation at the site of tissue gas exchange. Thus, hypoxia decreases MCA blood flow-velocity via vasodilation, while microcirculatory regional O₂ saturation (rSO₂) declines in response to reduced O₂ delivery [39]. However, in other cases, the two measures of cerebral perfusion may appear to be in conflict. For example, the vasopressor phenylephrine increases MCA velocity and decreases rSO₂, while the opposite occurs with administration of the hypotensive agent, sodium nitroprusside. This seeming paradox has been explained by drug-induced changes in MCA diameter [5].

The use of TCD to determine VMR is well established [40]. It describes the cerebral hemisphere-specific relationship between PaCO₂ and MCA blood flow or flow-velocity. Early knowledge of the patient’s VMR status may be of considerable value during perioperative anesthetic management [40]. First, VMR is a precondition for cerebral autoregulation [41]. With a normal VMR (i.e., 4 % Δ velocity/mmHgCO₂), TCD can define both the lower limit of autoregulation and the zero-flow pressure below which perfusion ceases [42]. A subnormal VMR and associated dysautoregulation identify pressure-passive hemispheric perfusion and an increased stroke risk, which may lead to adjustments in the anesthetic plan [43]. Second, the presence of normal VMR alerts anesthesia providers to the potential hazards of hypocapnia. Aggressive hyperventilation of CO₂-reactive patients following endotracheal intubation may decrease MCA flow-velocity by more than 30 % and produce ischemic EEG suppression [44]. Third, TCD identifies hemispheric asymmetry in VMR, which can lead to blood flow steal from the non-CO_2--reactive hemisphere in the presence of hypercapnia [45]. Fourth, knowledge of the VMR status is important during the planning for deep hypothermia. In both pediatric [38] and adult patients [46], optimal brain cooling with high cerebral blood flow appears to be best achieved through pH-stat acid–base management, while rewarming favors alpha-stat control. Establishment of these ideal conditions requires CO₂-reactive arteries in both cerebral hemispheres.

Vasomotor reactivity and cerebral autoregulation are related, but distinct, phenomena. Autoregulation requires normal VMR, but may become suppressed while cerebral artery CO₂ responsiveness remains intact. For example, volatile anesthetics [47] and other vasoactive agents [48] may diminish TCD-assessed autoregulation without affecting cerebral artery constriction with hypocapnia. Additionally, it should be appreciated that the widespread notion of a universally intact cerebral autoregulation with a lower limit of 50 mmHg is based on a 1953 study of 15 conscious pregnant women, half of whom were toxemic [49]. In fact, the 1973 article that published the now-familiar sigmoid autoregulatory graph of blood pressure vs. cerebral blood flow emphasized that the lower limit was highly variable and in hypertensive patients often above 100 mmHg [50].

Neurovascular coupling refers to the mechanism that adapts local cerebral blood flow to changes in associated neuronal activity in the healthy brain [51]. Van Alfen et al. [52] used TCD and EEG recording to illustrate the maintenance of NVC during brain cooling. In aortic surgery requiring hypothermic circulatory arrest, cooling to 25 °C partially suppressed cerebral metabolic activity and resulted in a burst-suppression EEG pattern. Despite unchanged extracorporeal blood flow, the TCD recording evidenced an oscillating flow-velocity trend, with peaks occurring approximately 5 s after the onset of each burst of EEG activity; velocity nadirs occurred with a similar delay following cessation of each burst.

Circulatory arrest may disrupt NVC due to vasoparesis. Resultant uncoupling observed with TCD monitoring during subsequent reperfusion and rewarming may portend developing brain injury if NVC is not restored [53].

Peca et al. [54] combined TCD with visual evoked potentials (VEPs) to compare NVC in healthy subjects and patients with amyloid angiopathy. During photic stimulation, uncoupling was evident in the patients because flow-velocity responsiveness was suppressed while VEP changes were the same as in healthy subjects [54].

Combined TCD and EEG recording during anesthetic induction demonstrated that NVC was maintained with propofol. In contrast, sevoflurane appeared to disrupt NVC with resulting luxury cerebral perfusion and EEG suppression [55]. Prompt flow adaptation to changing metabolic needs appears to be caused by an incompletely understood local vasodilation evoked by neuronal activation. The relative contribution of NVC and systemic factors to cerebral blood flow regulation is currently unknown [56].

Of the currently available neuromonitoring modalities, TCD is unique in its capacity to detect both gaseous and particulate emboli within the cerebral circulation. TCD has detected cerebral emboli associated with cardiac [57], carotid [58], laparoscopic [59], and hip and knee surgery [60]. During postoperative or critical care, such emboli have been observed in patients with (1) new mechanical aortic valves [61], (2) infective endocarditis [62], (3) atrial fibrillation [63], or (4) those with implanted ventricular assist devices [64]. The presence of persistent TCD-detected embolization may lead to the development of focal neurologic deficits. However, the clinical expression of embolic injury appears to be a threshold phenomenon that may require a persistent HITS rate of more than two per minute [65]. With the incorporation of this threshold concept, TCD-guided antiplatelet therapy has successfully reduced particulate-based HITS and associated neurodeficit signs following carotid endarterectomy [66].