This PDF file contains the front matter associated with SPIE Proceedings Volume 11945, including the Title Page, Copyright information, Table of Contents, and Conference Committee listings.

Cerebral blood flow (CBF) monitoring is crucial during cerebrovascular surgery to inform decision making. In cerebral aneurysm clipping cases, CBF monitoring is routinely used to confirm patency in vessels and determine successful aneurysmal obliteration. Current intraoperative tools for CBF monitoring such as indocyanine green angiography (ICGA) do not provide real-time and continuous assessment of CBF.

Introduction: Endovascular embolization is becoming an increasingly utilized method of treating a variety of neurovascular disorders, including aneurysms, arteriovenous malformations (AVMs), and tumors. Many of the existing limitations of this treatment are related to the embolization agents currently available, including material compaction or migration, disease recurrence, or off-target embolization. Hydrogels are a promising class of materials that may be utilized to address some of these concerns. Methods: We compounded hydrogel formulations that were low-viscosity, shear thinning, photo-sensitive, and radioopaque. We developed a method of intravascular micro-catheter hydrogel delivery with dynamic modulation of hydrogel physical characteristics at the tip of the catheter, via photo-crosslinking with an integrated UV emitting optical fibre. This allowed for rapid transition from liquid to solid state to block blood flow at the vascular target, as well as dynamic modulation to suit the needs of a variety of neurovascular disorders. We performed preliminary testing of this novel methodology in animal models of neurovascular disease. Results: With dynamic modulation of photo-crosslinking, we were able to deliver hydrogels with a viscosity range of up to 10^4 Pa*s. The technique allowed for successful deposition of the hydrogel precursor in animal models for aneurysms, AVMS, and tumors. Post-procedural angiography demonstrated satisfactory occlusion of target vessels without evidence of complications. Conclusions: This novel embolization method holds promise in improving the safety and efficacy of the endovascular treatment of a variety of different pathologies and should be investigated further with direct comparative studies.

Optical coherence tomography angiography (OCTA) has been widely used for neuroimaging with non-invasive and high-resolution advantages. However, the signals from the skull and the noise from the deep imaging areas reduce the microvascular clarity in the OCTA projections. Here we proposed a U-Net deep learning method to segment the superficial cortical area from the skull and other tissues for improving the quality of the OCTA projections. The peak signal-to-noise ratio (pSNR) and the average contrast-to-noise ratio (aCNR) were analyzed to evaluate the OCTA projection images. The results showed that the pSNR and aCNR values increased significantly and, thus, the image quality of the microvascular projections was improved after the cortical segmentation.

Diffuse optical tomography (DOT) is a widely used optical method for functional neuroimaging. When a dense grid of optodes is used, DOT can produce functional brain imaging maps that are comparable with fMRI in terms of spatial resolution. However, when the available number of sources and detectors are limited, it is important to understand where to place them to image a region of interest (ROI) with optimal coverage and sensitivity to the brain. Conventionally, the optode configuration is heuristically determined by the experimenter. Recently, a total sensitivity maximizing algorithm was proposed to answer the question as an optimization problem. However, in larger ROIs and complex geometries, optodes tend to crowd over only a small portion of the ROI and have therefore unsatisfactory coverage of the ROI. ArrayDesigner was proposed as an attempt to address these limitations by adding a coverage term, so that the algorithm can balance between sensitivity and coverage of the ROI. In this work, we demonstrate that such modification still may not suffice in certain geometries, and the inherent nature of sensitivity maximization can limit the uniformity of imaging. We propose a resolution- based optimization algorithm Grid Resolution with Optimized Uniformity Placement (GROUP)" that aims to provide high-resolution, uniform DOT imaging. We show simulated data using a realistic head model that GROUP can provide better uniformity of resolution and has the flexibility to tune the focus to voxels of different depths.

We investigated the effects of modulation frequency on image quality with FD-HD-DOT through simulations with a realistic noise model of functional activations in human head models, arising from 11 source modulation frequencies between CW and 1,000 MHz. Functional reconstructions were simulated in five fMRI-based head models covered by 158 light sources and 166 detectors and a realistic noise model was considered. We quantitatively evaluated image quality by assessing the localization error (LE), success rate, full width at half maximum (FWHM), and full volume at half maximum (FVHM) of recoveries. The results show that the optimized frequencies are about 300~500 MHz for image quality and 300 MHz for reliable field of view when accounting for realistic noise.

Because contemporary intraoperative tumor detection modalities, such as intraoperative MRI, are not ubiquitously available and can disrupt surgical workflow, there is an imperative for an accessible diagnostic device that can meet the surgeon’s needs in identifying tissue types. The objective of this paper is to determine the efficacy of a novel non-contact tumor detection device for metastatic melanoma boundary identification in a tissue-mimicking phantom, evaluate the identification of metastatic melanoma boundaries in ex vivo mouse brain tissue, and find the error associated with identifying this boundary. To validate the spatial and fluorescence resolution of the device, tissue-mimicking phantoms were created with modifiable optical properties. Phantom tissue provided ground truth measurements for fluorophore concentration differences with respect to spatial dimensions. Modeling metastatic disease, ex vivo melanoma brain metastases were evaluated to detect differences in fluorescence between healthy and neoplastic tissue. This analysis includes determining required-to-observe fluorescence differences in tissue. H&E staining confirmed tumor presence in mouse tissue samples. The device detected a difference in normalized average fluorescence intensity in all three phantoms. There were differences in fluorescence with the presence and absence of melanin. The estimated tumor boundary of all tissue phantoms was within 0.30 mm of the ground truth tumor boundary for all boundaries. Likewise, when applied to the melanoma-bearing brains from ex vivo mice, a difference in normalized fluorescence intensity was successfully detected. The potential prediction window for the tumor boundary location is less than 1.5 mm for all ex vivo mouse brain tumors boundaries. We present a non-contact, laser-induced fluorescence device that can identify tumor boundaries based on changes in laser-induced fluorescence emission intensity. The device can identify phantom ground truth tumor boundaries within 0.30 mm using instantaneous rate of change of normalized fluorescence emission intensity and can detect endogenous fluorescence differences in melanoma brain metastases in ex vivo mouse tissue.

Perception-action cycle-based motor learning theory postulates coupled action and perception for visuomotor learning. We hypothesized that perception-action-related brain connectivity will underpin visuomotor skill levels in a complex motor task based on this theory. We tested our hypothesis using multi-modal brain imaging on healthy human subjects (N=6 experts, N= 8 novice, all right-handed) during the performance of fundamentals of laparoscopic surgery (FLS) "suturing and intracorporeal knot-tying" task. We investigated dynamic directed brain networks using nonoverlapping sliding window-based spectral Granger causality (GC) from simultaneously acquired electroencephalogram (EEG), and functional near-infrared spectroscopy (fNIRS) signals. Our GC analysis on EEG signals showed the flow of information from the supplementary motor area complex (SMA) to the left primary motor cortex (LM1) that was statistically different (p <0.05) between the experts and novices. This result aligned with the perception action cycle theory where SMA is central to the orderly descent from the prefrontal to the motor cortex in Fuster's perception-action processing stages. The GC analysis of the fNIRS oxyhemoglobin signal revealed the connectivity from left to right primary motor cortex (LM1 to RM1) and LM1 to left prefrontal cortex (LPFC) that was significantly different (p <0.05) between the cohorts. Here, our preliminary results supported the involvement of perception-action-related directed brain connectivity in distinguishing the skill levels during a complex laparoscopic task that was measured with portable brain imaging during task performance. Future studies need to investigate the fusion of the EEG and fNIRS networks for the causal brain-behavior analysis of complex motor skill acquisition.

Optical neuroimaging is a promising tool to assess motor skills execution. Especially, functional near-infrared spectroscopy (fNIRS) enables the monitoring of cortical activations in scenarios such as surgical task execution. fNIRS data sets are typically preprocessed to derive a few biomarkers that are used to provide a correlation between cortical activations and behavior. Meanwhile, Deep Learning methodologies have found great utility in the data processing of complex spatiotemporal data for classification or prediction tasks. Here, we report on a Deep Convolutional model that takes spatiotemporal fNIRS data sets as input to classify subjects performing a Fundamentals of Laparoscopic Surgery (FLS) task used in board certification of general surgeons in the United States. This convolutional neural network (CNN) uses dilated kernels paired with multiple stacks of convolution to capture long-range dependencies in the fNIRS time sequence. The model is trained in a supervised manner on 474 FLS trials obtained from seven subjects and assessed independently by stratified-10-fold cross-validation (CV). Results demonstrate that the model can learn discriminatory features between passed and failed trials, attaining 0.99 and 0.95 area under the Receiver Operating Characteristics (ROC) and Precision-Recall curves, respectively. The reported accuracy, sensitivity, and specificity are 97.7%, 81%, and 98.9%, respectively.