Abstract 34: Transcranial Detection and Classification of Stroke Using a Noninvasive Handheld Portable Device

Abstract

Introduction: Existing paradigms for stroke diagnosis typically involve computed tomography or magnetic resonance imaging to classify ischemic versus hemorrhagic stroke variants, as treatment for these subtypes varies widely. Delays in diagnosis and transport of unstable patients may worsen neurological status. Here, we demonstrate feasibility of rapid and accurate bedside stroke detection using a novel, handheld portable eddy current damping imaging device in live human clinical ischemic and hemorrhagic stroke settings. Methods: Copper wire was wound around an 11.4cm plastic cylinder to create a large solenoid coil and connected to an inductance-to-digital converter. Institutional Review Board approval was obtained and patients with hemorrhagic or ischemic stroke were recruited. The sensor coil was tangentially rotated across 8 rows on the patient’s head circumferentially. Data was plotted as a 2D heatmap to predict lesion type and location. For 3D figures, the 2D image was processed to convert the stroke-affected area from white to red (hemorrhage) and black to white (ischemia). Results: Consent, positioning each patient, and scanning with the sensor took roughly 15 minutes from start to end for each participant enrolled in our study and occurred at the patient bedside (n=8). Figure 1a and 1d show the location and type of lesion, Figure 1b and 1e show the 2D prediction heatmap generated after scanning hemorrhagic and ischemic stroke respectively, and Figure 1c and 1f show 3D reconstructed images of stroke location and subtype. Conclusion: We show that diagnosis of stroke may potentially be reduced from several hours to minutes, with additional spatial localization of intracranial hemorrhage, thereby rapidly guiding time-sensitive medical decisions for clinical intervention such as tPA. The sensor additionally detects ischemic and hemorrhagic lesions located deep inside the brain, and its range can be selectively tuned during sensor design and fabrication.