Open Access
Add to BibSonomy
Abstract
Carotid endarterectomy (CEA) involves removal of plaque in the carotid artery to reduce the risk of stroke and improve cerebral perfusion. This study aimed to investigate the utility of assessing pulsatile blood volume and flow during CEA. Using a combined near-infrared spectroscopy/diffuse correlation spectroscopy instrument, pulsatile hemodynamics were assessed in 12 patients undergoing CEA. Alterations to pulsatile amplitude, pulse transit time, and beat morphology were observed in measurements ipsilateral to the surgical side. The additional information provided through analysis of pulsatile hemodynamic signals has the potential to enable the discovery of non-invasive biomarkers related to cortical perfusion.
© 2024 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Clinical management of cerebral blood flow (CBF) during CEA is typically performed by targeting the value of mean arterial blood pressure or cerebral perfusion pressure (CPP) that allows for adequate perfusion for the brain [1–5]. This approach focuses on the average perfusion received by the cerebral tissue, although recently, a focus not only on the average cerebral perfusion, but also the cerebral hemodynamic pulsatility has been investigated as an additional important factor in brain perfusion [6]. Pulsatile flow is a continuous forward flow caused by the elastic recoil of blood vessel walls. Alteration in the mechanical properties of vessel walls as a consequence of age (e.g. vascular stiffening [7]) or pathology (e.g., atherosclerosis) can lead to changes in the degree of pulsatility that is transferred to the microvasculature. Increases in amplitude of the pulsatile blood flow transferred to the cerebral microvasculature have been found to contribute to white matter damage and cognitive decline [8–12]. On the other hand, decreases in pulsatility, such as those seen in continuous flow cardiopulmonary bypass, are found to contribute to impaired microvascular flow and a decoupling of cerebral blood flow and regional brain oxygenation [13,14]. Continuous monitoring and management of pulsatile features of blood flow and blood volume [15] could provide complementary insights into the efficacy of cerebral and systemic microvascular blood flow and clinical perfusion management strategies.
To test the hypothesis that continuous monitoring of the features of pulsatile cerebral hemodynamics could provide a reliable marker of the quality of microvascular circulation, we analyzed the pulsatile hemodynamic features captured in noninvasive, diffuse optical data sets collected during carotid endarterectomy (CEA). CEA is a prophylactic surgery where built-up atherosclerotic plaque at the carotid bifurcation is removed to reduce the risk of ischemic stroke and restore normal cerebral blood flow [16]. During CEA, the three branches of the carotid artery are cross-clamped, reducing blood flow to the ipsilateral cerebral hemisphere. Blood flow is delivered to the cerebral hemisphere ipsilateral to the clamping from the contralateral circulation and the vertebral arteries through the circle of Willis [17,18], though a fully anatomically complete circle of Willis is present in only approximately 20% of the population [19]. While a fully complete circle of Willis is not necessary to provide adequate cerebral perfusion, insufficient collateral circulation can put patients at risk for ischemic stroke due to hypoperfusion [20]. In the most severe cases of flow insufficiency, a shunt can be placed in the ipsilateral carotid artery to satisfy cerebral oxygen requirements. Intraoperative strategies for assessing the quality of collateral circulation and the appropriateness of the use of a shunt include electroencephalography (EEG), carotid stump pressure (SP) measurement, and transcranial Doppler ultrasound (TCD) [21]. EEG monitors for cortical electrical dysfunction that comes as a consequence of hypoperfusion and ischemia [22]. SP invasively estimates cerebral perfusion by measuring the pressure waveforms in the internal carotid artery after clamping [23,24]. TCD is used to measure changes in the blood flow of the ipsilateral middle cerebral artery for the detection of cerebral hypoperfusion during CEA [25]. Each of these methods provides complementary information about cerebral blood flow and neurovascular coupling, though none are both easy to use and directly assess the microvascular cerebral blood flow and blood volume.
Diffuse optical modalities, such as near-infrared spectroscopy (NIRS) and diffuse correlation spectroscopy (DCS), are noninvasive techniques that allow for the measurement of microvascular blood volume and blood flow and can resolve their pulsatile components up to a depth of approximately 1-2 cm under the surface of the skin [26]. NIRS measures the attenuation of light at multiple wavelengths to assess oxy- (HbO) and deoxy- (HbR) hemoglobin changes, and their sum provides total (HbT) hemoglobin concentration changes which is proportional to cerebral blood volume (CBV) [27–29]. DCS uses long coherence-length light to evaluate the motion of scattering particles in tissue, which are primarily red blood cells. By computing the autocorrelation of the fluctuating speckle intensity and fitting the result with an appropriate model, blood flow in microvasculature is measured in terms of the DCS blood flow index (BFi) [30]. To maintain consistent nomenclature between reported NIRS and DCS data, we refer to the NIRS pulsatile signal in terms of a pulsatile blood volume index (pBVi), which is linearly proportional to the pulsatile total hemoglobin concentration (HbT) signal or the pulsatile optical density signal (ΔOD). While traditional implementations of DCS measurements exhibit lower signal-to-noise ratio (SNR) than NIRS for a given source-detector separation at the sampling rates needed to resolve pulsatile hemodynamics, recent advances in DCS technology with enhanced SNR as well as pulse averaging techniques have enabled devices to measure pulsatile BFi (pBFi) at high sampling rates [31–37].
In this pilot study, we used NIRS and DCS to noninvasively assess changes in pBVi, commonly called photoplethysmography (NIRS-PPG) [38], and pBFi of cerebral microvasculature as a consequence of unilateral carotid clamping during CEA. Specifically, we show the feasibility of detecting bilateral pBFi and pBVi in these patients. We characterized shape and amplitude of the pulsatile components and compared them across ipsilateral and contralateral hemispheres and different surgical periods. We also compared between shunted and non-shunted patients to assess the utility of pBFi and pBVi as potential biomarkers indicating the need for shunt placement.
2. Methods
2.1 Study subjects and procedure
A subset of patients previously reported was used for this pulsatile analysis [39]. From the original 24 patients, we included twelve patients (#12 to #23) in whom combined bilateral NIRS-DCS measurements were performed. Potential subjects were approached during their preoperative visit. The study was explained and all questions answered by a study physician prior to the day of surgery. Patients who decided to participate gave written informed consent. The optical probes were secured symmetrically to the patients’ left and right upper forehead before the surgery. This study was reviewed and approved by the Massachusetts General Brigham Institutional Research Board (IRB: #2015P002669). Patient demographics are described in Table. 1
2.2 Instrumentation
As previously described [39], a combined frequency domain NIRS-DCS instrument (MetaOx, ISS Inc. Champaign, IL) was used to monitor hemoglobin concentration and blood flow bilaterally. Each bilateral optical probe had colocalized source fibers for four NIRS wavelengths and one DCS wavelength. In each probe, one DCS single-mode, detector fiber was placed at a source-detector separation of 5 mm and three at 25 mm, and two NIRS fiber bundles were placed at 5 mm and 30 mm from the source (Fig. 1). The NIRS sources were split by wavelength: 672, 706, 759, and 830 nm on the right probe and 690, 726, 784, and 813 nm on the left probe. This combination of wavelengths and source-detector separations allowed for bilateral quantification of oxy- and deoxy- hemoglobin concentration changes and BFi in the scalp (short source-detector separation of 5 mm) and in the scalp + brain (large source-detector separation of 25 mm for DCS and 30 mm for NIRS). The difference between NIRS and DCS long separation distances was dictated by SNR limitations of DCS and by better depth sensitivity of DCS [40].
Fig. 1. a) Optical probe used in this study (enlarged). Short separation (blue square) has colocalized outputs to DCS and NIRS. b) Image of both optical probes on a subject’s forehead along with EEG probes (standard institutional monitoring).
NIRS lasers emitted <5 mW and were temporally multiplexed to cycle all eight lasers in 100 ms. Two long coherence length lasers, one for each side, continuously emitted 35 mW at 850 nm for the DCS measurements and were always on during data acquisition. The sampling rate was set to 10 Hz for both modalities. This application of the optical powers ensured the total light exposure remained below the maximal permissible exposure safety limits for skin [41]. Crosstalk between the two modalities was eliminated using optical short-pass filters in front of the NIRS detectors [42].
Clinically measured radial artery blood pressure (ABP) was collected in real time via MetaOx auxiliary channels, and patient characteristics, including preoperative Doppler ultrasound results, bilateral stenosis levels, and post-surgery outcomes, were obtained via the Epic electronic medical record system (Epic Systems, USA) after the procedure.
2.3 Optical data processing
As described previously [39], for both large and short source-detector separations, DCS BFi was obtained by fitting the normalized temporal intensity autocorrelation function (g2(τ)), with fixed absorption and reduced scattering coefficients (µa = 0.2 cm-1, µs’ = 8.0 cm-1 at 850 nm). For each NIRS wavelength and each source-detector pair, the detected intensity was converted to change in optical density relative to baseline (ΔOD) following the assumptions reported previously [39]. Changes in absorption were estimated using the modified Beer-Lambert law [28]. Finally, we obtained changes in HbO, HbR and HbT from multi-wavelength absorption changes.
2.4 Flow and volume pulsatility analysis
To refine the pulsatile hemodynamic signals, we employed a previously reported cardiac gated averaging technique to improve the SNR of pulsatile HbT and BFi signals [36]. Systolic peaks were identified in ABP time traces to identify heartbeat intervals for averaging. Owing to natural variations in heartbeat frequency, intervals between identified systolic peaks were not of equal length. To allow for averaging, each signal, (i.e. individual beats), in time intervals between peaks were interpolated to have the same number of samples (400 samples in each normalized beat). For pBVi signals, peak-to-peak intervals were extracted directly from the 10 Hz sampled signal, resampled into the common time base, and averaged. For pBFi signals, g2(τ) curves from the peak-to-peak intervals were identified and averaged in-sync with the other g2(τ) curves coming from the same position in the heartbeat. Following averaging, g2(τ) curves were fit for pBFi. The difference in approach for the calculation of the pBVi and pBFi signals was due to the reduced SNR of DCS signals, and the need for higher SNR of g2(τ) curves to accurately fit the BFi. For both modalities, a 30 s sliding average window with 50% overlap was used to compute pulsatile hemodynamic parameters. For this window duration, the minimum number of averaged beats across all subjects and all time points was 20 and the maximum was 40. From the resulting averaged signals, pulsatile amplitude and the time delay between the peak of arterial pulsation and peak of the pBFi and pBVi waveforms for each measurement side and source-detector separation were calculated. Pulsatile amplitude was computed as the maximum value of averaged signal minus the minimum value of the averaged pulse. Due to the unknown difference in pulse travel distance between the optical measurements on the forehead and the arterial blood pressure measurement in the radial artery, the absolute time delay measured between these signals are difficult to interpret, though, because of the common reference (i.e. ABP), delays between optical signals, as well as changes in delays through different surgical periods can be analyzed quantitatively. Time delay between bilateral measurements is representative of the difference in pulse transit time to the measured vascular bed. In all subjects, with the sliding window, we calculated these parameters at 15 s temporal resolution and provided their average values during the four possible surgical phases: (1) baseline period before clamping(30 min ending 5 min before the start of carotid clamping), (2) if shunting was required, clamped, pre-shunt period (immediately after clamping until placement of the shunt), (3) regardless of shunt placement status, clamped period (30 min during clamping starting 5 min after carotid cross-clamping), and (4) recovery period (30 min starting 5 min after clamp release).
2.5 Ratio metrics computed from the pulsatile hemodynamic signals
Assessment of pulsatile features using other techniques, including TCD and MRI, have been performed not only with raw values of pulsatile amplitudes, but also through normalized ratios [12,43,44]. Two commonly computed ratios are the pulsatility index (PI) [45], defined as the amplitude of the pulsation divided by the average value across the cardiac cycle, and the resistive index (RI) [46], defined as the amplitude of the pulsation divided by the maximum systolic value reached. Additionally, using the pulsatile blood flow signals, we compute an index of cerebrovascular resistance (pCVRi), calculated as the amplitude of the arterial blood pressure pulsation divided by the amplitude of the blood flow pulsation, and an index of cerebrovascular compliance (pCVCi), calculated as the amplitude of the amplitude of the blood volume pulsation divided by the amplitude of the arterial blood pressure pulsation [43]. Comparisons of the ratio metrics are performed for the long-source detector separation measurements between the baseline period and the recovery period.
2.6 Statistical analysis
We compared the relative changes in the pulsatile hemodynamic parameters (rpBFi and rpBVi) at each separation (short and long) for the bilateral measurements. Comparisons are made between the baseline, pre-clamp stage to the during clamp stage and the baseline, pre-clamp stage to the post clamp stage using paired t-tests. Normalized hemodynamic parameters were calculated as the average value during the surgical phase divided by the average value during the baseline period. We applied the Bonferroni correction to the level of significance that considers the number of comparisons performed (32) and the final level of significance for the two tails, paired t-test was α = 0.0016 [47].
3. Results
Figure 2 shows the distribution of amplitude changes of pBVi and pBFi both ipsilateral and contralateral to clamping for both the short and long source-detector separations. Each box and whisker plot represents the distribution of changes to the amplitude of the pBVi and pBFi signals during each of the surgical phases, with individual patient changes labeled with lines with colors corresponding to whether or not a shunt was used. Patients receiving a shunt and those without a shunt are plotted separately to interrogate possible differences between patient populations. On average, a > 40% reduction in the pulsatile amplitude was seen on the side ipsilateral to the clamping in both pBVi and pBFi both at short and large source-detector separations (Table 2). Restoration of pulsatile amplitude after the clamping was seen in most patients across all ipsilateral signals. Changes in the amplitudes on the contralateral side were largely variable, although there was a trend toward increases in pBFi and constant or decreases in pBVi during the clamping period.
Fig. 2. Box plots showing the changes in pulsatile BFi (left) and BVi (right) amplitude for measurements ipsilateral (top) and contralateral (bottom) to the surgical side for both long (top of each quadrant) and short (bottom of each quadrant) source-detector separations during different surgical phases. Individual patient changes are displayed with lines, with shunted patients in green and non-shunted patients in orange.
Table 1. Patient demographics.
Table 2. Summary results for average pulsatile hemodynamic changes.
We also compared the shapes of pBFi and pBVi waveforms. Figures 3 and 4 show the average bilateral pulsatile signals during the surgical periods at the long source detector separation for an exemplary non-shunted patient (Fig. 3, #18 [39]) and for a shunted patient (Fig. 4, #20 [39]). Error bars represent the standard deviations over the averaged period. The pulsatile signals are time shifted for better visibility such that the diastolic minimum pBFi value occurs at t = ∼0.1. The relative timing between pBFi and pBVi signals are maintained. In both patients, on the contralateral side, minimal changes to the shape of the pulsatile curves were seen, indicating both timing and amplitude of pBVi and pBFi were unaffected by the changes caused by clamping. Conversely, the side ipsilateral to clamping had changes in shape, amplitude, and timing for both pBFi and pBVi measurements during the surgical phases in both patients. For both shunted and non-shunted patients, there was an observed decrease in the amplitude (Fig. 2) of pBFi and pBVi during the clamping phase relative to the baseline. Reduction in amplitude of pBFi in shunted patients was comparable to reductions seen in non-shunted patients, though the amplitude of pBVi signals in shunted patients were reduced more than what was observed for the non-shunted patients. Reported shapes of pulsatile signal from these two exemplary subjects were representative of the population of pulsatile shapes measured in both shunted and non-shunted patients.
Fig. 3. Large separation pulsatile waveforms of blood flow (top) and blood volume (bottom) on the ipsilateral (left) and contralateral (right) sides of a non-shunted patient (#18). During the clamping period, the side ipsilateral to the clamp had a reduced pulsatile signal amplitude in both NIRS and DCS measurements, and the peak of the waveform was delayed relative to the pre-clamp period. Upon the release of the clamp, in this patient, the blood flow amplitude and peak time on the side ipsilateral to the clamp returned to baseline, although with a slightly modified shape. Changes during the recovery period were variable between subjects.
Fig. 4. Large separation pulsatile waveforms of blood flow (top) and blood volume (bottom) on the ipsilateral (left) and contralateral (right) sides of a patient (#20) who required carotid shunting. For the NIRS measurement ipsilateral to clamping, before shunt placement, amplitude of the pulsatile signals is reduced more than is seen in patients not requiring shunt placement. Upon placement of the shunt, the pBVi signal amplitude returned to levels that are observed in pBVi amplitude of patients not requiring a shunt, though the pBFi increase was seen to reach pre-clamp level.
Peak (highest value in each pulse) arrival time relative to the pre-clamping period was also compared across all pBVi and pBFi measurements. Figure 5 shows the distribution of differences in peak arrival times for all hemodynamic signals in the analyzed subset of patients. The later arrival of the pulsatile waveform observed on the ipsilateral side during clamping may be due to the extra length in the vasculature that is traveled to reach that hemisphere via collateral circulation. In shunted patients, the shift in arrival time caused by clamping was reduced following shunt placement.
Fig. 5. Box plots showing the changes in peak arrival time relative to the pre-clamp period for pulsatile BFi (left) and BVi (right) amplitude ipsilateral (top) and contralateral (bottom) to clamping for both long (top of each quadrant) and short (bottom of each quadrant) source-detector separations during different surgical phases. For all signals ipsilateral to the clamp, we observed an increase in the delay of arrival, consistent with an increase in travel time through collateral circulation. Signals contralateral to the clamped side have considerable variability.
We summarize results presented in Figs. 2–5 in Table 2, detailing the average ratio of pulsatile amplitude across all non-shunted patients for both ipsilateral and contralateral hemispheres, average time difference between pulse arrival times in the ipsilateral and contralateral hemispheres, and changes in heart rate and blood pressure between the different surgical phases. The measured pulsatile amplitudes of ipsilateral NIRS and DCS signals decreased by approximately two-fold during clamping, with a slightly more severe drop observed in the DCS measurements. On average, contralateral signals were observed to increase in amplitude, likely due to clinically induced hypertension using phenylephrine during the surgical period to help perfuse brain ipsilateral to the carotid clamp [48]. On average, an increase in the pulsatile amplitude for both pBVi and pBFi was seen in the long separation measurements after the removal of the clamp; however, as was seen in Fig. 2, there was a greater degree of variability between patients. The pulse delay of ipsilateral signals before clamping relative to that during clamping showed the expected delay, consistent with an increased distance for the pressure wave to travel. Contralateral delays remained unchanged on average, and ipsilateral delays returned to baseline after unclamping. During clamping, heart rate was slightly elevated on average and returned to baseline after unclamping.
Comparisons of the changes in PI and RI measured in the baseline period and recovery period are shown in Figs. 6 and 7, respectively. Though the distributions of the baseline and recovery periods are overlapping, in general, for measurements ipsilateral to the clamped side, increases in the median values of pulsatility index and resistive index are observed for both NIRS and DCS measurements. In the measurements contralateral to the clamped side, changes in the distributions are less consistent, and the median value of the parameter remains roughly the same. For the NIRS measurements, because of the small amplitude of the pulsation relative to the assumed baseline total hemoglobin concentration (80 µM [39]), both the PI and RI are lower in magnitude as compared to the DCS derived metrics, and are nearly equal, as the maximum systolic value is nearly equal to the mean value.
Fig. 6. Comparison of the distribution of the pulsatility index for (A.) DCS long source-detector separation measurements and (B.) NIRS long source-detector separation measurements. Values for DCS PI are within the range of previously reported normative values of PI [37], with slight elevations observed on the side ipsilateral to clamping. Ipsilateral signals for both DCS and NIRS are seen to, on average, increase following the surgical procedure, while the response is more variable in the contralateral signals.
Fig. 7. Comparison of the distribution of the resistive index for (A.) DCS long source-detector separation measurements and (B.) NIRS long source-detector separation measurements. The NIRS results presented in (B.) are seen to be very similar to the results presented in Fig. 6.B, owing to the relatively small pulsation amplitude. As was the case for the PI, following the surgical procedure, the values of RI ipsilateral to the clamping were seen to increase on average while the values contralateral had a more varied response.
In addition to the metrics of PI and RI, using the invasively measured blood pressure signal, estimates of cerebrovascular resistance (pCVRi) and cerebrovascular compliance (pCVCi) can be computed. In Fig. 8, comparisons of pCVRi (Fig. 8.A) and pCVCi (Fig. 8.B) are shown for the baseline and recovery period for the long source-detector separation measurements. A reduction in pCVRi is seen for both ipsilateral (median reduction of 22%) and contralateral (median reduction of 28%) measurements. The response for pCVCi is mixed for the ipsilateral and contralateral signals. A ∼40% increase in the median value was increase was observed for the ipsilateral signal, while the median value of the contralateral signal remained constant, though the distribution of values trended toward an increase.
Fig. 8. Comparison of the distribution of (A.) pCVRi for the long source-detector separation measurements and (B.) pCVCi for the long source-detector separation measurements. On average, decreases in the pCVRI are observed for both ipsilateral and contralateral measurements, while on average modest increases in pCVCi are observed for both ipsilateral and contralateral measurements.
4. Discussion and conclusion
NIRS and DCS can be combined and co-localized at the tissue in the same spatial location. Although the sensitivity to the cortical signal is lower for NIRS than for DCS [40], the co-localized probe implies that the same underlying brain area is probed for BFi and HbT measurements yielding more information regarding the characteristics of hemodynamic pulsatility for the sampled tissue volume. Use of low-profile, easy to position optical probes allows for non-invasive, continuous monitoring during surgery without the need of an expert operator.
In this study, we successfully extracted pulsatile waveforms of cerebral blood volume and blood flow in patients undergoing CEA. Observed changes in hemodynamic pulsatility trended similarly to the transient changes at the time of clamping in the steady-state CBF and HbT results presented previously [39]. We report the steady-state response in CBFi and HbT for this subset of patients in the supplemental information (Table S1 and Fig. S1). While the size of the study is a limitation for the comparison of the shunted and non-shunted patient groups, shunted patients displayed the two most extreme reductions in long separation pBVi before shunt placement, possibly pointing to an effective biomarker for the need for shunt placement. Sensitivity to brain perfusion at the microvascular level offered by diffuse optical techniques could make NIRS and DCS helpful in assessing the need for selective shunting during carotid clamping, although a more extensive study is required to confirm the sensitivity and specificity of these signal changes. We observed that shunt placement effectively restored blood flow pulsatility in the long separation measurement to pre-clamp levels, though did not do so for either the short separation measurements or for the long separation, blood volume measurement. Due to the small sample size and small number of adverse events [39], the significance of the observed difference is unclear, though could be investigated in a larger population. Estimations of PI and RI from the long source-detector separation DCS measurements yielded results that fall reasonably within the range of normative values presented previously [37], though deviations are observed. Further investigation of the pulsatile signals across the lifespan as well as in other pathologies could help frame the results presented here in the context of vascular aging as well as vascular pathology [7,49]. From the estimates of pCVRi and pCVCi, we observe changes following the procedure in measurements both ipsilateral and contralateral to the clamping side that are indicative of a greater transfer of blood flow and volume to the cerebral vasculature. We do not necessarily expect that the microvascular properties are altered by the surgical procedure, and the observed results may reflect an increase in the amplitude of blood pressure that is allowed to pass through the less restricted carotid artery following the surgery. In addition to quantifying the amplitude of pulsatility and the ratio metrics, we also showed the ability to track the delay of the pulsatile signal relative to arterial blood pressure, and the changes caused by carotid clamping. A clear shift in the pulse peak time was observed for all signals ipsilateral to the clamped side, though differentiation between shunted and non-shunted patients is less clear in the timing signals.
Using multi-beat averaging, we were able to overcome the limitation of 0.1 s resolution (10 Hz acquisition) to yield finer detail, as seen in the Figs. 3 and 4, though a device with a higher sampling rate would improve the visibility of informative pulsatile features [49] and improve the precision of the peak delay estimations. Relative delays between the NIRS and DCS signals, as can be seen by comparing the same side NIRS and DCS measurements in Figs. 3 and 4, could also be an interesting feature to examine, as has been done previously for absorbance and speckle signals [50]. Somewhat surprisingly, no correlation was found between the reduction in hemodynamic pulsatility during clamping and the level of stenosis in the contralateral side [39]. This may suggest that vessels that anastomose with the circle of Willis or the contralateral carotid artery itself compensate for increased stenosis.
Estimation of pulse amplitude could also be affected by the reduced sampling rate. For NIRS measurements used in this work, this is relatively unlikely, as most of the NIRS pulsatile signal is comprised of the fundamental frequency with smaller contributions by harmonic frequencies, for which the 10 Hz sampling rate is sufficient to fully capture the signal. For DCS, the shape of the pulsatile signal is more complex, and higher harmonics of the fundamental frequency that could be smoothed by the 0.1s integration period of the DCS measurement contribute more to the shape of the pulse. Across all subjects, 95% of the individual heartbeat fundamental frequencies fell between 46.2 and 75.9 beats per minute with a mean of 57.9 beats per minute. Using the frequency content of a pulsatile blood flow signal measured at high temporal resolution [35], we can estimate the effect of averaging on the measured amplitude. Based on the range of frequencies measured and the frequency content present in the DCS signals, we expected an attenuation of the amplitude between 2% and 8.9%. This could become problematic for amplitude estimation, because the changing heart rate alters the extracted amplitude. However, as presented in Table 2, the relative change in heart rate between surgical conditions was minimal, and it is likely that the range of average heart rates across all subjects contributed to the spread. As with timing features, we anticipate that the use of a faster sampling device will allow more accurate measurements of hemodynamic pulsatility.
This study has shown the potential of NIRS-DCS for monitoring the quality of cerebral perfusion via deep tissue pulsatility measures. With both increases and reductions in hemodynamic pulsatility implicated in the progression of cerebrovascular disease [6,51], use of the combined NIRS-DCS approach provides a potential noninvasive method to assess the risk of complications associated with excessive or insufficient pulsatility [52,53]. Furthermore, as NIRS-DCS investigates pulsatility in the microvasculature, pulsatile perfusion at the level of oxygen exchange is interrogated, and a complementary set of features may be visible that are not found in measurements of conduit vasculature. This is significant because several recent studies have suggested a role of microvasculature in cognitive decline and aging [54,55].
We hope to apply this analysis to data collected at faster sampling rates to extract more information from the shape of the pulsatile signal, as has been done in previous studies [34]. Specifically, increasing temporal resolution could reveal the dicrotic notch (features including time after peak, amplitude relative to peak, presence/absence) that is seen in invasive blood pressure measurements, elucidating finer details concerning pulsatility, potentially providing insight into heart health (vasodilation and aortic valve function) [56]. Several NIRS devices with high sampling rates have been demonstrated [57,58], and recent advances in DCS techniques, including use of heterodyne detection [31,32,59], time-resolved detection [60–62], and shifts in the wavelength used [63], provide methods to perform fast DCS measurements while maintaining sufficient SNR at the long source-detector separations required to be sensitive to cerebral blood flow.
Funding
Eunice Kennedy Shriver National Institute of Child Health and Human Development (R01HD091067); National Institute of General Medical Sciences (R01GM116177).
Acknowledgments
We would like to thank Drs. Mirela Simon, Glen LaMuraglia, Alyssa Martin, and the neurovascular OR team for assistance with measurements, and Dr. John Sunwoo for stimulating discussion.
Disclosures
At the time of the study, MAF had a financial interest in 149 Medical, Inc., a company developing DCS technology for assessing and monitoring CBF in newborn infants, which is now dissolved. MAF’s interests were reviewed and managed by Massachusetts General Hospital and Mass General Brigham in accordance with their conflict-of-interest policies. A.I.Z., K.K., K.C.W., E.T.P., and M.B.R. have nothing to report.
Data availability
The data that support the findings of this study are available from the corresponding authors upon reasonable request.
Supplemental document
See Supplement 1 for supporting content.
References
1. J. Biller, W. M. Feinberg, J. E. Castaldo, et al., “Guidelines for carotid endarterectomy,” Circulation 97(5), 501–509 (1998). [CrossRef]
2. K. H. Kim, C. H. Lee, Y. J. Son, et al., “Post-carotid endarterectomy cerebral hyperperfusion syndrome: is it preventable by strict blood pressure control?” J Korean Neurosurg Soc 54(3), 159 (2013). [CrossRef]
3. M. Smith, “Cerebral perfusion pressure,” BJA: British Journal of Anaesthesia 115(4), 488–490 (2015). [CrossRef]
4. M. van der Jagt, “Fluid management of the neurological patient: a concise review,” Critical Care 20(1), 126 (2016). [CrossRef]
5. F. S. Taccone, G. Citerio, P. Le Roux, et al., “Advanced monitoring of systemic hemodynamics in critically ill patients with acute brain injury,” Neurocrit Care 21(S2), 38–63 (2014). [CrossRef]
6. Y. Shi, M. J. Thrippleton, G. W. Blair, et al., “Small vessel disease is associated with altered cerebrovascular pulsatility but not resting cerebral blood flow,” J. Cereb. Blood Flow Metab. 40(1), 85–99 (2020). [CrossRef]
7. X. Xu, B. Wang, C. Ren, et al., “Age-related Impairment of Vascular Structure and Functions,” Aging Dis 8(5), 590 (2017). [CrossRef]
8. T. Vikner, A. Eklund, N. Karalija, et al., “Cerebral arterial pulsatility is linked to hippocampal microvascular function and episodic memory in healthy older adults,” J Cereb Blood Flow Metab 41(7), 1778–1790 (2021). [CrossRef]
9. A. Scuteri, P. M. Nilsson, C. Tzourio, et al., “Microvascular brain damage with aging and hypertension: Pathophysiological consideration and clinical implications,” J Hypertens 29(8), 1469–1477 (2011). [CrossRef]
10. R. S. Vasan, M. I. Short, T. J. Niiranen, et al., “Interrelations between arterial stiffness, target organ damage, and cardiovascular disease outcomes,” J Am Heart Assoc 8(14), 141 (2019). [CrossRef]
11. T. A. D. Jolly, G. A. Bateman, C. R. Levi, et al., “Early detection of microstructural white matter changes associated with arterial pulsatility,” Front Hum Neurosci 7, 782 (2013). [CrossRef]
12. S. H. Pahlavian, X. Wang, S. Ma, et al., “Cerebroarterial pulsatility and resistivity indices are associated with cognitive impairment and white matter hyperintensity in elderly subjects: A phase-contrast MRI study,” J. Cereb. Blood Flow Metab. 41(3), 670–683 (2021). [CrossRef]
13. C. M. Veraar, H. Rinösl, K. Kühn, et al., “Non-pulsatile blood flow is associated with enhanced cerebrovascular carbon dioxide reactivity and an attenuated relationship between cerebral blood flow and regional brain oxygenation,” Critical Care 23(1), 426 (2019). [CrossRef]
14. M. P. O’Neil, J. C. Fleming, A. Badhwar, et al., “Pulsatile versus nonpulsatile flow during cardiopulmonary bypass: microcirculatory and systemic effects,” Ann Thorac Surg 94(6), 2046–2053 (2012). [CrossRef]
15. M. E. Wagshul, P. K. Eide, and J. R. Madsen, “The pulsating brain: a review of experimental and clinical studies of intracranial pulsatility,” Fluids Barriers CNS 8(1), 5–23 (2011). [CrossRef]
16. S. J. Howell, “Carotid endarterectomy,” Br J Anaesth 99(1), 119–131 (2007). [CrossRef]
17. J. R. Romero, A. Pikula, T. N. Nguyen, et al., “Cerebral collateral circulation in carotid artery disease,” Curr Cardiol Rev 5(4), 279–288 (2009). [CrossRef]
18. Z. Vrselja, H. Brkic, S. Mrdenovic, et al., “Function of circle of Willis,” J. Cereb. Blood Flow Metab. 34(4), 578–584 (2014). [CrossRef]
19. D. L. Felten, M. K. O’Banion, and M. E. Maida, Netter’s Atlas of Neuroscience (Elsevier Health Sciences, 2016).
20. A. F. AbuRahma, A. Y. Mousa, P. A. Stone, et al., “Correlation of intraoperative collateral perfusion pressure during carotid endarterectomy and status of the contralateral carotid artery and collateral cerebral blood flow,” Ann Vasc Surg 25(6), 830–836 (2011). [CrossRef]
21. A. F. AbuRahma, A. Y. Mousa, and P. A. Stone, “Shunting during carotid endarterectomy,” J Vasc Surg 54(5), 1502–1510 (2011). [CrossRef]
22. B. Foreman and J. Claassen, “Quantitative EEG for the detection of brain ischemia,” Crit Care 16(2), 216 (2012). [CrossRef]
23. J. P. Archie, “Technique and clinical results of carotid stump back-pressure to determine selective shunting during carotid endarterectomy,” J Vasc Surg 13(2), 319–327 (1991). [CrossRef]
24. A. Kordzadeh, O. A. Abbassi, I. Prionidis, et al., “The role of carotid stump pressure in carotid endarterectomy: a systematic review and meta-analysis,” Ann Vasc Dis 13(1), 28–37 (2020). [CrossRef]
25. W. S. Yun, “Cerebral monitoring during carotid endarterectomy by transcranial Doppler ultrasonography,” Ann Surg Treat Res 92(2), 105 (2017). [CrossRef]
26. E. M. Buckley, A. B. Parthasarathy, P. E. Grant, et al., “Diffuse correlation spectroscopy for measurement of cerebral blood flow: future prospects,” Neurophotonics 1(1), 011009 (2014). [CrossRef]
27. M. Thavasothy, M. Broadhead, C. Elwell, et al., “A comparison of cerebral oxygenation as measured by the NIRO 300 and the INVOS 5100 near-infrared spectrophotometers,” Anaesthesia 57(10), 999–1006 (2002). [CrossRef]
28. D. T. Delpy, M. Cope, P. Van Der Zee, et al., “Estimation of optical pathlength through tissue from direct time of flight measurement,” Phys. Med. Biol. 33(12), 1433–1442 (1988). [CrossRef]
29. W. L. Chen, J. Wagner, N. Heugel, et al., “Functional near-infrared spectroscopy and its clinical application in the field of neuroscience: advances and future directions,” Front. Neurosci. 14, 724 (2020). [CrossRef]
30. D. A. Boas, L. E. Campbell, and A. G. Yodh, “Scattering and imaging with diffusing temporal field correlations,” Phys. Rev. Lett. 75(9), 1855–1858 (1995). [CrossRef]
31. M. B. Robinson, M. Renna, N. Ozana, et al., “Portable, high speed blood flow measurements enabled by long wavelength, interferometric diffuse correlation spectroscopy (LW-iDCS),” Sci. Rep. 13(1), 8803 (2023). [CrossRef]
32. W. Zhou, O. Kholiqov, J. Zhu, et al., “Functional interferometric diffusing wave spectroscopy of the human brain,” Sci. Adv. 7(20), eabe0150 (2021). [CrossRef]
33. N. Ozana, A. I. Zavriyev, D. Mazumder, et al., “Superconducting nanowire single-photon sensing of cerebral blood flow,” Neurophotonics 8(03), 35006 (2021). [CrossRef]
34. A. Ruesch, J. Yang, S. Schmitt, et al., “Estimating intracranial pressure using pulsatile cerebral blood flow measured with diffuse correlation spectroscopy,” Biomed. Opt. Express 11(3), 1462 (2020). [CrossRef]
35. K. C. Wu, A. Martin, M. Renna, et al., “Enhancing diffuse correlation spectroscopy pulsatile cerebral blood flow signal with near-infrared spectroscopy photoplethysmography,” Neurophotonics 10(03), 035008 (2023). [CrossRef]
36. K.-C. Wu, J. Sunwoo, F. Sheriff, et al., “Validation of diffuse correlation spectroscopy measures of critical closing pressure against transcranial Doppler ultrasound in stroke patients,” J. Biomed. Opt. 26(03), 36008–36009 (2021). [CrossRef]
37. T. M. Urner, K. R. Cowdrick, R. O. Brothers, et al., “Normative cerebral microvascular blood flow waveform morphology assessed with diffuse correlation spectroscopy,” Biomed. Opt. Express 14(7), 3635–3653 (2023). [CrossRef]
38. T. Y. Abay and P. A. Kyriacou, “Reflectance photoplethysmography as noninvasive monitoring of tissue blood perfusion,” IEEE Trans. Biomed. Eng. 62(9), 2187–2195 (2015). [CrossRef]
39. K. Kaya, A. I. Zavriyev, F. Orihuela-Espina, et al., “Intraoperative cerebral hemodynamic monitoring during carotid endarterectomy via diffuse correlation spectroscopy and near-infrared spectroscopy,” Brain Sci. 12(8), 1025 (2022). [CrossRef]
40. J. Selb, D. A. Boas, S.-T. Chan, et al., “Sensitivity of near-infrared spectroscopy and diffuse correlation spectroscopy to brain hemodynamics: simulations and experimental findings during hypercapnia,” Neurophotonics 1(1), 015005 (2014). [CrossRef]
41. American National Standard for Safe Use of Lasers, ANSI Z136.1-2007 (2007).
42. S. A. Carp, P. Farzam, N. Redes, et al., “Combined multi-distance frequency domain and diffuse correlation spectroscopy system with simultaneous data acquisition and real-time analysis,” Biomed. Opt. Express 8(9), 3993 (2017). [CrossRef]
43. M. Holmgren, A. Wåhlin, T. Dunås, et al., “Assessment of cerebral blood flow pulsatility and cerebral arterial compliance with 4D flow MRI,” J. Magn. Reson. Imaging 51(5), 1516–1525 (2020). [CrossRef]
44. J. Naqvi, K. H. Yap, G. Ahmad, et al., “Transcranial Doppler ultrasound: A review of the physical principles and major applications in critical care,” Int. J. Vasc. Med. 2013, 1–13 (2013). [CrossRef]
45. N. De Riva, K. P. Budohoski, P. Smielewski, et al., “Transcranial doppler pulsatility index: What it is and what it isn’t,” Neurocrit Care 17(1), 58–66 (2012). [CrossRef]
46. R. O. Bude and J. M. Rubin, “Relationship between the resistive index and vascular compliance and resistance,” Radiology 211(2), 411–417 (1999). [CrossRef]
47. R. A. Armstrong, “When to use the Bonferroni correction,” Ophthalmic and Physiological Optics 34(5), 502–508 (2014). [CrossRef]
48. M. D. Stoneham and J. P. Thompson, “Arterial pressure management and carotid endarterectomy,” Br J Anaesth 102(4), 442–452 (2009). [CrossRef]
49. F. A. J. Relander, A. Ruesch, J. Yang, et al., “Using near-infrared spectroscopy and a random forest regressor to estimate intracranial pressure,” Neurophotonics 9(04), 045001 (2022). [CrossRef]
50. M. Ghijsen, T. B. Rice, B. Yang, et al., “Wearable speckle plethysmography (SPG) for characterizing microvascular flow and resistance,” Biomed. Opt. Express 9(8), 3937 (2018). [CrossRef]
51. G. F. Mitchell, “Aortic stiffness, pressure and flow pulsatility, and target organ damage,” J. Appl. Physiol. 125(6), 1871–1880 (2018). [CrossRef]
52. E. L. Wu, M. Kleinheyer, and A. Ündar, “Pulsatile vs. continuous flow,” in Mechanical Circulatory and Respiratory Support (2018).
53. E. E. Van Der Wall, “Mate for the heart: pulsatile or continuous?” Netherlands Heart J. 18(2), 59 (2010). [CrossRef]
54. W. R. Brown and C. R. Thore, “Review: Cerebral microvascular pathology in aging and neurodegeneration,” Neuropathol. Appl. Neurobiol. 37(1), 56–74 (2011). [CrossRef]
55. O. De Montgolfier, A. Pinçon, P. Pouliot, et al., “High systolic blood pressure induces cerebral microvascular endothelial dysfunction, neurovascular unit damage, and cognitive decline in mice,” Hypertension 73(1), 217–228 (2019). [CrossRef]
56. M. Nirmalan and P. M. Dark, “Broader applications of arterial pressure wave form analysis,” Continuing Education in Anaesthesia, Critical Care and Pain 14(6), 285–290 (2014). [CrossRef]
57. K. C. Wu, D. Tamborini, M. Renna, et al., “Open-source FlexNIRS: A low-cost, wireless and wearable cerebral health tracker,” NeuroImage 256, 119216 (2022). [CrossRef]
58. A. Y. Rwei, W. Lu, C. Wu, et al., “A wireless, skin-interfaced biosensor for cerebral hemodynamic monitoring in pediatric care,” Proc. Natl. Acad. Sci. U. S. A. 117(50), 31674–31684 (2020). [CrossRef]
59. W. Zhou, M. Zhao, and V. J. Srinivasan, “Interferometric diffuse optics: recent advances and future outlook,” Neurophotonics 10(01), 013502 (2022). [CrossRef]
60. N. Ozana, N. Lue, M. Renna, et al., “Functional time domain diffuse correlation spectroscopy,” Front. Neurosci. 16, 1123 (2022). [CrossRef]
61. S. Samaei, P. Sawosz, M. Kacprzak, et al., “Time-domain diffuse correlation spectroscopy (TD-DCS) for noninvasive, depth-dependent blood flow quantification in human tissue in vivo,” Sci. Rep. 11(1), 1817 (2021). [CrossRef]
62. V. Parfentyeva, L. Colombo, P. Lanka, et al., “Fast time-domain diffuse correlation spectroscopy with superconducting nanowire single-photon detector: system validation and in vivo results,” Sci. Rep. 13(1), 11982 (2023). [CrossRef]
63. S. A. Carp, D. Tamborini, D. Mazumder, et al., “Diffuse correlation spectroscopy measurements of blood flow using 1064 nm light,” J. Biomed. Opt. 25(09), 97003–97004 (2020). [CrossRef]
