Real-Time Non-invasive Assessment of Cerebral Hemodynamics With Diffuse Optical Spectroscopies in a Neuro Intensive Care Unit: An Observational Case Study

Prevention of secondary damage is an important goal in the treatment of severe neurological conditions, such as major head trauma or stroke. However, there is currently a lack of non-invasive methods for monitoring cerebral physiology. Diffuse optical methods have been proposed as an inexpensive, non-invasive bedside monitor capable of providing neurophysiology information in neurocritical patients. However, the reliability of the technique to provide accurate longitudinal measurement during the clinical evolution of a patient remains largely unaddressed. Here, we report on the translation of a hybrid diffuse optical system combining frequency domain diffuse optical spectroscopy (FD-DOS) and diffuse correlation spectroscopy (DCS) for real-time monitoring of cerebral physiology in a neuro intensive care unit (neuro-ICU). More specifically, we present a case study of a patient admitted with a high-grade aneurysmal subarachnoid hemorrhage, who was monitored throughout hospitalization. We show that the neurophysiological parameters measured by diffuse optics at the bedside are consistent with the clinical evolution of the patient at all the different stages following its brain lesion. These data provide support for clinical translation of DOS/DCS as a useful biomarker of neurophysiology in the neuro-ICU, particularly in locations where other clinical resources are limited.


INTRODUCTION
One of the main goals in the care of patients with severe neurological conditions, such as a major head trauma or stroke, is to prevent secondary damage. Additional insults caused by primary damage affect and may be affected by both the systemic and cerebral physiologies (1)(2)(3)(4). To this end, close patient monitoring following acute brain lesions is essential to prevent secondary damage, and it may directly improve mortality and morbidity rates (5)(6)(7). Yet, monitoring of neurocritical patients has been limited to intermittent systemic physiological monitors and/or snapshot measurements through computed tomography (CT) or magnetic resonance imaging (MRI) (8)(9)(10)(11). Continuous cerebral monitoring is expensive, and it commonly requires invasive catheters, which limits its use to few, more severe cases, and it is seldom used in low-budget environments (8).
Among the few potential non-invasive techniques to probe cerebral physiology at the bedside, Transcranial Doppler ultrasound (TCD) is generally used to monitor cerebral blood flow velocity (CBF v ) in the large arteries (12). However, TCD is operator-dependent, not suitable for long-term monitoring, and is not sensitive to changes at the microvasculature. In addition, ∼10% of the subjects do not have appropriate acoustic windows (13,14). Electroencephalography (EEG) is another commonly used non-invasive technique to assess electrophysiological activity of the brain. However, transcranial EEG has poor spatial resolution and is very sensitive to artifacts (15).
Some studies have employed frequency-(FD-DOS) and timedomain (TD-DOS) diffuse optical spectroscopy to measure absolute tissue oxygenation. Either FD-DOS or TD-DOS can be combined with diffuse correlation spectroscopy (DCS) for measurements of CBF and metabolism. In particular, DCS-based measurements of have been previously validated against several gold standard CBF measures (37). The combination of DOS and DCS has been explored to monitor patients in several clinical scenarios, e.g., for neonatal monitoring (34,(52)(53)(54)(55), during cerebrovascular interventions (18,21,22,27,28,56), as well as for monitoring of neurocritical patients (16,17,31). However, these studies mostly focused on the effects of specific events and interventions. Concerning the practical clinical application of diffuse optics, the important question regarding the relationship between neurophysiology, as measured by DCS/DOS, and the clinical evolution of patients during hospitalization remains largely unaddressed.

CASE DESCRIPTION
We describe the case of a 62-year-old female patient who was admitted to the neuro intensive care unit (neuro-ICU) following aneurysmal subarachnoid hemorrhage (aSAH). She was an obese smoker with inconsistent adherence to her anti-hypertensive regimen. The patient was recruited during admission to the neuro-ICU of the Clinical Hospital at the University of Campinas, Brazil.
The patient was admitted after a severe headache for 3 days associated with nausea, vomiting and elevated blood pressure. She rapidly progressed to coma right after admission. The first CT scan after admission revealed a right middle cerebral artery (MCA) aSAH with Grade V on the Hunt & Hess scale and Grade III on the Fisher Scale. The TCD measurement performed on the first day of hospitalization did not show any signs of vasospasm (mean CBFv of 73 and 83 cm/s, for the right and left MCA, respectively). In the end of the first day after hospitalization the patient underwent an emergency ventriculostomy with EVD placement due to high intracranial pressure.
The CT scan obtained in the third and fourth days after admission revealed a diffuse cerebral edema with collapse of lateral ventricles, as well as an extensive ischemic area on the ipsilesional hemisphere. The TCD performed in the fourth day found mean CBFv of 64.2 and 68.3 cm/s on the right and left MCAs, respectively. In the same day, the patient evolved with non-photo reacting pupils and absent corneal reflex. The clinical examination remained unchanged even after analgesic and sedative drug withdrawal. With this prognostic, the clinical staff opted for comfort care. The patient died on the ninth day after hospitalization.

Experimental Protocol
The experimental protocol was designed so that the optical measurements would not interfere with any clinical practices. Therefore, the timing of each optical monitoring session was purposely chosen to avoid conflicts with any planned clinical intervention. Most of the monitoring sessions happened in the early afternoon, prior to family visitation hours. No treatment decisions were made based on the optical measurements.
The optical system used in this study consists of a hybrid instrument employing a DCS module and a DOS module (Supplementary Figure 1). For details about the optical instrumentation and the optical probe, see Supplementary Methods. In this work, we opted to focus on a single DCS source-detector separation (2.5 cm) and two DOS wavelengths (690 and 846 nm). These separations allowed the optical device to probe the external surface of the prefrontal cortex (PFC), as it was shown in previous studies (19,38). We developed a user-friendly graphical user interface (GUI) based on current clinical monitors to control the operation of each module and to display in real-time CBF, OEF, and CMRO 2 .
We utilized the FD-DOS data to estimate the absolute oxy-and deoxy-hemoglobin concentrations (HbO and HbR, respectively). Total hemoglobin concentration was calculated as the sum of HbO and HbR (i.e., HbT = HbO + HbR). Tissue oxygen saturation (StO 2 ) was calculated as the fraction of HbO considering all hemoglobin concentrations (i.e., StO 2 = HbO/HbT). Oxygen extraction fraction (OEF) was computed as OEF = (SaO 2 − StO 2 )/(0.75 * SaO 2 ), where SaO 2 is the arterial blood oxygen saturation (39,55). DOS data was combined with DCS data to provide cerebral blood flow (CBF), and metabolism. From CBF and OEF, we estimated the cerebral metabolic rate of oxygen (CMRO 2 ) as CMRO 2 = Hgb * CBF * OEF (39,40,55). All neurophysiological parameters were calculated for each time point continuously every 6 s. The whole analysis procedure is detailed in the Supplementary Methods and can be found elsewhere (32,33,57).
Optical measurements were taken for five of the 9 days of the patient hospitalization for at least 1 h per day. The optical probe was held on the forehead with an elastic strap. Each brain hemisphere was measured sequentially, for at least 30 min on each side. All monitoring sessions were supervised by a specialized neurologist. Mean arterial pressure (MAP), arterial blood saturation (SaO 2 ), hemoglobin concentration (Hgb) and CBF v from TCD were retrospectively recovered from the medical records and were taken at most an hour before or after the optical measurements.
Due to the lack of continuous measurement of systemic physiology, we opted to report the average data points of each neurophysiological parameter calculated across each optical monitoring session. Supplementary Figure 2 shows the continuous measurement of all the optical-derived parameters for one representative monitoring session. Figure 1 shows a timeline of the main events during the patient hospitalization, along with the daily averages of all the cerebral measurements obtained with our diffuse optical system along with MAP and the CT scans performed during the whole hospitalization period.

Optical Monitoring
We started monitoring the patient with our diffuse optical system on the second day following admission, ∼32 h after hospitalization. At that time, MAP was 90 mmHg and we found lower CBF and CMRO 2 on the ipsilesional side of the PFC when compared to the contralesional side, consistent with lower metabolism following a brain hemorrhage. The OEF and the cerebral blood volume (CBV, as estimated by HbT) were similar between the two hemispheres On the third day of hospitalization, MAP increased to 122 mmHg, and we observed an overall increase in CBF and CMRO 2 . The increase in CBF and CMRO 2 was comparatively higher in the ipsilesional side. In addition, an increase in ipsilesional OEF and a decrease in ipsilesional HbT were also measured, suggesting high metabolic demand due to low blood availability in tissue. No optical monitoring was taken between the fourth and the fifth days due to intermittent clinical care to assist the patient.
On the sixth day of hospitalization, OEF was below 40% in both hemispheres. Compared to the second day of hospitalization, CBF increased by 40 and 51% in the ipsilesional and contralesional sides, respectively. However, the increase in CBF was not followed by any significant changes in CMRO 2 (Figure 1). From the second to the sixth day after hospitalization, HbT increased in the contralesional side and decreased in the ipsilesional side. MAP at the sixth day was 85 mmHg. For the following 2 days, ipsilesional CBF and CMRO 2 further decreased, respectively, by 82 and 69% (compared to day 6), and OEF dramatically increased by ∼76%. CBV (as estimated by HbT) decreased in the contralesional side, reaching the levels of the ipsilesional side. The ipsilesional CBV remained relatively stable between the second and the ninth days of hospitalization. MAP decreased to 67 and 45 mmHg during the last two monitoring sessions, respectively.

DISCUSSION
Diffuse optics has been previously validated in medical settings as an alternative bedside monitoring tool, and it holds potential to personalize clinical management. Despite limitations in the penetration depth inherent to any diffuse optical technique (32,37,38), both DOS and DCS are capable of monitoring cerebral hemodynamics non-invasively and in real-time, without the use of exogenous contrasts, ionizing radiation, and without interfering with current clinical practices. Unlike global systemic physiological monitors, the principles of light propagation limit the measurement to a region of interest, which can be useful to monitor specific regions, such as an ischemic penumbra. In addition, due to its portability and low cost, diffuse optical methods can be very useful to clinicians in places where clinical care conditions are restricted, such as in military fields and in developing regions. Even in better clinical conditions, access to daily neuroimaging data for longer periods of time may not be readily available.
Here, we report the results from an observational case report concerning a patient severely affected by aSAH, who was admitted almost 3 days after the event. For simplicity, we limited our measurements to the forehead, and thus we were only sensitive to focal changes in the PFC, which was sufficient to elucidate the main events during the evolution of the patient's brain injury. Since we focused on the feasibility of performing longitudinal measurements in the neuro-ICU, we minimized both the size of the optical probe and the monitoring time per day to reduce the interference with clinical interventions. However, we note that these choices are not intrinsic limitations of the optical system. The optical probes are small and customized so that they can be temporarily detached from patients for any necessary reason. Since our optical system measures absolute values, we are not limited to a baseline or reference period. Also, the system's high portability allows its relocation alongside the patient with minimal time and effort. Additionally, long-term continuous monitoring (i.e., several hours per day) poses no extra-challenge in the present system. In fact, previous studies have reported diffuse optical measurements at the ICU for as long as 8 h per day, both in adults (16,31) and in neonates (58).
The results from this pilot observational case study provide evidence that neurophysiological information derived from DOS and DCS is consistent with the triphasic evolution of SAH: hypoperfusion, hyperemia, and vasospasm (59,60). During the first day of monitoring, the patient had poor ipsilesional CBF and CMRO 2 but normal OEF when compared to the contralesional side, consistent with a hypoperfusion phase. In a previous PET study, Yundt et al. found similar trends of reduced CBF and CMRO 2 , but unaffected OEF, during the early phases of SAH (61,62). This pattern of low CBF and CMRO 2 but unaffected OEF may be related to an ongoing ischemia, to the start of the vasospasm phase or to a sign that the injury-related neurologic deficits have not yet peaked. Based on the admission CT scan (Figure 1), our results are most likely related to an ongoing ischemia. The initial SAH insult may impact mitochondrial function and diminish oxidative metabolism in favor of anaerobic metabolism (63), which may further explain our findings of low CBF and CMRO 2 , but unaffected OEF.
At the third day after hospitalization, we found an increase in OEF at the ipsilesional side, which was associated with a bilateral increase in CBF and CMRO 2 (with a comparatively higher ipsilesional increase). Previously, Yokose et al. found decreased ipsilesional oxygen saturation (which corresponds to increased OEF) in severe SAH patients with angiographic vasospasm with TD-DOS (64). Thus, the higher OEF during the third day after hospitalization could be an indicative of vasospasm. The fact that the patient evolved with an ipsilesional ischemia between the fourth and fifth days is another indicative of vasospasm.
Taken together, our results corroborate Yokose et al.'s (64), and reinforces the argument that decreases in tissue oxygenation (or, similarly, increases in OEF) may be a more sensitive measure of vasospasm than CBF v measured by TCD. Of note, prolonged elevation of ICP may also induce ischemia in aSAH patients; in this study we did not assess ICP, therefore we cannot confirm the real cause for ischemia in this patient.
We also measured increased CBF and decreased OEF in the contralesional side of the patient between the second and the sixth days following hospitalization, with no significant difference in MAP. The increase in CBF with lower OEF is an indicative of cerebral hyperemia that may have been secondary to an inflammatory mediated process causing vasodilation of the microvascular bed (63,65). The cerebral hyperemia in the contralesional side may also be explained by an attempt to restore metabolic balance in the affected tissue. However, by this time the patient had already developed a large ischemic insult, and only the healthy tissue was capable of increasing CBF and appropriately oxygenating the tissue. Additionally, the ipsilesional ischemia may explain the fact that CBV (as measured by HbT) remained approximately constant across the different monitoring days (besides the first monitoring session).
Overall, the optically derived hemodynamic parameters were sensitive to the mismatch between the physiology of each hemisphere. We note, though, that further clinical studies with larger populations are required to assess the accuracy of the technique to find specific events, and/or to measure the effects of potential treatments. Together, our results reinforce that diffuse optics may provide useful information at the bedside, which could aid clinicians to make decisions based on each patient's neurophysiology. The real-time access to patient's neurophysiological condition may positively impact clinical care and patient outcome (5)(6)(7). In addition, our results show the feasibility of performing diffuse optical monitoring in a restricted environment in a developing country, where access to other methods of cerebral monitoring are not always readily available.
Even though the optical results show strong agreement with clinical outcome, it is worth noting that the optical parameters can be more accurate. In this work, we modeled the head as a semi-infinite medium for the DCS/DOS data analysis. By using the simplest model for analysis in order to get realtime monitoring, our results do not account for extracortical contributions. Thus, it is likely that we have underestimated the cerebral physiology changes (33,37,38,66). In the future, multilayer models can be used to isolate extracortical contributions from the optical signal (67-71). However, these models either require an extra step in the monitoring setup or may not yet be reliable for clinical use. Nonetheless, the quantification of the neurophysiological changes does not affect the comparison between hemispheres, since extracerebral contributions are expected to similarly affect both hemispheres. Last, the opticalderived CBF has physical units (i.e., cm 2 /s) rather than the more usual clinical units (i.e., ml/100 g of tissue/min). Some authors have proposed the use of an indocyanine-green (ICG) bolus to recover absolute CBF from the optical signal, which may be used to convert the continuous CBF index from DCS to the absolute clinical units (72)(73)(74).

CONCLUSION
To summarize, in this study we presented an observational case report recorded during the translation of a hybrid diffuse optical system for real-time monitoring of cerebral hemodynamics inside a neuro-ICU. Our results suggest that optics may provide supporting information to monitor secondary damage following brain injury, and that optics can be safely and non-invasively applied early during hospitalization, at the bedside and with minimal interference with standard clinical practices. The optically-derived hemodynamic parameters provide neurophysiological information at the microvasculature, which was consistent with the clinical outcome. Although we focused on one single case, our results support that diffuse optics can be a reliable tool for bedside monitoring and encourage further clinical validations with DCS/DOS in the neuro-ICU. Translation of diffuse optics to clinical settings can positively impact patient outcome and can be especially useful in lowbudget hospitals and in remote areas, where cerebral physiology monitoring is essential but not readily available.

DATA AVAILABILITY STATEMENT
The datasets generated for this study are available on request to the corresponding author.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Ethics Committee of the University of Campinas (Comitê de Ética em Pesquisa). The patients/participants provided their written informed consent to participate in this study. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.

AUTHOR CONTRIBUTIONS
AQ, LV, MK, RF, and RM collected all the data. RF wrote the first draft of the manuscript and conducted the final analysis. RF and RM designed the data analysis protocol. All authors assisted with data interpretation, contributed to manuscript revision, read, and approved the submitted version.

FUNDING
We acknowledge the support by the São Paulo Research Foundation (FAPESP) through Proc. 2012/02500-8 (RM), 2014/25486-6 (RF) and 2013/07559-3. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.