This relationship between cerebrovascular resistance and resistance in the umbilical arteries provides the basis for fetal CBF monitoring using Doppler ultrasound. The cerebral/placental ratio (CPR) is also known as the cerebral/umbilical ratio and is a primary indicator of fetal blood flow distribution (25–30). This metric is a ratio of the relative pulsatility indices (PI) in the cerebral and placental circulations (as measured in the umbilical artery, UmA) defined by the equations,

However, limitations to this method of CA monitoring exist. In the third trimester, low fetal CPR is associated with high uterine artery Doppler indices and is thought to be a consequence of impaired placental perfusion that increases the risk for stillbirth (31). However, it may also result from cerebrovascular vasodilation, which is common in fetuses growing poorly and is discussed further in the section on intrauterine growth restriction (IUGR). The use of CPR as a screening tool in otherwise healthy pregnancies has been shown to only weakly predict adverse outcomes and there is concern that overuse of this metric may expose women and neonates to unnecessary interventions such as earlier gestational age delivery (32).

Similar to prenatal Doppler ultrasound monitoring of CBF, postnatal measurements can be useful to determine cerebrovascular dilatation as well. The primary metric used postnatally is the resistive index (RI), which is derived from the systolic and diastolic velocities in the cerebral arteries as described,

Normative values for newborn cerebral artery RI are well defined and typically fall between 0.55 and 0.8 (33, 34). Although RI is not sensitive to the angle of insonation, there are limitations to its use including variations based on location of measurement (e.g., anterior fontanelle or transtemporal) (33), and differences based on technique (35). As such, a recent systematic review concluded that there is no clear evidence to support following RI in preterm infants (36), although RI may still have utility in neonates with hypoxic-ischemic encephalopathy (HIE) (37).

Advanced magnetic resonance imaging (MRI) technologies that contribute to adult research and care are now finding utility in pregnant people and the fetus. Using advanced MR imaging techniques that combine fetal cardiac MRI with T2 mapping to determine tissue oxygen saturations have enhanced our understanding of fetal circulation (38). These experiments effectively replicate in the human fetus, those classic studies in the chronically instrumented pregnant ewe that form the basis for our current understanding of fetal and placental hemodynamics (39–42). Several key differences in the human fetus have been described by Seed and colleagues, who are pioneering the application of these new technologies (43–46).

T2 mapping is an MRI method that allows identification of small blood vessels within tissue. When combined with phase contrast MRI, which quantifies velocity of moving fluids, accurate measurements of fetal blood flow are possible. One major obstacle to this imaging is the necessity of timing image acquisition with the cardiac cycle (47–51). Cardiac gating is used to time data acquisition with specific phases of the cardiac cycle, and MRI-compatible fetal ECG devices are now available (48, 51–55). This technology has now been applied to fetal disease states including IUGR (54) as well as the fetus with cardiac malformations (38).

Modern NIRS is based on the pioneering work of Professor Frans F. Jöbsis, and since his landmark study in 1977 (56), its use has expanded exponentially. NIRS has the advantage of being non-invasive, inexpensive, and continuous, therefore it is the most commonly used technique to measure regional cerebral saturation (ScO₂) in neonates despite the lack of large randomized clinical trials to support its clinical impact. The NIRS signal represents the oxygen saturation of Hg in the underlying brain tissue and is derived from the difference in absorption of near-infrared light by the oxygenated and deoxygenated heme groups (57). However, the depth of tissue interrogation can be relatively low and likely only provides information on the cortex (58, 59). Using NIRS to estimate CBF relies on the assumptions that the overwhelming majority of oxygen in blood is bound to hemoglobin, total hemoglobin is proportional to cerebral blood volume, and there is uniform oxygen consumption within the brain. The NIRS ScO₂ value represents a “weighted average” of arterial, venous, and capillary oxygenation with approximately 75–80% of the signal originating from the venous measurement (60). In general, ScO₂ values between 65 and 90% are considered normal in healthy term neonates (61, 62).

Only when coupled with a measure of CPP can NIRS impart information about CA. Using NIRS to determine periods of intact and impaired CA requires additional data as well as careful interpretation. Systems that provide time-synchronization of the continuous signals from invasive blood pressure monitors, NIRS, peripheral oxygen saturation, heart rate, and respiratory rate are enhancing the utility of NIRS as a monitor for CA.

Wavelet transform coherence analysis was developed as an alternative tool to Fourier transform to analyze non-stationary signals. Its application to our understanding of neonatal CBF has evolved over the past 10 years (63–65). Due to the non-stationary ScO2 signal provided by NIRS, traditional Fourier transform is inadequate to create a unified analysis across the time-frequency domain without dividing the data into arbitrary “bins” or frequency intervals that require an artificial simplification of the signal into a pseudo-stationary form for analysis. The use of wavelet transform coherence analysis does not require such simplification, but instead allows analysis of signal coherence across a spectrum of the time-frequency domain. The application of wavelet transform coherence analysis has been used extensively for neonates with HIE and is further discussed in that section.

Multiple other analytic techniques utilize NIRS or its component values to determine the correlation between CBF and MAP. The cerebral oximetry index uses the regional cortical oxygen saturation and has been validated in animal models of hypotension (66, 67) followed by clinical trials in pediatric cardiac surgery (68, 69). The pressure reactivity index has been used extensively in traumatic brain injury patients and requires invasive intracranial pressure monitoring, thus, it is not suited for neonates. The hemoglobin volume phase index was developed by Lee et al. using the regional hemoglobin from the NIRS calculation to estimate CBF, and is a non-invasive correlate of the pressure reactivity index (70–72). This method has been used to establish an individualized optimal MAP and time spent below this optimal MAP correlated with adverse outcomes (73). The significant differences in the calculated limits of autoregulation based on each of these methodologies complicates their use in clinical practice and necessitates validation of each methodology in specific populations and disease states.

Despite their widespread use, no randomized studies to date have systematically investigated potential benefits or risks associated with specific ScO₂ target ranges or provided validation of these analytical tools in any neonatal population. Normative NIRS signals vary considerably by commercial sensor and by individual patient physiology. Only recently investigators have reported on the mathematical conversions to allow comparison of NIRS data between devices (74, 75). Due to these limitations, the European Society of Paediatric and Neonatal Intensive Care recommends against routine clinical use of NIRS in all children with hemodynamic instability and there is no agreement from the group on the significance of a decline in ScO₂ from baseline or a lower limit of normal (76).

Early studies on CBF in neonates used a modified xenon-133 clearance method that relied on external scintillation detectors (77–79). These earlier discoveries related to the influence of Hg, hypoglycemia, and blood pressure on neonatal CBF have informed much of the contemporary investigations into neonatal CBF homeostasis. Recently developed methods to estimate CBF based on heart rate variability show promise since invasive monitoring of MAP is not always possible or desirable in all neonates (80). Heart rate variability refers to the variation in time intervals between each heartbeat and is controlled by the autonomic nervous system. This variability is affected by hypoxia and other conditions that compromise CBF, thus making the metric a possible predictor of disturbed CBF (81, 82). One key limitation to using heart rate variability as a clinical tool is correlating it with other physiologic parameters of cerebral oxygenation and the real-time data analysis required for it to be useful (83).

Another method of CBF monitoring is superior vena cava (SVC) flow which is based on the physiologic principle that cerebral venous return contributes as much as 70–80% to total SVC flow. This technique has been applied in preterm populations and may correlate with adverse outcomes (84), although it is subject to influence by multiple factors. In one study, low SVC flow was associated with lower gestational age, higher upper body vascular resistance, larger diameter patent ductus arteriosus (PDA) shunts, and higher mean airway pressure (85). Together, these variables can make interpretation of SVC flow difficult. In addition, SVC flow only correlates weakly with estimations of CBF by cerebral NIRS (86), adding complexity to its use. With further refinement this technique may contribute to the overall assessment of CBF, and is the subject of ongoing investigations to determine its predictive validity on long-term outcomes in preterm neonates (87).

CPR is used in pregnancies complicated by IUGR to quantify the redistribution of cardiac output with increased flow to the fetal brain (traditionally called a “brain-sparing” effect) (88, 89). However, this term may be a misnomer. In a secondary analysis of the large multicenter Prospective Observational Trial to Optimize Pediatric Health in IUGR (PORTO) Study, investigators showed that this redistribution of blood flow to the brain is associated with adverse perinatal outcomes including intraventricular hemorrhage, periventricular leukomalacia, HIE, necrotizing enterocolitis, bronchopulmonary dysplasia, sepsis, and death (90). In fact, fetuses with IUGR and CPR <1 had poorer neurodevelopmental outcomes at 3 years compared to pregnancies with abnormal umbilical artery Doppler alone in the follow up report of 375 patients from the PORTO Study (91). These studies clearly demonstrate that cerebral vasodilation does not result in normalization of brain development.

The cause of this hemodynamic disturbance is most frequently related to placental insufficiency and can lead to deleterious alterations in vasoreactivity of cerebral vasculature when prolonged (92). In fact, IUGR fetuses with normalization of MCA-PI after prolonged vasodilation are at increased risk for fetal demise (93). Postnatally, neonates who experienced IUGR have persistent vasodilation of cerebral vasculature (94). This can result in higher ScO₂ values by NIRS likely due to a combination of increased cerebral oxygen delivery and a decreased cerebral tissue oxygen consumption (95). The key factor in using this technique is to not rely soley on the single timepoint measurement; rather, a trend in CPR is required to determine where on the trajectory a fetus is in regard to CA.

A fetus is typically evaluated for fetal anemia (defined as Hg levels less than mean for gestational age (GA) (96) or multiples of the median (MoM) for GA) (97–99) when it is noted to have abnormal fluid collections (ascites, pericardial effusion, hydrothorax, and/or skin edema) or reduced movements. The gold standard for the diagnosis of fetal anemia is fetal blood sampling (FBS), a procedure that imparts a high risk for fetal loss, particularly in hydropic fetuses (100). Doppler ultrasonography can detect fetal anemia based on an increase in the peak velocity of systolic blood flow in the MCA (MCA-PSV), mitigating the risk of fetal loss (97, 100). This non-invasive metric relies on the fetal cerebral autoregulatory capacity to increase CBF in the setting of hypoxia, similar to that seen in fetal IUGR.

The basic principle used by Doppler for the diagnosis of fetal anemia, relates to the observed increase in blood velocity which stems from the increased fetal cardiac output. Due to relative maintenance of cardiac output for small changes in Hg, there is no strong correlation between MCA-PSV and fetal Hg within the normal range. If significant Hg decreases occur, however, the MCA-PSV increases and can be used to determine the Hg value with a fair degree of accuracy (100). An MCA-PSV > 1.5 MoM is used as a cut-off for severely anemic fetuses. In non-hydropic fetuses, sensitivity of a single measurement of MCA-PSV varies between 75 and 95%, with about 10% of false-positive results, and this increases if the study is performed after 35 weeks GA (101). Use of the multiple time points of MCA-PSV can decrease the false positive values (101). As such, for high-risk fetuses, MCA-PSV are typically performed at regular intervals depending on risk assessment. Fetuses with anemia are at risk for multiple complications and may benefit from intrauterine transfusions, underscoring the importance of assessment at regular intervals in those deemed high-risk.

Although much has been learned about CBF in preterm neonates, many questions remain including how variations in CBF are related to neurologic pathologies, and the role of various CBF monitoring techniques, particularly in the setting of disturbed hemodynamics such as hypotension, PDA, anemia, and changes in respiratory status. In all neonates, immediately following birth there is an initial increase in CBF followed by a slight downward trend until around 72 h of life. This rise in CBF after birth is more pronounced in preterm infants (102, 103). Immaturity of the autonomic nervous system, especially the parasympathetic control of CBF is particularly underdeveloped in preterm infants and is the presumed cause of these differences between term and preterm infants (104–106). Loss of this autonomic regulation dampens the adaptive cerebral reactivity, making preterm infants susceptible to passive cerebral circulation and putting them at risk for low CBF in the setting of hypotension (107, 108). When accompanied by PDA, this further compromises CBF and the ability to maintain intact CA (109, 110).

Respiratory status changes also affect CBF more acutely in the preterm neonate as they are prone to rapid fluctuations in P_aCO₂ after birth due to respiratory distress syndrome and surfactant administration. In addition, fluctuations in MAP acutely impact cerebrovascular responsiveness to P_aCO₂ (111). Together, these circumstances put preterm infants at increased risk of impaired CA. Dysfunction of CA has been proposed as the mechanism for preterm brain injuries such as IVH and leukomalacia (112–115). The correlation between NIRS metrics of intact vs. impaired CA and subsequent IVH was recently demonstrated in a cohort of preterm infants studied in the first 4 days of life (116). Infants who developed IVH or died showed a loss of cerebral reactivity with fluctuation in blood pressures, resulting in both hypoperfusion and reperfusion brain injury (116).

Several studies to date have asked the question of whether routine NIRS monitoring in preterm neonates improves outcomes. There is no consensus as researchers from several centers have reported opposing results (117–120). Comparisons between such studies are difficult, however, and differences are likely due to contrasting approaches in the management of preterm neonates based on S_cO₂ measures. Initial results in the SafeBoosC II randomized controlled trial of NIRS in neonates born extremely preterm did not show an impact on EEG outcomes (121) or neurodevelopmental outcome at age 2 years (119), but an additional Phase III randomized-controlled trial is underway (122). A practical guide to the use of NIRS in neonates was recently published by Vesoulis et al. and the authors conclude that there is a need for large studies of comparative approaches (123), which the SafeBoosC III trial will be a strong contributor. These studies must occur while equipoise still exists.

There is a sense of urgency to improve our understanding of CA in neonates with HIE as their fluctuations in CBF and maintenance of adequate CPP have been shown to correlate with secondary injury (148–154). Avoidance of post-hypoxic reperfusion injury is a key area of concern in these infants (155). Doppler assessment of cerebral RI with measurements < 0.55 indicate loss of CA and correlate with worse outcomes (156, 157). However, in the setting of active therapeutic hypothermia, RI is somewhat less predictive (positive predictive outcome 60%, negative predictive outcome, NPV 78%) (158), but in late cooling RI < 0.55 has a NPV of 86%, which increases to 89% after rewarming (159).

Many studies underway are using NIRS coupled with invasive MAP monitoring to determine the upper limits of CA in asphyxiated neonates in order to develop individualized optimal MAP parameters and improve long-term prognostication (151, 160). With adjuvant therapies on the horizon, this will enhance the impact of therapeutic hypothermia (or replace its use in those patients in whom it is contraindicated) (161–165). Wavelet transform coherence analysis is being used to differentiate those neonates labeled as mild HIE who may benefit from therapeutic hypothermia based on their inability to maintain intact CA (Figure 2). Chalak et al. has further developed the wavelet coherence analysis to quantify neurovascular coupling by analyzing the relation between regional changes in aEEG signals and MAP. This analysis may provide new avenues for classification of HIE injury that will promote a more targeted therapeutic approach, particularly due to the heterogeneity of the HIE diagnosis. Currently, these techniques are limited by the time and computationally intensive analysis required. Therefore, further development of these bedside time-synchronized hemodynamic monitoring tools to include real-time analytics for neonates with HIE is of utmost importance.

Hemoglobin volume phase index, which relies on an estimated relative total hemoglobin from NIRS along with continuous MAP to assess CA, was recently used in a cohort of neonates with HIE. Those with higher hemoglobin volume phase values at 24 h of life (indicating impaired CA) had a higher rate of death or severe brain injury by MRI (166). The author's proposed cutoff of hemoglobin volume phase index had a sensitivity of 86% and specificity of 74% for death or severe brain injury by MRI. This type of early prognostication tools may facilitate the use of adjuvant therapies in specific HIE sub-populations and/or allow for improved decision-making about goal-directed care at a much earlier timepoint than standard treatments currently allow. Next steps should include larger scale multicenter prospective clinical trials to determine the generalizability of this metric.

CBF monitoring of the fetus will be improved significantly through advances in MRI technologies. Currently, the research scans that show the most promise for fetal and placental perfusion are limited to specific institutions that have the expertise required for these customized sequences and their interpretation. Image data sharing between institutions and countries is a major barrier. Commercialization of these MRI scanning sequences would improve their accessibility and data sharing capabilities. The development of artificial intelligence for automated region of interest assignment will be key to improve the pipeline of discovery as will user-friendly interfaces for analysis. These technologies will not replace the inexpensive and readily-available method of Doppler ultrasound for CBF monitoring anytime soon, thus studies to maximize our knowledge and utilization of this tool are also warranted. The CEPRA study (Dutch Trial Registry NTR trial NL7557) is an example of how the fetal Doppler CBF monitoring field is moving forward, the goal of which is to determine the utility of CPR as an indicator for immediate delivery in pregnant people reporting loss of fetal movement (167). These types of multicenter studies are essential to have an impact on clinical practice of fetal monitoring technologies that have already been studied for many years. In contrast, studies of CBF using fetal MRI are in their infancy and require robust sharing of data and technical expertise between centers to move the field forward.

For postnatal CBF monitoring, NIRS dominates most recent research and will likely continue to be a favored technology in the neonate. However, in order to move forward with the widespread use of NIRS and its real-time analysis with other hemodynamic monitoring signals to estimate CA, significant investment is required to perform multicenter randomized controlled trials validating its ability to improve outcomes in specific neonatal populations. In neonates with CHD, equipoise is likely lost for clinical studies of NIRS since it is already widely used in these patients in cardiac intensive care units and its benefit in that population has been demonstrated in small studies. Clinical trials of NIRS and time-synchronized analytical tools are being studied in other populations, however, such as neonates with HIE (64, 65, 168) and those born preterm (121, 122, 169).

In addition, the use of NIRS to estimate the upper and lower limits of CA in vulnerable neonatal populations requires additional study and a focus on a precision-medicine approach, which is underway in ongoing clinical trials (170). The optimal MAP to maintain stable CBF are unique to individuals and highly related to their specific pathology and their hemodynamic adaptations (160). In the fetal and preterm neonatal populations the ability to autoregulate CBF is not yet fully mature. For those with CHD, the limits of CA are highly related to the degree of somatic and cerebral hypoxia that drives minute to minute changes in CBF. In each of these cases, uniform thresholds of what is considered adequate MAP are not able to provide the type of precision medicine that is needed to improve patient outcomes. Instead, new tools to analyze and integrate the multitude of data points that clinicians are tasked with interpreting are essential to allow a precision-medicine approach and move the field forward.

Technologies that integrate multiple monitoring modalities into a time synchronized visual representation will allow a new level of interpretation and improve our individualized approach to care. Specifically, in terms of CA, an accurate assessment of upper and lower limits of CA must be made for each patient in order to precisely titrate medications and interventions. Next-level monitoring tools are required to facilitate the rapid recognition of impending hemodynamic instability. Likewise, these predictive analytics will be useful to encourage faster weaning from vasoactive medications, mechanical ventilators, and other treatments. As with all technologic advances in patient monitoring, the question of whether these analytics of the future make a positive impact on patient care must be asked and carefully investigated.