While mechanical thrombectomy is highly effective in revascularizing cerebral vessels after stroke, interventionalists are only able to achieve first-pass recanalization in less than one third of cases. We hypothesize that this limitation is in part due to the inability to localize the thrombus relative to interventional devices under angiography.
To augment the interventionalist awareness of clot location, we present a contact-detection method which provides audio feedback indicating to the interventionalist whether or not their catheter is in contact with the thrombus. This contact-detection method integrates into the existing clinical workflow by using standard, off-the-shelf aspiration catheters, which we equip with proximal pressure sensing and a motorized syringe. With this augmented catheter setup, we trained a support vector machine (SVM) classification model to detect whether or not a soft thrombus was embedded in the catheter tip based on proximal pressure readings.
This contact-detection method was evaluated in both a benchtop verification and user validation studies. Benchtop verification on a silicon phantom model indicated the efficacy (99.67% accuracy) of this approach subject to variabilities including vessel size, vessel tortuosity, and heart rate. The system was then deployed into operating room, to be used by five trained neurointerventionalists. Compared to the interventionalists estimate of thrombus location from angiographic presence alone, the sensing method was 2.86 times more likely to correctly detect contact. This shows that the proposed method improves expert interventionalists’ awareness of the thrombus location compared to existing clinical information. Beyond the scope of this study, we would like to determine whether this method may improve first-pass recanalization rates.