Optimization of time domain diffuse correlation spectroscopy parameters for measuring brain blood flow

1060 nm interferometric near-infrared spectroscopy

Recently, interferometric near-infrared spectroscopy (iNIRS) has emerged to measure diffuse light field fluctuations with time-of-flight (TOF) resolution. Yet, current iNIRS implementations suffer from low signal-to-noise ratio (SNR). Longer wavelengths, with lower photon energy, lower reduced scattering in biological tissues, and higher permissible exposures, have the potential to increase SNR. Here, we investigate iNIRS at 1060 nm. Across various forehead locations, we find that the autocorrelation SNR is improved 3.7-9.3 times compared to 855 nm and 6.0-33.5 times compared to 773 nm at TOFs of 800-1000 ps. Physical system parameters account for much of this improvement, but the tissue response may also play a role. We conclude that wavelengths near 1060 nm can potentially improve iNIRS measurements of TOF-resolved speckle fluctuations.

Fast blood flow index reconstruction of diffuse correlation spectroscopy using a back-propagation-free data-driven algorithm

This study introduces a fast and accurate online training method for blood flow index (BFI) and relative BFI (rBFI) reconstruction in diffuse correlation spectroscopy (DCS). We implement rigorous mathematical models to simulate the auto-correlation functions (g 2) for semi-infinite homogeneous and three-layer human brain models. We implemented a fast online training algorithm known as random vector functional link (RVFL) to reconstruct BFI from noisy g 2. We extensively evaluated RVFL regarding both speed and accuracy for training and inference. Moreover, we compared RVFL with extreme learning machine (ELM) architecture, a conventional convolutional neural network (CNN), and three fitting algorithms. Results from semi-infinite and three-layer models indicate that RVFL achieves higher accuracy than the other algorithms, as evidenced by comprehensive metrics. While RVFL offers comparable accuracy to CNNs, it boosts training speeds that are 3900-fold faster and inference speeds that are 19.8-fold faster, enhancing its generalizability across different experimental settings. We also used g 2 from one- and three-layer Monte Carlo (MC)-based in-silico simulations, as well as from analytical models, to compare the accuracy and consistency of the results obtained from RVFL and ELM. Furthermore, we discuss how RVFL is more suitable for embedded hardware due to its lower computational complexity than ELM and CNN for training and inference.

Towards high-performance deep learning architecture and hardware accelerator design for robust analysis in diffuse correlation spectroscopy

This study proposes a compact deep learning (DL) architecture and a highly parallelized computing hardware platform to reconstruct the blood flow index (BFi) in diffuse correlation spectroscopy (DCS). We leveraged a rigorous analytical model to generate autocorrelation functions (ACFs) to train the DL network. We assessed the accuracy of the proposed DL using simulated and milk phantom data. Compared to convolutional neural networks (CNN), our lightweight DL architecture achieves 66.7% and 18.5% improvement in MSE for BFi and the coherence factor β, using synthetic data evaluation. The accuracy of rBFi over different algorithms was also investigated. We further simplified the DL computing primitives using subtraction for feature extraction, considering further hardware implementation. We extensively explored computing parallelism and fixed-point quantization within the DL architecture. With the DL model's compact size, we employed unrolling and pipelining optimizations for computation-intensive for-loops in the DL model while storing all learned parameters in on-chip BRAMs. We also achieved pixel-wise parallelism, enabling simultaneous, real-time processing of 10 and 15 autocorrelation functions on Zynq-7000 and Zynq-UltraScale+ field programmable gate array (FPGA), respectively. Unlike existing FPGA accelerators that produce BFi and the β from autocorrelation functions on standalone hardware, our approach is an encapsulated, end-to-end on-chip conversion process from intensity photon data to the temporal intensity ACF and subsequently reconstructing BFi and β. This hardware platform achieves an on-chip solution to replace post-processing and miniaturize modern DCS systems that use single-photon cameras. We also comprehensively compared the computational efficiency of our FPGA accelerator to CPU and GPU solutions.

A comprehensive overview of diffuse correlation spectroscopy: Theoretical framework, recent advances in hardware, analysis, and applications

Diffuse correlation spectroscopy (DCS) is a powerful tool for assessing microvascular hemodynamic in deep tissues. Recent advances in sensors, lasers, and deep learning have further boosted the development of new DCS methods. However, newcomers might feel overwhelmed, not only by the already-complex DCS theoretical framework but also by the broad range of component options and system architectures. To facilitate new entry to this exciting field, we present a comprehensive review of DCS hardware architectures (continuous-wave, frequency-domain, and time-domain) and summarize corresponding theoretical models. Further, we discuss new applications of highly integrated silicon single-photon avalanche diode (SPAD) sensors in DCS, compare SPADs with existing sensors, and review other components (lasers, sensors, and correlators), as well as data analysis tools, including deep learning. Potential applications in medical diagnosis are discussed and an outlook for the future directions is provided, to offer effective guidance to embark on DCS research.

Time-domain diffuse correlation spectroscopy at large source detector separation for cerebral blood flow recovery

Time-domain diffuse correlation spectroscopy (td-DCS) enables the depth discrimination in tissue's blood flow recovery, considering the fraction of photons detected with higher time of flight (TOF) and longer pathlength through the tissue. However, the recovery result depends on factors such as the instrument response function (IRF), analyzed TOF gate start time, gate width and the source-detector separation (SDS). In this research we evaluate the performance of the td-DCS technique at three SDSs of 1.5, 2 and 2.5 cm to recover cerebral blood flow (CBF). To do that we presented comprehensive characterization of the td-DCS system through a series of phantom experiments. First by quality metrices such as coefficient of variation and contrast-to-noise ratios, we identified optimal time gate(s) of the TOF to extract dynamics of particles. Then using sensitivity metrices, each SDS ability to detect dynamics of particles in superficial and deeper layer was evaluated. Finally, td-DCS at each SDS was tested on healthy volunteers during cuff occlusion test and breathing tasks. According to phantom measurements, the sensitivity to estimate perfusion within the deep layer located at depth of 1.5 cm from the surface can be increased more than two times when the SDS increases from 1.5 cm to 2.5 cm.

Interferometric near-infrared spectroscopy (iNIRS) reveals that blood flow index depends on wavelength

Blood flow index (BFI) is an optically accessible parameter, with unit distance-squared-over-time, that is widely used as a proxy for tissue perfusion. BFI is defined as the dynamic scattering probability (i.e. the ratio of dynamic to overall reduced scattering coefficients) times an effective Brownian diffusion coefficient that describes red blood cell (RBC) motion. Here, using a wavelength division multiplexed, time-of-flight- (TOF) - resolved iNIRS system, we obtain TOF-resolved field autocorrelations at 773 nm and 855 nm via the same source and collector. We measure the human forearm, comprising biological tissues with mixed static and dynamic scattering, as well as a purely dynamic scattering phantom. Our primary finding is that forearm BFI increases from 773 nm to 855 nm, though the magnitude of this increase varies across subjects (23% ± 19% for N = 3). However, BFI is wavelength-independent in the purely dynamic scattering phantom. From these data, we infer that the wavelength-dependence of BFI arises from the wavelength-dependence of the dynamic scattering probability. This inference is further supported by RBC scattering literature. Our secondary finding is that the higher-order cumulant terms of the mean squared displacement (MSD) of RBCs are significant, but decrease with wavelength. Thus, laser speckle and related modalities should exercise caution when interpreting field autocorrelations.

Comparing the performance potential of speckle contrast optical spectroscopy and diffuse correlation spectroscopy for cerebral blood flow monitoring using Monte Carlo simulations in realistic head geometries

Significance

The non-invasive measurement of cerebral blood flow based on diffuse optical techniques has seen increased interest as a research tool for cerebral perfusion monitoring in critical care and functional brain imaging. Diffuse correlation spectroscopy (DCS) and speckle contrast optical spectroscopy (SCOS) are two such techniques that measure complementary aspects of the fluctuating intensity signal, with DCS quantifying the temporal fluctuations of the signal and SCOS quantifying the spatial blurring of a speckle pattern. With the increasing interest in the use of these techniques, a thorough comparison would inform new adopters of the benefits of each technique.

Aim

We systematically evaluate the performance of DCS and SCOS for the measurement of cerebral blood flow.

Approach

Monte Carlo simulations of dynamic light scattering in an MRI-derived head model were performed. For both DCS and SCOS, estimates of sensitivity to cerebral blood flow changes, coefficient of variation of the measured blood flow, and the contrast-to-noise ratio of the measurement to the cerebral perfusion signal were calculated. By varying complementary aspects of data collection between the two methods, we investigated the performance benefits of different measurement strategies, including altering the number of modes per optical detector, the integration time/fitting time of the speckle measurement, and the laser source delivery strategy.

Results

Through comparison across these metrics with simulated detectors having realistic noise properties, we determine several guiding principles for the optimization of these techniques and report the performance comparison between the two over a range of measurement properties and tissue geometries. We find that SCOS outperforms DCS in terms of contrast-to-noise ratio for the cerebral blood flow signal in the ideal case simulated here but note that SCOS requires careful experimental calibrations to ensure accurate measurements of cerebral blood flow.

Conclusion

We provide design principles by which to evaluate the development of DCS and SCOS systems for their use in the measurement of cerebral blood flow.

Quantification of blood flow index in diffuse correlation spectroscopy using a robust deep learning method

Significance

Diffuse correlation spectroscopy (DCS) is a powerful, noninvasive optical technique for measuring blood flow. Traditionally the blood flow index (BFi) is derived through nonlinear least-square fitting the measured intensity autocorrelation function (ACF). However, the fitting process is computationally intensive, susceptible to measurement noise, and easily influenced by optical properties (absorption coefficient μ a and reduced scattering coefficient μ s ' ) and scalp and skull thicknesses.

Aim

We aim to develop a data-driven method that enables rapid and robust analysis of multiple-scattered light's temporal ACFs. Moreover, the proposed method can be applied to a range of source-detector distances instead of being limited to a specific source-detector distance.

Approach

We present a deep learning architecture with one-dimensional convolution neural networks, called DCS neural network (DCS-NET), for BFi and coherent factor (β ) estimation. This DCS-NET was performed using simulated DCS data based on a three-layer brain model. We quantified the impact from physiologically relevant optical property variations, layer thicknesses, realistic noise levels, and multiple source-detector distances (5, 10, 15, 20, 25, and 30 mm) on BFi and β estimations among DCS-NET, semi-infinite, and three-layer fitting models.

Results

DCS-NET shows a much faster analysis speed, around 17,000-fold and 32-fold faster than the traditional three-layer and semi-infinite models, respectively. It offers higher intrinsic sensitivity to deep tissues compared with fitting methods. DCS-NET shows excellent anti-noise features and is less sensitive to variations of μ a and μ s ' at a source-detector separation of 30 mm. Also, we have demonstrated that relative BFi (rBFi) can be extracted by DCS-NET with a much lower error of 8.35%. By contrast, the semi-infinite and three-layer fitting models result in significant errors in rBFi of 43.76% and 19.66%, respectively.

Conclusions

DCS-NET can robustly quantify blood flow measurements at considerable source-detector distances, corresponding to much deeper biological tissues. It has excellent potential for hardware implementation, promising continuous real-time blood flow measurements.

Diffuse Correlation Spectroscopy: A Review of Recent Advances in Parallelisation and Depth Discrimination Techniques

Diffuse correlation spectroscopy is a non-invasive optical modality used to measure cerebral blood flow in real time, and it has important potential applications in clinical monitoring and neuroscience. As such, many research groups have recently been investigating methods to improve the signal-to-noise ratio, imaging depth, and spatial resolution of diffuse correlation spectroscopy. Such methods have included multispeckle, long wavelength, interferometric, depth discrimination, time-of-flight resolution, and acousto-optic detection strategies. In this review, we exhaustively appraise this plethora of recent advances, which can be used to assess limitations and guide innovation for future implementations of diffuse correlation spectroscopy that will harness technological improvements in the years to come.

Comparison of numerical-integration-based methods for blood flow estimation in diffuse correlation spectroscopy

BACKGROUND AND OBJECTIVE

Diffuse correlation spectroscopy (DCS) is an optical blood flow monitoring technology that has been utilized in various biomedical applications. In signal processing of DCS, nonlinear fitting of the experimental data and the theoretical model can be a hindrance in real-time blood flow monitoring. As one of the approaches to resolve the issue, INISg1, the inverse of numerical integration of squared g1 (a normalized electric field autocorrelation function), that could surpass the state-of-the-art technique at the time in terms of signal processing speed, has been introduced. While it is possible to implement INISg1 using various numerical integration methods, no relevant studies have been performed. Meanwhile, INISg1 was only tested within limited experimental conditions, which cannot guarantee the robustness of INISg1 in various experimental conditions. Thus, this study aims to introduce variants of INISg1 and perform a thorough comparison of the original INISg1 and its variants.

METHODS

In this study, based on the right Riemann sum (RR) and trapezoid rule (TR) of numerical integration, INISg1_RR and INISg1_TR are suggested. They are thoroughly compared with the original INISg1 using model-based simulations that offer us control of most of the experimental conditions, including integration time, β, and photon count rate.

RESULTS

Except for some extreme cases, INISg1 performed more robustly than INISg1_RR and INISg1_TR. However, in extreme conditions, variants of INISg1 performed better than INISg1. With the same condition, the signal processing speed of INISg1 was 1.63 and 1.98 times faster than INISg1_RR and INISg1_TR, respectively.

CONCLUSION

This study shows that INISg1 is robust in most cases and the study can be a guide for researchers using INISg1 and its variants in different types of DCS applications.

scatterBrains: an open database of human head models and companion optode locations for realistic Monte Carlo photon simulations

Significance

Monte Carlo (MC) simulations are currently the gold standard in the near-infrared and diffuse correlation spectroscopy (NIRS/DCS) communities for generating light transport paths through tissue. However, realistic and diverse models that capture complex tissue layers are not easily available to all; moreover, manually placing optodes on such models can be tedious and time consuming. Such limitations may hinder the adoption of representative models for basic simulations and the use of these models for large-scale simulations, e.g., for training machine learning algorithms.

Aim

We aim to provide the NIRS/DCS communities with an open-source, user-friendly database of morphologically and optically realistic head models, as well as a succinct software pipeline to prepare these models for mesh-based Monte Carlo simulations of light transport.

Approach

Sixteen anatomical models were created from segmented T1-weighted magnetic resonance imaging (MRI) head scans and converted to tetrahedral mesh volumes. Approximately 800 companion scalp surface locations were distributed on each model, comprising full head coverage. A pipeline was created to place custom source and optical detectors at each location, and guidance is provided on how to use these parameters to set up MC simulations.

Results

The models, head surface locations, and all associated code are freely available under the scatterBrains project on Github.

Conclusions

The NIRS/DCS community benefits from having shared resources for conducting MC simulations on realistic head geometries. We hope this will make MRI-based head models and virtual optode placement easily accessible to all. Contributions to the database are welcome and encouraged.

Assessing depth sensitivity in laser interferometry speckle visibility spectroscopy (iSVS) through source-to-detector distance variation and cerebral blood flow monitoring in humans and rabbits

Recently, speckle visibility spectroscopy (SVS) was non-invasively applied on the head to monitor cerebral blood flow. The technique, using a multi-pixel detecting device (e.g., camera), allows the detection of a larger number of speckles, increasing the proportion of light that is detected. Due to this increase, it is possible to collect light that has propagated deeper through the brain. As a direct consequence, cerebral blood flow can be monitored. However, isolating the cerebral blood flow from the other layers, such as the scalp or skull components, remains challenging. In this paper, we report our investigations on the depth-sensitivity of laser interferometry speckle visibility spectroscopy (iSVS). Specifically, we varied the depth of penetration of the laser light into the head by tuning the source-to-detector distance, and identified the transition point at which cerebral blood flow in humans and rabbits starts to be detected.

Fast time-domain diffuse correlation spectroscopy with superconducting nanowire single-photon detector: system validation and in vivo results

Time-domain diffuse correlation spectroscopy (TD-DCS) has been introduced as an advancement of the "classical" continuous wave DCS (CW-DCS) allowing one to not only to measure depth-resolved blood flow index (BFI) but also to extract optical properties of the measured medium without using any additional diffuse optics technique. However, this method is a photon-starved technique, specially when considering only the late photons that are of primary interest which has limited its in vivo application. In this work, we present a TD-DCS system based on a superconducting nanowire single-photon detector (SNSPD) with a high quantum efficiency, a narrow timing response, and a negligibly low dark count noise. We compared it to the typically used single-photon avalanche diode (SPAD) detector. In addition, this system allowed us to conduct fast in vivo measurements and obtain gated pulsatile BFI on the adult human forehead.

Interferometric diffusing wave spectroscopy imaging with an electronically variable time-of-flight filter

Diffuse optics (DO) is a light-based technique used to study the human brain, but it suffers from low brain specificity. Interferometric diffuse optics (iDO) promises to improve the quantitative accuracy and depth specificity of DO, and particularly, coherent light fluctuations (CLFs) arising from blood flow. iDO techniques have alternatively achieved either time-of-flight (TOF) discrimination or highly parallel detection, but not both at once. Here, we break this barrier with a single iDO instrument. Specifically, we show that rapid tuning of a temporally coherent laser during the sensor integration time increases the effective linewidth seen by a highly parallel interferometer. Using this concept to create a continuously variable and user-specified TOF filter, we demonstrate a solution to the canonical problem of DO, measuring optical properties. Then, with a deep TOF filter, we reduce scalp sensitivity of CLFs by 2.7 times at 1 cm source-collector separation. With this unique combination of desirable features, i.e., TOF-discrimination, spatial localization, and highly parallel CLF detection, we perform multiparametric imaging of light intensities and CLFs via the human forehead.

Numerical approach to quantify depth-dependent blood flow changes in real-time using the diffusion equation with continuous-wave and time-domain diffuse correlation spectroscopy

Diffuse correlation spectroscopy (DCS) is a non-invasive optical technique that can measure brain perfusion by quantifying temporal intensity fluctuations of multiply scattered light. A primary limitation for accurate quantitation of cerebral blood flow (CBF) is the fact that experimental measurements contain information about both extracerebral scalp blood flow (SBF) as well as CBF. Separating CBF from SBF is typically achieved using multiple source-detector channels when using continuous-wave (CW) light sources, or more recently with use of time-domain (TD) techniques. Analysis methods that account for these partial volume effects are often employed to increase CBF contrast. However, a robust, real-time analysis procedure that can separate and quantify SBF and CBF with both traditional CW and TD-DCS measurements is still needed. Here, we validate a data analysis procedure based on the diffusion equation in layered media capable of quantifying both extra- and cerebral blood flow in the CW and TD. We find that the model can quantify SBF and CBF coefficients with less than 5% error compared to Monte Carlo simulations using a 3-layered brain model in both the CW and TD. The model can accurately fit data at a rate of <10 ms for CW data and <250 ms for TD data when using a least-squares optimizer.

Functional Time Domain Diffuse Correlation Spectroscopy

Time-domain diffuse correlation spectroscopy (TD-DCS) offers a novel approach to high-spatial resolution functional brain imaging based on the direct quantification of cerebral blood flow (CBF) changes in response to neural activity. However, the signal-to-noise ratio (SNR) offered by previous TD-DCS instruments remains a challenge to achieving the high temporal resolution needed to resolve perfusion changes during functional measurements. Here we present a next-generation optimized functional TD-DCS system that combines a custom 1,064 nm pulse-shaped, quasi transform-limited, amplified laser source with a high-resolution time-tagging system and superconducting nanowire single-photon detectors (SNSPDs). System characterization and optimization was conducted on homogenous and two-layer intralipid phantoms before performing functional CBF measurements in six human subjects. By acquiring CBF signals at over 5 Hz for a late gate start time of the temporal point spread function (TPSF) at 15 mm source-detector separation, we demonstrate for the first time the measurement of blood flow responses to breath-holding and functional tasks using TD-DCS.

Complete head cerebral sensitivity mapping for diffuse correlation spectroscopy using subject-specific magnetic resonance imaging models

We characterize cerebral sensitivity across the entire adult human head for diffuse correlation spectroscopy, an optical technique increasingly used for bedside cerebral perfusion monitoring. Sixteen subject-specific magnetic resonance imaging-derived head models were used to identify high sensitivity regions by running Monte Carlo light propagation simulations at over eight hundred uniformly distributed locations on the head. Significant spatial variations in cerebral sensitivity, consistent across subjects, were found. We also identified correlates of such differences suitable for real-time assessment. These variations can be largely attributed to changes in extracerebral thickness and should be taken into account to optimize probe placement in experimental settings.

First-in-clinical application of a time-gated diffuse correlation spectroscopy system at 1064 nm using superconducting nanowire single photon detectors in a neuro intensive care unit

Recently proposed time-gated diffuse correlation spectroscopy (TG-DCS) has significant advantages compared to conventional continuous wave (CW)-DCS, but it is still in an early stage and clinical capability has yet to be established. The main challenge for TG-DCS is the lower signal-to-noise ratio (SNR) when gating for the deeper traveling late photons. Longer wavelengths, such as 1064 nm have a smaller effective attenuation coefficient and a higher power threshold in humans, which significantly increases the SNR. Here, we demonstrate the clinical utility of TG-DCS at 1064 nm in a case study on a patient with severe traumatic brain injury admitted to the neuro-intensive care unit (neuroICU). We showed a significant correlation between TG-DCS early (ρ = 0.67) and late (ρ = 0.76) gated against invasive thermal diffusion flowmetry. We also analyzed TG-DCS at high temporal resolution (50 Hz) to elucidate pulsatile flow data. Overall, this study demonstrates the first clinical translation capability of the TG-DCS system at 1064 nm using a superconducting nanowire single-photon detector.

Development of a Monte Carlo-wave model to simulate time domain diffuse correlation spectroscopy measurements from first principles

SIGNIFICANCE

Diffuse correlation spectroscopy (DCS) is an optical technique that measures blood flow non-invasively and continuously. The time-domain (TD) variant of DCS, namely, TD-DCS has demonstrated a potential to improve brain depth sensitivity and to distinguish superficial from deeper blood flow by utilizing pulsed laser sources and a gating strategy to select photons with different pathlengths within the scattering tissue using a single source-detector separation. A quantitative tool to predict the performance of TD-DCS that can be compared with traditional continuous wave DCS (CW-DCS) currently does not exist but is crucial to provide guidance for the continued development and application of these DCS systems.

AIMS

We aim to establish a model to simulate TD-DCS measurements from first principles, which enables analysis of the impact of measurement noise that can be utilized to quantify the performance for any particular TD-DCS system and measurement geometry.

APPROACH

We have integrated the Monte Carlo simulation describing photon scattering in biological tissue with the wave model that calculates the speckle intensity fluctuations due to tissue dynamics to simulate TD-DCS measurements from first principles.

RESULTS

Our model is capable of simulating photon counts received at the detector as a function of time for both CW-DCS and TD-DCS measurements. The effects of the laser coherence, instrument response function, detector gate delay, gate width, intrinsic noise arising from speckle statistics, and shot noise are incorporated in the model. We have demonstrated the ability of our model to simulate TD-DCS measurements under different conditions, and the use of our model to compare the performance of TD-DCS and CW-DCS under a few typical measurement conditions.

CONCLUSION

We have established a Monte Carlo-Wave model that is capable of simulating CW-DCS and TD-DCS measurements from first principles. In our exploration of the parameter space, we could not find realistic measurement conditions under which TD-DCS outperformed CW-DCS. However, the parameter space for the optimization of the contrast to noise ratio of TD-DCS is large and complex, so our results do not imply that TD-DCS cannot indeed outperform CW-DCS under different conditions. We made our code available publicly for others in the field to find use cases favorable to TD-DCS. TD-DCS also provides a promising way to measure deep brain tissue dynamics using a short source-detector separation, which will benefit the development of technologies including high density DCS systems and image reconstruction using a limited number of source-detector pairs.

Scanning interferometric near-infrared spectroscopy

In diffuse optics, quantitative assessment of the human brain is confounded by the skull and scalp. To better understand these superficial tissues, we advance interferometric near-infrared spectroscopy (iNIRS) to form images of the human superficial forehead blood flow index (BFI). We present a null source-collector (S-C) polarization splitting approach that enables galvanometer scanning and eliminates unwanted backscattered light. Images show an order-of-magnitude heterogeneity in superficial dynamics, implying an order-of-magnitude heterogeneity in brain specificity, depending on forehead location. Along the time-of-flight dimension, autocorrelation decay rates support a three-layer model with increasing BFI from the skull to the scalp to the brain. By accurately characterizing superficial tissues, this approach can help improve specificity for the human brain.

Superconducting nanowire single-photon sensing of cerebral blood flow

Significance: The ability of diffuse correlation spectroscopy (DCS) to measure cerebral blood flow (CBF) in humans is hindered by the low signal-to-noise ratio (SNR) of the method. This limits the high acquisition rates needed to resolve dynamic flow changes and to optimally filter out large pulsatile oscillations and prevents the use of large source-detector separations ( ≥ 3    cm ), which are needed to achieve adequate brain sensitivity in most adult subjects. Aim: To substantially improve SNR, we have built a DCS device that operates at 1064 nm and uses superconducting nanowire single-photon detectors (SNSPD). Approach: We compared the performances of the SNSPD-DCS in humans with respect to a typical DCS system operating at 850 nm and using silicon single-photon avalanche diode detectors. Results: At a 25-mm separation, we detected 13 ± 6 times more photons and achieved an SNR gain of 16 ± 8 on the forehead of 11 subjects using the SNSPD-DCS as compared to typical DCS. At this separation, the SNSPD-DCS is able to detect a clean pulsatile flow signal at 20 Hz in all subjects. With the SNSPD-DCS, we also performed measurements at 35 mm, showing a lower scalp sensitivity of 31 ± 6 % with respect to the 48 ± 8 % scalp sensitivity at 25 mm for both the 850 and 1064 nm systems. Furthermore, we demonstrated blood flow responses to breath holding and hyperventilation tasks. Conclusions: While current commercial SNSPDs are expensive, bulky, and loud, they may allow for more robust measures of non-invasive cerebral perfusion in an intensive care setting.