Automatic subject-specific extraction of cerebrovascular features
Summary: The goal of this study is to understand the effects of stroke on cerebral hemodynamics and to predict tissue viability in Stroke patients. We extract the patient specific geometry of the cerebral vascular network from CTA and MRA scans using image processing methods to obtain the 3D vessel volume. We also study the hemodynamics of cerebral vasculature using a computational model of the blood flow and perform comparisons with medical imaging measurements using 4D flow MRI and Transcranial Doppler Ultrasound. We then use novel machine learning techniques to extract predictive metrics from these patterns in clinical severity and risk assessment studies.
Top image:An overview of the cerebral vessel extraction process. We use Hessian based filtering to enhance the vessels in the grayscale images, followed by active-contours based segmentation to get the 3D vessel volume from which we extract network geometry consisting of centerlines, diameters and other features such as tortuosity etc.
Bottom image: A visual comparison of the cerebral vasculature of two healthy subjects and two stroke patients. We also performed a quantitative study to analyze the differences in geometry between the two groups, along with changes induced by the regular aging process.
End to end stroke triage using cerebrovascular morphology and machine learning
Summary: Rapid and accurate triage of acute ischemic stroke (AIS) is essential for early revascularization and improved patient outcomes. Response to acute reperfusion therapies varies significantly based on patient-specific cerebrovascular anatomy that governs cerebral blood flow. We present an end-to-end machine learning approach for automatic stroke triage. Employing a validated convolutional neural network (CNN) segmentation model for image processing, we extract each patient’s cerebrovasculature and its morphological features from baseline non-invasive angiography scans. These features are used to detect occlusion’s presence and the site automatically, and for the first time, to estimate collateral circulation without manual intervention. We then use the extracted cerebrovascular features along with commonly used clinical and imaging parameters to predict the 90 days functional outcome for each patient. The CNN model achieved a segmentation accuracy of 94% based on the Dice similarity coefficient (DSC). The automatic stroke detection algorithm had a sensitivity and specificity of 92% and 94%, respectively. The models for occlusion site detection and automatic collateral grading reached 96% and 87.2% accuracy, respectively. Incorporating the automatically extracted cerebrovascular features significantly improved the 90 days outcome prediction accuracy from 0.63 to 0.83. The fast, automatic, and comprehensive model presented here can improve stroke diagnosis, aid collateral assessment, and enhance prognostication for treatment decisions, using cerebrovascular morphology.
Cerebral hemodynamics and physics-informed machine learning
Summary: The objective of this proposal is to develop a non-invasive diagnostic tool for early detection of cerebral arterial vasospasm that combines routine clinical imaging and deep machine learning for improved diagnosis. Cerebral vasospasm (CVS) is a common (over 70%) and disabling consequence of non-traumatic subarachnoid hemorrhage (SAH).
