Optical Cerebral Hemodynamic Monitoring for the Detection of Large Vessel Occlusion - Endovascular therapy has revolutionized the treatment of acute stroke with large vessel occlusion (LVO) but is only available at a minority of stroke centers. Prehospital LVO detection provides an opportunity to route patients to thrombectomy-capable stroke centers and thereby reduce treatment times and improve outcomes. Emergency medical services (EMS) rely on examination-based stroke severity scales to identify patients with potential LVOs, but diagnostic accuracy is suboptimal, and implementation has been inconsistent. There is a need for an inexpensive, easy-to-use, portable instrument that can quickly and reliably detect LVO during prehospital care. To that end, our group recently validated a novel non-invasive optical imaging modality that leverages principles of speckle contrast optical spectroscopy to monitor cerebral blood flow (CBF). Innovative features of this optical technique facilitate high frequency data collection with excellent signal-to-noise, such that morphology of the CBF waveform can be well characterized. The cortical CBF waveform is particularly abnormal in patients with LVO, so after consolidating the optical scan into an easy to perform 70-second bedside study, we evaluated 135 patients who presented to the Emergency Department with clinical concern for acute stroke. The optical CBF waveform data were used to train a deep transformer neural network to recognize waveform features that discriminated LVO status. This optical classification achieved excellent sensitivity and specificity for LVO detection and outperformed two commonly used examination-based prehospital stroke severity scales. With this proposal, we have integrated the optical scan into the acute stroke alert workflow at three Emergency Departments within Penn Medicine, and we will validate the optical LVO detection algorithm by performing the 70-second optical scan on sequential patients being evaluated for potential acute stroke within 24 hours of onset. Urgent vascular imaging performed during the stroke alert workflow will provide the gold standard LVO categorization which will allow us to quantify the diagnostic performance of the optical LVO classification. Diagnostic performance will be compared with that of currently used prehospital stroke severity scales. This proposal will all also evaluate the feasibility of portable optical imaging in the prehospital environment through a collaborative effort with a large EMS provider that works closely with Penn Medicine. Optical instruments will be installed in several EMS vehicles, and the 70-second optical scan will be integrated into their prehospital workflow, such that all patients with acute onset focal deficit within 24 hours will be scanned during the prehospital stroke assessment. Diagnostic performance will be quantified, but this work will primarily be used to demonstrate feasibility, highlight potential technical challenges in this environment, and reveal opportunities to streamline workflow. This proposal provides critical data to spur the translation of optical CBF imaging for portable LVO detection.
