This PDF file contains the front matter associated with SPIE Proceedings Volume 11639 including the Title Page, Copyright information, and Table of Contents.

Welcome and Introduction to SPIE Conference 11639 Optical Tomography and Spectroscopy of Tissue XIV

Understanding and detecting viruses with surface-enhanced Raman Spectroscopy

Ultrasound (US)-guided diffuse optical tomography (DOT) has demonstrated potential value for screening and treatment monitoring of breast cancers. However, in clinical cases, the chest wall, bad probe-tissue contact, and tissue heterogeneity can create image artifacts, causing misinterpretation of lesion images. In the current work, realistic and flexible threedimensional numerical breast phantoms were generated using the Virtual Imaging Clinical Trials for Regulatory Evaluation (VICTRE) tools developed by U.S. Food and Drug Administration (FDA). By selecting physical attributes and tissue optical properties, the VICTRE breast phantoms were, for the first time, adopted in DOT for in silico studies. Monte Carlo simulations were conducted to generate the forward data. Edge artifacts (hot spots on the edge of the regions of interests) were found on the reconstructed images when there was a mismatch between the lesion-side breast and the contralateral reference-side breast. We propose a fully automated, connected components analysis-based algorithm that can remove these edge artifacts and improve lesion reconstruction.

Noninvasive diffuse optical spectroscopy (DOS) is promising as an adjunct diagnostic imaging technique for distinguishing benign and malignant breast lesions. However, most DOS approaches require normalizing lesion optical biomarkers to healthy values in reference tissue since major tissue constituents exhibit large inter-patient variations. Our goal is to determine whether absolute measurements of minor optical absorbers such as collagen and methemoglobin (metHb) can successfully distinguish benign and malignant breast lesions. In this retrospective clinical study, 30 lesions with BIRADS score >= 3 (12 benign and 18 malignant lesions from 28 subjects) measured with broadband quantitative DOS were analyzed for their oxy-, deoxy- hemoglobin, water, lipids, collagen, and metHb concentrations as well as their optical scattering characteristics. Amongst all absolute DOS parameters (i.e., without normalization to healthy tissue) considered, only absolute metHb was observed to be significant for lesion discrimination (0.43±0.18 μM for benign vs 0.87±0.32 μM for malignant, p = 0.0002). Our findings demonstrate that lesion metHb concentration measured by quantitative DOS can improve noninvasive optical diagnosis of breast malignancies.

Diffuse optical methods have demonstrated significant potential to monitor breast cancer treatment response. Unfortunately, many diffuse optical tools involve lengthy acquisition, slow data processing, or both, limiting their clinical utility. Here, we present a new frequency domain diffuse optical device capable of providing real-time tissue oxygenation maps. This device is made possible through the integration of three enabling technologies: real-time 3D probe tracking, rapid translation of raw data to optical properties, and new high-speed acquisition electronics. We have demonstrated the ability of this instrument to identify inclusions in phantoms as well as to make proof-of-concept measurements in breast tissue.

The use of technologies still under development (Silicon PhotoMultipliers and Time-to-Digital Converter) for the upgrade of our time-resolved optical mammograph highlighted the crucial need for a thorough and progressive instrument characterization (from extensive laboratory tests on phantoms to in vivo measurements), potentially of general interest. Despite satisfying results on phantoms, in vivo tests urged an instrumental change, with the adoption of a just released high-throughput TDC to obtain high-quality breast scans (much wider scan area, better day-by-day reproducibility, improved signal quality, simpler set-up). A clinical study on monitoring of neoadjuvant chemotherapy is now ongoing at the San Raffaele Hospital, Milan.

Optical imaging techniques have emerged as a possible alternative to predict pathological complete response (pCR) in breast cancer patients undergoing neoadjuvant chemotherapy (NAC). Our team developed a so-called diffuse optical tomographic breast imaging system (DOTBIS) which does not require the use of contrast agents or compression and enables imaging of the whole breast volume using low intensity near infrared light capable to measure tissue concentration of total hemoglobin (ctTHb). In this retrospective study, we evaluated 55 stage II-III BC patients in the neoadjuvant setting who received weekly paclitaxel x 12, followed by dose-dense adriamycin/cyclophosphamide every 2 weeks x 4. DOTBIS images were acquired from the patient whole breast volume at 6 different time points: at baseline (TP0); two weeks after the first taxane infusion (TP1); after four infusions of taxane (TP2); at the end of the taxane regimen and before starting AC cycle (TP3); after two AC infusions (TP4); and at the end of NAC and before surgery (TP5). In order to evaluate whether pCR status influences the change of ctTHb over time, we designed a multilevel mixed-effect model. pCR was defined as no invasive tumor cells from the breast and axillary tissue at surgery (ypT0 ypN0). Changes in ctTHb levels compared to baseline (TP0) values were statistically significant different between pCR (n = 20) and non-pCR (n=35) at all time points except at TP1 and at the end of the taxane cycle (TP3).

A machine learning (ML) model with physical constraints is introduced to perform diffuse optical tomography (DOT) reconstruction. Here, for the first time, we combine ultrasound-guided DOT with ML to facilitate DOT reconstruction. Our method has two key components: (i) An unsupervised auto-encoder with transfer learning is adopted for clinical data without a ground truth, and (ii) physical constraints are implemented to achieve accurate reconstruction. Both qualitative and quantitative results demonstrate that the accuracy of the proposed method surpasses that of the existing model. In a phantom study, compared with the Born conjugate gradient descent (CGD) reconstruction method, the ML method improves the reconstructed maximum absorption coefficient by 18.3% on high contrast phantom and by 61.3% on low contrast phantom, with improved depth distribution of absorption maps. In a clinical study, better contrast was obtained from a treated breast cancer pre- and post- treatment.

In this paper, we describe a deep convolutional neural network (DNN) model trained with simulated breast diffuse optical tomography data with realistic noise characteristics to solve the inverse problem in a fast single-pass feed forward reconstruction. In addition to an AUTOMAP-inspired network structure, our DNN model, a.k.a. FDU-Net, is also comprised of a U-Net to further improve the image quality. We demonstrate that our FDU-Net model can successfully recover nearly the full contrast of inclusions with accurate localization at millisecond-scale speed, outperforming the conventional finite element-based (FEM) methods. Trained with cases with a single spherical inclusion, the FDU-Net model can also recover multi-inclusions and irregular-shaped cases, demonstrating advantages of generalization.

To improve non-invasively the specificity in the diagnosis of breast cancer after a positive screening mammography or doubt/suspicious ultrasound examination, the SOLUS project developed a multimodal imaging system that combines: Bmode ultrasound (US) scans (to assess morphology), Color Doppler (to visualize vascularization), shear-wave elastography (to measure stiffness), and time domain multi-wavelength diffuse optical tomography (to estimate tissue composition in terms of oxy- and deoxy-hemoglobin, lipid, water, and collagen concentrations). The multimodal probe arranges 8 innovative photonic modules (optodes) around the US transducer, providing capability for optical tomographic reconstruction. For more accurate estimate of lesion composition, US-assessed morphological priors can be used to guide the optical reconstructions. Each optode comprises: i) 8 picosecond pulsed laser diodes with different wavelengths, covering a wide spectral range (635-1064 nm) for good probing of the different tissue constituents; ii) a large-area (variable, up to 8.6 mm² ) fast-gated digital Silicon Photomultiplier; iii) the acquisition electronics to record the distribution of time-of-flight of the re-emitted photons. The optode is the basic element of the optical part of the system, but is also a stand-alone, ultra-compact (about 4 cm³ ) device for time domain multi-wavelength diffuse optics, with potential application in various fields.

A new low-cost imaging system has been developed for MRI-guided Near-Infrared Spectral Tomography (MRI-NIRST) for breast cancer detection. In this new system, 8 flexible sensing strips with 4 source fibers and 4 photodetectors will be applied directly to the breast, from nipple towards chest-wall, to cover the entire breast. For each source illumination, data will be collected at 32 detector locations. The feasibility of this new imaging system was demonstrated using 3D simulation breast phantoms that created from real breast MR images. Our results indicated that total hemoglobin in a tumor smaller than 1 mm could be recovered with this system.

We propose to use neural networks to learn and replace Monte Carlo (MC) simulations. Our neural networks are not only convenient to use but also are shown to be extremely time-saving compared to MC simulations with comparable accuracy. Furthermore, we employ the machine learning method called transfer learning to perform calibration between MC simulated diffuse reflectance and that measured by Diffuse Reflectance Spectroscopy (DRS). The transfer learning model is able to predict DRS measured diffuse reflectance spectrum by training the model using a small amount of DRS data.

Performance assessment and standardization are indispensable for instruments of clinical relevance in general and clinical instrumentation based on photon migration/diffuse optics in particular. In this direction, a multi-laboratory exercise was initiated with the aim of assessing and comparing their performances. 29 diffuse optical instruments belonging to 11 partner institutions of a European level Marie Curie Consortium BitMap1 were considered for this exercise. The enrolled instruments covered different approaches (continuous wave, CW; frequency domain, FD; time domain, TD and spatial frequency domain imaging, SFDI) and applications (e.g. mammography, oximetry, functional imaging, tissue spectroscopy). 10 different tests from 3 well-accepted protocols, namely, the MEDPHOT2 , the BIP3 , and the nEUROPt4 protocols were chosen for the exercise and the necessary phantoms kits were circulated across labs and institutions enrolled in the study. A brief outline of the methodology of the exercise is presented here. Mainly, the design of some of the synthetic descriptors, (single numeric values used to summarize the result of a test and facilitate comparison between instruments) for some of the tests will be discussed.. Future actions of the exercise aim at deploying these measurements onto an open data repository and investigating common analysis tools for the whole dataset.

In patients with peripheral artery disease (PAD), plaque is accumulating in arteries which leads to a reduction in blood supply to the extremities. In advanced stages, surgical intervention is required to reopen the arteries and restore limb perfusion. During this procedure, it is important to correctly identify which areas of the foot lack perfusion. The standard procedure to obtain this information is X-ray angiography, which is performed repeatedly during the intervention. The disadvantage of this procedure is the relatively high radiation dose and extensive use of contrast agents. To reduce this problem, we evaluate in this pilot study (involving 4 patients) the ability of vascular optical spectroscopy (VOS) to detect the X-ray contrast agent permeating the angiosomes in the foot. We show that the contrast agent can be detected by optical measurements as it temporarily replaces the blood in the different angiosomes, which leads to a 1% to 5% change in the signal amplitude. In addition, measurements of the blood pooling in the foot were performed before the intervention. We observed a strong correlation between the angiosomes that showed a worsen state in the measurement done before the intervention and the absence of angiographic contrast agent signals during the intervention itself. Among the 4 patients monitored, 2 showed a response to the contrast agent in their angiosomes and they corresponded to the patients with a relatively better perfusion in the pre-intervention measurements.

One of the chief applications of diffuse optical spectroscopy is the measurement of chromophore concentrations in biological tissue, which requires measurements of tissue absorption. To achieve absolute absorption measurements, two chief confounds must be accounted for: instrumental contributions and tissue scattering. To account for instrumental contributions, a preliminary calibration on a phantom of known optical properties is typically done. The need for a calibration is eliminated by self-calibrating or dual-slope techniques using specially designed probe geometries. A technique that is capable of measuring tissue scattering is frequency domain near-infrared spectroscopy. However, it is typically not implemented for a spectrum of wavelengths due to instrumental complexity. Here we present a technique that combines self-calibrating frequency-domain at two wavelengths, to account for tissue scattering, and dual-slope continuous-wave broadband diffuse reflectance spectroscopy to achieve spectral measurements of absolute absorption between 600 nm and 1064 nm without any need for calibration. We apply this technique to two human tissues in vivo to determine concentrations of oxy-hemoglobin, deoxy-hemoglobin, lipids, and water. We found that the quality of the spectral fits may be significantly improved by the inclusion of a wavelength-independent background absorption. This leads to a discussion on the origin of this background absorption, and on the meaning of the chromophore concentrations that are recovered from spectral analysis. Current work is seeking to further understand and possibly correct for this apparent background absorption.

The ability to non-invasively monitor in-vivo the human muscle and adipose tissue is of great practical use and hence of growing interest in the fields of clinical diagnostics and preventive medicine. Optical methods, such as diffuse optical spectroscopy (DOS) applied in the near-infrared spectral region could be of great interest in clinical scenario. In this work, we present a pilot study based on multi-distance broadband time-domain diffuse optical spectroscopy (TD DOS) to characterize in vivo the subcutaneous adipose tissue (abdominal region) and the vastus lateralis muscle (thigh region). The study was performed using a fully automated portable TD DOS instrument on a set of 24 healthy adult volunteers. The optical properties of these two tissue types were obtained over the broad wavelength range of 600-1100 nm. The results suggest a clear influence of the stratified nature of the two regions considered, namely the abdomen and thigh, on the recovered optical properties. This work demonstrates how multi-distance broadband diffuse optical spectroscopy could be complimentary in fields like the non-invasive spectroscopy of adipose tissue and the standard DOS-based muscle oximetry.

Diffuse optical techniques have been used to assess joint inflammation in rheumatoid arthritis (RA); however, there is scant data on the use of time-domain (TD) techniques for this purpose. We conducted TD simulations in a realistic finger joint and investigated how TD parameters changed in response to simulated joint inflammation. An MRI image of a healthy human finger was segmented into 7 tissue types and manipulated to create 4 models representing various degrees of RA disease activity. NIRFAST was used to simulate TD light propagation through the models at 800nm using an array of light sources/detectors placed across the proximal interphalangeal (PIP) joint. Parameters were extracted from the resulting TPSFs using statistical moments (number of photons, mean time-of-flight, centralized variance), temporal binning (early, middle, late temporal windows), and Fourier decomposition to produce a series of 2D images. Spatial frequency components were then extracted from each of the images and used to detect differences between the 4 models of disease activity. Spatial frequency amplitudes of high-frequency temporal Fourier component images differentiated between models with 100% accuracy and with the highest percent differences (i.e., performance). However, once Poisson noise was added to the TPSFs, amplitudes of early photon and total number of photon images at the 0.14mm⁻¹ spatial frequency had the highest performance for accuracy cut-offs of 95% and 100%, respectively. Future work will focus on validating results with experimental TD-DOI data and assessing how combinations of various parameters can be used to increase performance while maintaining accuracy.

Oxygenation monitoring of deep biological tissues with a bedside non-invasive method is still a crucial issue that is not resolved in routine clinics. It is particularly important in several major clinical applications with significant societal and financial implications such as the management of severe traumatic brain injury or stroke, the monitoring of hemodynamic parameters during or after birth or the post-operative monitoring after breast reconstruction surgery. However, the chances of recovery are directly proportional to the rapidity of diagnosis and subsequent remedial treatment. We propose an optical Time-Resolved probe, which monitor noninvasively the level of oxygenation in deep tissues. Our solution combines a measurement device, a stethoscope-type probe and an analysis software that provides biomarkers of tissue oxygenation. It is validated on a cohort of 15 pigs with the detection and identification of all arterial and venous events induced by the experiments’ protocol. We implement two analysis methods– one is real-time and aim at detecting and identifying the occlusion; it is based on a Time-Resolved windowing approach. The other post-processes the data and focuses on the quantification purpose; it is based on a tomographic approach.

The LUCA device combines clinical ultrasound, time-domain near infrared and diffuse correlation spectroscopies with the aim of improving thyroid cancer screening sensitivity and specificity. The preliminary clinical campaign on patients (n=31) with thyroid nodules and healthy controls (n=11) allowed the characterization of the precision of the instrument and demonstrated that using a couple of biomarkers the muscle-to-nodule contrast allows an area under the curve of 0.92 for single-nodule patients and 0.77 for all patients in differentiating benign and malignant nodules in a receiver operating characteristic curve. We will present the updated results from the ongoing study.

We have addressed the challenge of investigating the lung using diffuse optics along four different directions, namely: 1) broadband time-domain diffuse optics (TD-DO) in the 600-1100 nm range to derive the mean chest optical properties; 2) Monte Carlo simulations to investigate the depth-sensitivity of TD-DO measurements assuming a layered structure of the chest; 3) single-wavelength TD-DO using a high power pulsed laser on 5 healthy volunteers on a dynamic protocol; 4) single-wavelength TD-DO measurements using a novel large area Silicon Photomultiplier (SIPM) detector module permitting acquisitions at 12 cm source-detector distance.

Frequency domain diffuse optical spectroscopy (FD-DOS) is a non-invasive technique used to determine the optical absorption and reduced scattering coefficients of biological tissue. Incident light is modulated at radio frequencies and the amplitude and phase shift induced by tissue are analyzed. Here, we evaluate different modulation frequency selection schemes using simulated data. We used noise-free data to evaluate the inverse model, and noisy data using instrument-specific noise models. We found that careful tuning of the inversion algorithm parameters is required to achieve good results and adding modulation frequencies can sometimes result in worse performance depending on the instrument.

Currently, no bedside tool measures placental hemoglobin properties. Thus, we developed a hybrid optical/ultrasound (US) system to measure and study placental oxygen dynamics noninvasively. The system combines a Frequency-Domain Diffuse Optical Spectroscopy (FD-DOS) instrument that is capable of probing tissue as deep as 5 cm below the surface, with a commercial US device to provide synchronized tissue morphological information. Multi-layer models of photon diffusion are then utilized to reconstruct placental hemoglobin properties from FD-DOS data with morphological constraints from US. Phantom experiments, simulations, and studies in human subjects demonstrate that this instrument quantifies in vivo placental hemodynamics non-invasively.

Extracorporeal membrane oxygenation (ECMO) is an important therapy for critically ill children but survivors have neurodevelopmental impairments. Cerebral inflammatory response resulting in brain edema is observed on ECMO. This pathologic response may adversely impact the quantitative accuracy of diffuse optical spectroscopy (DOS) neuromonitoring (including commercial NIRS) which commonly assumes a 75% water fraction. Using fresh brain tissue desiccation, we directly quantified the severity of cerebral edema in pediatric swine following cardiac arrest, CPR and 22-24 hours of ECMO therapy. The fractional error in DOS quantification of cerebral hemodynamics from assuming 75% water fraction was determined to be <5%.

Time-domain fNIRS facilitates the elimination of the influence of extracerebral, systemic effects on measured signals since it contains time-of-flight information that is related to the penetration depth. Employing perturbation and MonteCarlo simulations, we quantitatively characterized and compared the performance of measurands based on moments and time windows of time-of-flight distributions. We extend our analysis to investigate whether higher moments and MellinLaplace (ML) moments promise improvements in performance. The comparison is based on spatial sensitivity profiles as well as metrics for relative contrast, contrast-to-noise ratio (CNR), depth selectivity, and the product of CNR and depth selectivity for layered absorption changes. The influence of reduced scattering coefficient, thickness of the superficial layer, and source-detector distance was analyzed. The third central moment performs similarly to variance and is worth considering for data analyzes. Higher order ML moments perform similarly to time windows and they likewise provide variable depth selectivity.

Despite its advantages in terms of safety, low cost and portability, the reliability of functional near-infrared spectroscopy (fNIRS) is challenged by substantial signal contamination from hemodynamic changes in the extracerebral layer (ECL). The time-resolved (tr) variant of NIRS can improve the sensitivity to the brain by recording the distribution of times-offlight (DTOF) of diffusely reflected photons that contain both time and intensity information. trNIRS data can be analyzed to obtain signals related to absorption changes at different depths within the medium; however, it can still be affected by ECL contamination. To further improve the isolation of the brain signal, this study adapted regression analysis, commonly used with short-channel functional NIRS, to trNIRS. Signals related to the early-arriving photons (0th moment, gates), selected based on sensitivity analysis, were used as the regressors, given their inherent sensitivity to superficial tissue. Performance of the regression was optimized using data from previously published studies that used trNIRS to measure oxygenation responses to hypercapnia caused by a rapid increase in end-tidal carbon dioxide pressure (PETCO2). To assess the effect of the regression approach, correlations between reconstructed hemoglobin signals and modelled hemodynamic response function were calculated. The results confirmed that the regression approach successfully removed large residue signals observed in the oxyhemoglobin signals.

In this work we provide some examples of sensitivity of the dual slope method to localized absorption changes in the layered geometry. Reasonably, this model geometry better represents many types of tissue. The sensitivity is shown in a two- and three-layer geometry for alternating current (AC) and phase data for both point-like and layered-like absorption perturbations. Contrary to the homogeneous medium geometry, where the ratio of deep to superficial tissue sensitivity of phase is always greater than that of AC, this is not always the case in the layered geometry. Therefore, depending on the targeted tissue, subject and protocol, in some cases it might be preferable to use AC dual-slopes, whereas in other cases phase dual-slope may be a better choice.

We have recently introduced a Dual-Slope (DS) method implemented with Frequency-Domain (FD) NearInfraRed Spectroscopy (NIRS). Its chief advantages are a preferential sensitivity to deeper tissue and the suppression of instrumental artifacts. Thus-far, the work using the DS method has focused on local measurements of tissue hemodynamics. The next step in the development of DS is the measurement of spatially resolved absorption changes of tissue in vivo. To achieve this, we designed a DS imaging array for applications on human tissue. We utilized this array to measure maps of absolute optical properties in human skeletal muscle, and muscle hemodynamics during venous occlusion. In each case, spatial maps were created. The maps of baseline absolute optical properties showed higher scattering in connective tissue and higher absorption in muscle tissue. The analysis of hemodynamics found a greater blood accumulation during venous occlusion in muscle. Overall, muscle hemodynamics were shown to be spatially variable over a large area, suggesting the importance of imaging (as opposed to single-location) measurements. The preliminary data on human subjects with this new DS imaging array pave the way for applications in functional NIRS (fNIRS) for mapping brain activation.

Time-resolved Near-Infrared Spectroscopy (trNIRS) methods typically use multiple wavelengths and source-detector distances in conjunction with a solution of the diffusion approximation to quantify tissue blood content and oxygenation. This approach can be both computationally intensive and costly, as multiple detectors are required. We propose a novel two-layer fitting approach for multi-wavelength trNIRS, which uses a single detector while providing accurate estimates of cerebral oxygen saturation (ScO2) and hemoglobin content. The method uses a multi-step fitting algorithm to establish rough estimates of the absorption and scattering coefficients in the extracerebral layer and the brain, and subsequently refine those estimates, to improve accuracy while reducing crosstalk and complexity. Validation was conducted using Monte Carlo simulations in a realistic adult head model with appropriate optical properties at 680nm, 750nm, 800nm, and 830nm. The detector was located 30 mm anteriorly from the source, which was placed 50 mm above the right temple. Scalp oxygen saturation (SO2) (50%, 60%, and 70%) and ScO2 (40%-80%, 2% increments) were varied independently. The recovered ScO2 had a difference (mean±standard deviation) of 2.31±2.93% from inputted values, and cerebral total hemoglobin was recovered with a difference of 2.94±3.47%. Such high accuracy demonstrates the robustness of this computationally efficient two-layer fitting approach for analyzing multi-wavelength trNIRS measurements acquired with a single detector. Future work will involve validating the technique in tissue mimicking phantoms and animal studies.

Thanks to recent developments in time-domain near-infrared spectroscopy (TD-NIRS), it is now possible to monitor brain activity during freely moving exercises. A wearable single-channel TD-NIRS instrument has been exploited to retrieve oxy- and deoxygenated hemoglobin absolute concentrations during gaiting experiments. It was possible to retrieve quantitative results concerning functional brain activation in the motor cortex areas, with high rejection of motion artifacts.

Diffuse correlation spectroscopy (DCS) is an optical modality used to measure an index of blood flow in biological tissue. This blood flow index depends on both the red blood cell flow rate and density (i.e., hematocrit), although the functional form of hematocrit dependence is not well delineated. Herein, we develop and validate a novel tissue-simulating phantom containing hundreds of microchannels to investigate the influence of hematocrit on blood flow index. For a fixed flow rate, we demonstrate a significant inverse relationship between hematocrit and blood flow index that must be accounted for to accurately estimate blood flow under anemic conditions.

Diffuse correlation spectroscopy (DCS) is an emerging diffuse optical technique that quantifies microvascular blood flow. In spite of the wide range of clinical/research applications, DCS instrumentation is not yet standardized. In this study, we have analyzed the effect of experimental parameters as the measurement duration time and the number of detectors, at different photon count-rates, on the precision of a DCS experiment. This provides a recipe for finding device and experimental settings that optimize the precision while balancing cost and temporal resolution.

Jones fractures, which occur in the proximal fifth metatarsal of the foot carry a significant risk of nonunion. Nonunions increase the patient’s financial burden, risk of infection, and may take years to heal completely. Exploiting the fact that a reason for this higher risk of nonunion is inadequate blood supply, we will use diffuse optics to monitor the fracture healing process. We recently developed a speckle contrast optical tomography system which we will combine and co-register with a frequency domain-diffuse optical tomography system. This combination will provide a powerful instrument optimized for monitoring patient risk of nonunion during fracture healing.

Phenylephrine is commonly used in cardiac surgery is common to increase mean arterial pressure without affecting cardiac output; however, its effects on the microcirculation of the brain and in skeletal muscle are unclear. The objective of this study is to use hyperspectral near infrared spectroscopy (h-NIRS) and diffuse correlation spectroscopy (DCS) to monitor the microcirculation during a phenylephrine bolus would be able to discern the timing and effects of a phenylephrine bolus on brain and skeletal muscle microvasculature. The h-NIRS subsystem uses two spectrometers and a halogen light source, and the DCS subsystem uses a long coherence diode laser (785 nm) and a single photon counting module. Probes were positioned on the left hind limb and the top of the skull, with a source-detector distance of 10mm. Data were collected from Sprague Dawley rats (n = 1, 158 g). Nine microvascular challenges were monitored by recording a two-minute baseline, then injecting 0.1mL of phenylephrine (0.1 ug/mL) intravenously, followed by collecting data for 4 additional minutes. Oxygen saturation increased by 9% in the brain and 2% in the muscle. Blood flow increased by 60% in the brain, and 17% in the muscle. This study is the first report on the use of h-NIRS and DCS to investigate the effects of phenylephrine in the microvasculature. Dissimilitude in flow response may be due to differences in regulation mechanisms. Future work will include acquiring data from both subsystems simultaneously to provide further insight into the relationship between oxygen saturation and blood flow.

Plasmonic photothermal therapy is an emerging complementary method for treating tumors usually studied pre-clinically in mice with small subcutaneous tumors. In this project, it is conducted on a more realistic orthotopic xenograft renal carcinoma mouse model, while tissue optical properties are monitored non-invasively by diffuse optics. Hybrid diffuse reflectance spectroscopy and diffuse correlation spectroscopy are used to quantify tissue hemodynamics, the accumulated plasmonic nanoparticle concentration and therapy induced physiological changes. The results show the need for personalization and the predictive capabilities of the monitoring, which, in turn, would allow for therapy optimization.

Diffuse correlation spectroscopy (DCS) is an optical technique which, by studying the speckle intensity fluctuations of coherent light diffused in a turbid medium, retrieves information regarding the scatterers motion. In the case of biological tissues, the particles of interest are the red blood cells, from which is possible to measure non-invasively microvascular blood flow (BF). However, being based on a continuous-wave light source, depth discrimination is achievable only by using multiple source-detector separations. On the other hand, time-domain (TD) DCS is a novel approach which exploits a pulsed yet coherent light source to discriminate the intensity fluctuations at different photon time-of-flights. This additional information is beneficial for in vivo applications, due to the physical relationship between photon time-of-flight and mean depth penetration. TD-DCS is typically performed in the spectral range between 700 and 800 nm. Here, we explore TD-DCS in a new spectral range compared to the typical one, moving to the spectral region beyond the water absorption peak (i.e., >970 nm). We performed liquid phantom and in vivo experiments on the human muscle at a wavelength of 1000 nm. Also, the possible advantages in terms of depth sensitivity and signal-to-noise ratio are discussed.

Intracranial pressure (ICP) is a critical biomarker measured invasively with the risk of complications. There is a need for non-invasive methods to estimate ICP. Diffuse correlation spectroscopy (DCS) allows the non-invasive measurement of pulsatile, microvascular cerebral blood flow which contains information about ICP. Recently, our proof-of-concept study used machine-learning to deduce ICP from DCS signals to estimate ICP resulting in excellent linearity and a reasonable accuracy (±4 mmHg). Here, we extend to a multi-center (three centers) data set of adults with acute brain injury (N=34). We will present the results from the complete data set as new data flows in.

Monitoring critical closing pressure (CrCP) can be a useful and noninvasive measure of intracranial pressure (ICP), especially in patients with high risk factors for brain injury. We monitored five patients undergoing cardiac ablation procedures using diffuse correlation spectroscopy (DCS). We utilized the prolonged diastolic events that occur during this procedure to validate non-invasive measurements of CrCP with DCS. to estimate the gold standard CrCP during long diastolic events induced during the procedure and compared them to estimations from normal pressure and flow waveforms prior to each event.

We present a novel approach to speckle contrast optical spectroscopy that utilizes multi-wavelength, multi-distance and multi-exposure time datasets to simultaneously estimate the blood flow and optical properties in the underlying tissue. This approach offers improvements in depth penetration over existing techniques and provides additional information content regarding the hemoglobin concentration and oxygen saturation of the tissue. We demonstrate the accuracy of this approach through computational simulations at source-detector separations of 1~2 cm and compare performance with existing multi-distance and multi-exposure time approaches.

The premature brain embodies an underdeveloped vascular system, which can lead to poor cerebral blood flow (CBF), impaired metabolism, and subsequent brain injury. NNeMo (Neonatal NeuroMonitor) is an in-house built brain monitor that provides continuous and simultaneous measurements of CBF, tissue saturation (StO2), and metabolism. Nine premature infants were monitored for 6 h on day 1 and 3 of life. An oscillatory signal was observed in CBF and StO2 which diminished by day 3; metabolic response was not impacted by minor fluctuations in perfusion. Hemodynamic neuromonitoring could aid in predicting the onset of cerebral hemorrhaging or gauging brain injury severity.

Hydrocephalus is a disorder of cerebral spinal fluid (CSF) physiology that results in increased intracranial pressure (ICP). It is commonly treated via surgical placement of a shunt in the ventricles to divert CSF. Diffuse optical measurements of cerebral perfusion and oxygen extraction were acquired before and after surgical shunt placement in neonates with hydrocephalus. An invasive ICP measurement was made at the time of shunt placement. Shunting increased cerebral perfusion and decreased oxygen extraction only in infants with elevated ICP. This suggests abnormally low perfusion in patients with elevated ICP, and normal perfusion in patients without elevated ICP.

Near-infrared spectroscopy (NIRS) combined with diffuse correlation spectroscopy (DCS) provides a non-invasive approach for monitoring oxygenation, cerebral blood flow (CBF) and the cerebral metabolic rate of oxygen (CMRO₂); however, these methods are vulnerable to signal contamination from the extracerebral layer (ECL). The aim of this work was to evaluate methods of reducing the impact of this contamination using time-resolved (tr) NIRS and multi-distance (MD) DCS. Experiments involved healthy participants, and oxygenation and CBF changes in response to hypercapnia were measured. A pneumatic tourniquet was used to impede scalp blood flow to assess ECL contamination. Responses acquired with and without the tourniquet demonstrated that trNIRS technique substantially reduced scalp contributions in the oxygenation signals, while blood flow responses from the scalp and brain could be separated by analyzing MD DCS data with a multi-layer model. Finally, no change in CMRO₂ during hypercapnia was observed, despite the large increases in CBF and oxygenation. These results indicate that NIRS/DCS techniques can accurately monitor cerebral blood flow and metabolism, highlighting the potential of these techniques for neuromonitoring

Spatial frequency domain imaging (SFDI) is an imaging modality that projects spatially modulated light patterns to determine optical property maps for absorption and reduced scattering of biological tissue via a pixel-by-pixel data acquisition and analysis procedure. The light interaction theory behind SFDI is based upon homogenous properties, with forward models calculated via analytical solutions or Monte-Carlo, also used for the optical property recovery, using only a pixel-independent nature. This is known to be limited for samples with high heterogeneity, with an increased error observed for varying optical property boundaries. NIRFAST is an image modelling and reconstruction tool based upon FEM of the diffusion model that simulates complex heterogenic tissue interactions from single and multi-wavelength systems and is routinely used in a variety of clinical and pre-clinical applications. NIRFAST has been adapted for SFDI, allowing for pixel-dependent heterogenic simulations. Image reconstruction using existing methodologies is compared to data generated from complex models with NIRFAST to quantify the optical property reconstruction accuracy, whilst heterogenous models of varying optical property values and depths further demonstrate SFDIs parameter recovery capabilities. It is shown that pixel-dependent light interaction in tissue plays an important part of accurate optical map recovery and can affect quantitative accuracy. This work demonstrates full raw image SFDI simulations for heterogenous samples working towards the use of modelbased image reconstruction to allow a coupled, pixel-dependent SFDI image modelling and parameter recovery.

Regular blood lipid screening is recommended for those suffering from cardiovascular disease (CVD). Current blood lipid measurements require invasive blood draws and lab-based testing. We introduce here shortwave infrared spatial frequency domain imaging (SWIR SFDI) as a non-invasive technique for blood lipid monitoring. We will report results from an analysis of SFDI measurement parameters (spatial frequency, wavelength, etc.) and light propagation models that result in accurate optical properties and chromophore extractions in the SWIR. Additionally, we will present results from a healthy volunteer study of longitudinal postprandial blood lipid changes measured in superficial blood vessels with SWIR SFDI.

While many diffuse optical techniques are available to characterize the optical properties of biological tissues, a gold standard has not been proposed yet for referencing the results of these different technologies. In this context, we implemented and validated a multi-modal referencing platform for diffuse optical spectroscopic instruments, consisting of a benchtop time-resolved diffuse optical spectroscopic system, together with a spatial frequency domain imaging (SFDI) station. To validate the systems capabilities, the BitMap (“Brain Injury and Trauma Monitoring using Advanced Photonics”) European collaboration has been carried out, aiming at assessing the performances of several instruments dedicated to diffuse optical spectroscopic purposes.

In this study, spatial frequency domain imaging (SFDI) and temporal frequency domain imaging (TFDI) are combined to observe superficial and deep tissues simultaneously using a time-resolving CMOS image sensor. SFDI is an established non-invasive wide-field imaging method for superficial or shallow tissues. On the other hand, time-resolved spectroscopy based near-infrared spectroscopy (TRS-NIRS) is suitable for deep tissue measurement while it is based on point or multipoint measurement. Recently, time-resolving CMOS image sensors based on the single-photon avalanche diode and charge modulators have emerged. To take advantage of their area-imaging capability, we propose a spatio-temporal frequency domain imaging (STFDI=SFDI+TFDI) method, where pulsed binary stripe patterns with a pitch, p, and a duty ratio of 1/N are sequentially projected onto the tissue. While the projected pattern is shifted N times with a step of p/N, time-resolved images are captured for every shift. SFDI is conducted by performing fast Fourier transform (FFT) at each pixel after integrating them in time for each shift. For TFDI, the detected light in the middle of the stripes is analyzed by FFT in time for each shift. Based on the obtained amplitude and phase for specific harmonics orders, the absorption and scattering coefficients are estimated. This concept was verified by a GPU-based Monte Carlo simulator, MCX, with a two-layer skin model. We also experimentally confirmed the difference in the measured reflectance and phase for SFDI and TFDI when the thickness of the first layer was changed.

Tissue oxygenation (StO₂), which is the fraction of oxygenated hemoglobin in biological tissues, is an important biomarker that can reveal information about tissue viability and underlying pathologies. The continuous monitoring of StO₂ is also useful for surgical guidance and patient management. In recent years, Spatial Frequency Domain Imaging (SFDI) has emerged as an elegant solution for mapping wide-field StO₂. However, conventional SFDI requires capturing a sequence of images at different spatial frequencies and wavelengths, resulting in slow acquisition times and challenges with moving objects. Model-based single-snapshot techniques have shortened the acquisition time but introduce image artifacts and decrease accuracy. Here we propose a deep-learning technique for real-time StO₂ mapping from snapshot structured light images. We train content-aware generative adversarial networks (OxyGAN) on pairs of structured light input at 659nm and 851nm wavelengths and StO₂ ground truth predicted by conventional SFDI. We demonstrate that OxyGAN is not only capable of rapid data acquisition and processing but is also more accurate than a model-based benchmark. We also compare OxyGAN to a hybrid model that uses separate networks to estimate optical absorption at two wavelengths followed by Beer-Lambert fitting. The end-to-end OxyGAN approach shows better performance in terms of both speed and accuracy. We additionally demonstrate real-time OxyGAN by applying it to videos of in vivo tissues. OxyGAN has the potential to enable wide-field, real-time, and accurate tissue oxygenation measurements in many clinical applications.

The feasibility of the spatial frequency domain imaging (SFDI) for early treatment assessment in a mouse femoral injury model was investigated. For mice with and without treatment, the revascularization process was monitored by SFDI, which provides the total hemoglobin concentration (THC) of the injured bone and surrounding soft tissues. The biomechanical strength of the healed femurs was measured after 4 weeks of healing. The correlation between the SFDI-derived THC and the biomechanical strength will be investigated. Such correlation will be used to develop a prediction model to estimate the bone healing outcome at an early time point.

Mild hypothermic cardiopulmonary bypass (CPB) is used during neonatal cardiac surgery. To elucidate potential risk factors for brain injury, diffuse optical spectroscopy, diffuse correlation spectroscopy, and cerebral microdialysis techniques were employed to monitor cerebral hemodynamics during mild hypothermic CPB in a swine model. Optical metrics were further compared with microdialysis metrics. Stable microdialysis metabolite levels, a significant decrease in CMRO2, and trending (but not significant) decreases in CBF and increases in OEF were observed. This suggests that the optical metrics may be more sensitive to neurologic injury during CPB than microdialysis.

During surgery with cardiopulmonary bypass (CPB), maintaining adequate cerebral blood flow (CBF) is paramount to prevent adverse neurological outcome; tissue damage can occur if CBF reduction is sufficient to impair energy metabolism. Ten adult patients undergoing cardiothoracic surgery with CPB received perfusion and metabolic neuromonitoring using a novel optical system combining diffuse correlation spectroscopy and broadband near-infrared spectroscopy. CPB onset resulted in large increases in CBF and significant drops in mean arterial pressure and metabolism. No changes were observed transitioning off CPB. Real-time assessment of cerebral perfusion and metabolism could alert clinicians to relevant hemodynamic events before brain injury occurs.

Functional diffuse correlation spectroscopy was used during various attention-demanding motor tasks, and test whether cerebral blood flow (CBF) changes differentiate between age groups or are related to mild cognitive impairment (MCI). Sixty-nine subjects were measured during single (ST) and dual motor tasks (DT); seventeen Cognitively Healthy Younger (<65y.o.,CH-Y), sixteen Cognitively Healthy Older Adults (CH-O), fifteen subjects amnestic - MCI, single-domain (aMCI - SD) and twenty-one amnestic - MCI, multi-domain (aMCI - MD). Results show elevated CBF during cognitively demanding DT statistically different (p <0.001) from reference test. No statistical difference was observed between CH and aMCI subjects (p = 0.36).

We report a time-domain reflectance diffuse optical tomography (TD-RDOT) system for providing three-dimensional images of hemoglobin concentration, tissue oxygen saturation, water and lipid contents of breast cancer from reflectance measurements. A scan area of 5 × 5 grid points with a 10-mm spacing is marked on the breast surface so that the tumor is just below the center of the area. The breast scan is performed by measuring the temporal profiles of six wavelengths at each grid point using a time-domain diffuse optical spectroscopy (TD-DOS) system and a hand-held probe. The TDDOS system that we developed is capable of measuring water and lipid contents and hemoglobin concentration. The hand-held probe is designed to measure the breast in reflectance mode with a source-to-detector separation of 20 mm. The three-dimensional distributions of the tissue parameters are restored using an iterative image reconstruction method. As a preliminary clinical demonstration, a breast cancer patient with a tumor size of approximately 20 mm was examined with the TD-RDOT. The reconstructed images show that the breast cancer had high hemoglobin concentration and water content, and low tissue oxygen saturation and lipid content. The results indicate that the TD-RDOT system has the potential to provide diagnostically relevant information on the tissue characteristics of the tumor at the bedside.

A new photo-detector (PD) and source fiber based wearable MRI-guide near infrared spectroscopic tomographic imaging (MRg-NIRST) system has been designed and is underdevelopment. In this new system, eight flex circuit strips, each with 6 PDs and 6 side-firing fibers will be applied to the breast, prior to coil/biopsy positioning. The intensity of the diffused light will be acquired at 48x68 source-detector positions at 12 wavelengths. The 3D optical image reconstruction will be guided through the structural information obtained from T2 MRI and the scattering properties associated with MRI defined breast density.

The goal of this study was to develop reliable technique for measuring optical properties of renal calculi. As the first step, the technique was tested on artificial (Bego) stones. Absorption coefficients and reduced scattering coefficients of samples at 1940 nm were evaluated from total transmittance and diffuse reflectance measurements using an Inverse Monte Carlo technique. Optical properties were determined before and after soaking samples in distilled water. Before soaking, samples had absorption coefficients of 0.12±0.03 /mm, and reduced scattering coefficients of 7.72±2.34 /mm. After soaking, absorption coefficients were 0.21±0.04 /mm, and reduced scattering coefficients were 8.53±3.90 /mm. Future studies aim to determine optical properties of natural stones.

Near-infrared (NIR) optical imaging systems are developed in an effort to maximize signal to noise ratio (SNR). Recently, a SmartPhone Oxygenation Tool (SPOT) was developed to obtain tissue oxygenation maps from dual-wavelength NIR images. Herein, several noise removal techniques are implemented to the NIR data obtained with SPOT in an effort to improve SNR for low-cost effective imaging of tissue oxygenation maps. NIR and oxygenation changes obtained from standard occlusion studies on normal subjects are compared with and without the noise removal techniques and the improvement in SNR is assessed.

The use of phase (Φ) data collected in Frequency-Domain Near-InfraRed Spectroscopy (FD-NIRS) has not been widespread in measurements of skeletal muscle and has mainly been applied to measure absolute optical properties. We show that single-distance (SD) Φ has a deeper sensitivity compared to SD intensity (I) and can be more sensitive to oxygen consumption in skeletal muscle underneath superficial adipose tissue. We also show the potential benefit of single-slope (SS) or dual-slope (DS) I or Φ in muscle studies.

High Density Diffuse Optical Tomography (HD-DOT) as applied in fNIRS is largely limited to continuous wave (CW) data. Using a single modulation frequency, frequency domain (FD) HD-DOT has recently demonstrated better localisation of focal activation as compared to CW data. This study shows that combining CW and FD measurements and multiple modulation frequencies increases imaging performance in fNIRS. This work evaluates the benefits of multiple modulation frequencies, combining different frequencies as well as CW data in fNIRS HD-DOT. A slab geometry model was used, with activation occurring within a cortex layer. CW and FD measurements were simulated at 78, 141 and 203 MHz. The localization error, full width half maximum and effective resolution were evaluated. Across the average of the three metrics, at 141 MHz, FD performed 8.4% better than CW and the combination of CW and FD was 21.7% better than CW. FD measurements at 203 MHz performed 5% better than 78 MHz. Moreover, the three combined modulation frequencies of CW and FD performed up to 3.92% better than 141 MHz alone. This work shows that combining CW and FD measurements offer better performance than FD alone, with higher modulation frequencies increasing accuracy. Combining CW and FD measurements at multiple modulation frequencies yields the best overall performance.

Diffuse optical tomography is a non-invasive medical imaging method that quantitatively reconstructs threedimensional optical absorption and scattering coefficient distribution of imaging tissue domain. Functional diffuse optical tomography (fDOT) system uses multiple-wavelength near-infrared (NIR) light sources, and by reconstructing the variation of the absorption coefficient inside the tissue for these wavelengths, it maps the concentration of various constituents such as oxygenated and deoxygenated blood, water and lipids inside the tissue. This manuscript discusses the development of a 3D tissue scanning system based on fDOT principle. A configurable optical fiber probe was designed that can be placed on a tissue surface to inject modulated NIR light into the tissue from one of the three wavelengths (660, 735, and 850 nm) by light-emitting diodes (LED). The intensity of the back-reflected diffused light is measured using detector fibers coupled with a set of photodetectors. A microcontroller-based electronic circuit system and a dedicated DAQ unit perform system control, data acquisition, noise-free lock-in detection, and wireless data transmission to a host computer, equipped with a graphics processing unit (GPU). The GPU system executes a 3D fDOT spectroscopic image reconstruction software in real-time. Extensive simulation studies have been carried out for the 3D fDOT image reconstruction to study the concentration of chromophores such as oxyhemoglobin, deoxyhemoglobin, and water, and to characterize tumor located deep inside the tissue. The results from two different experimental setups validated the system. The promising results pave the way for the development of a low-cost 3D fDOT handheld system for the real-time onsite functional imaging of biological tissue.

Diffuse Optical Tomography (DOT) is an emerging non-invasive medical imaging method that can measure structural and functional properties of tissue, enabling it to detect abnormalities and tumors that other methods would miss. It may be described, in layman’s terms, as “3D ultrasound – but in color, and low resolution”. We present a fully parallel source and detector frequency-multiplexing method that enabled us to implement a fast 3D continuous-wave DOT (CW-DOT) system using low-cost embedded systems. Six simultaneously driven NIR LEDs and 24 photodiodes were fiber-optically coupled to a cylindrical phantom. A current driver board was developed to drive each LED with either a precise pre-set current or none. This was connected to an Arduino Due, which triggered each LED digitally with a precise frequency (0.5 – 2.5 kHz). The Due’s hardware timers were used to ensure frequency precision. This “division of labor” for precision in frequency and intensity allowed by square-wave modulation enabled significant reductions in cost and complexity and removed the need for multiple DACs for generating precise sinewaves. Signals from the 24 photodiodes were sampled simultaneously and processed using Fast Fourier Transform by a single-board computer (Raspberry Pi 3). We also describe a simple algorithm to mathematically remove the frequency crosstalk observed in the source LEDs, provided that all source intensities are also measured throughout. Experimental studies demonstrate a 2400x speed-up in the CW-DOT scan (1 s/scan) compared to the pre-existing serial system, which took 40 min/scan.

While CBCT provides excellent guidance capability, it is less adept at localizing soft tissue targets growing in a low image contrast environment. To advance image-guided irradiation for soft tissue targets, we developed quantitative bioluminescence tomography(QBLT) for small animal radiation research platform(SARRP). We presented a comprehensive approach to systematically tackle the challenging of BLT for in vivo target delineation, quantify its uncertainties in localization, and demonstrate the practicality for radiation guidance. Our QBLT-guided system provides a unique opportunity for investigators to guide conformal irradiation for soft tissue targets. The accurate image-guidance is expected to increase rigorous and reproducibility of scientific discovery.

Filtered backprojection (FBP) reconstruction is a simple and rapid technique for reconstructing tomographic data. Optical projection tomography (OPT) for instance, makes use of this technique to facilitate three dimensional visualization of optically clear biological tissues. However, for the case of nontransparent specimens where scattering dominates, more representative algorithms are required to model the behavior of light through the sample such that the object of interest can be recreated. In this work, model-based iterative techniques are investigated for use with an angle-restricted fluorescence OPT system for the specific application of imaging lymph nodes. Through physical and simulated phantoms, the effects of model inaccuracies in Monte Carlo-generated system matrices was evaluated. Findings demonstrated the importance of accurately capturing detector response, and that sample optical properties were more influential than sample geometry in affecting the reconstructed results.

The well-known red shift of fluorescence spectra in thick tissue causes significant cross-talk and inaccurate quantification of fluorophores using multispectral imaging (MSI). However, fluorescence lifetimes (FLT) are largely unaffected by tissue light propagation and results in superior quantitation accuracy. Here we present a systematic experimental comparison of quantitative MSI and FLT of multiple fluorescent molecules in tissue. In experiments using one or two fluorescence species in mixture with tissue autofluorescence, MSI showed significantly higher error rates (>20%) than FLT (<5%). These results show promise for in vivo FLT multiplexing based imaging of receptor expression in cancer and other diseases.