Early Diagnosis of Stroke and Internal Hemorrhage via Deep Neural Network Inference of Microwave Signals

Abstract

A Compact automatic diagnostic system for internal hemorrhage is described. Utilizing microwave sensing and Deep Neural Network Inference, the technology detects brain stroke and internal abdominal hemorrhage. A sensitivity of 92.9% in detection of internal abdominal hemorrhage is demonstrated. A similar procedure leads to stroke detection with 90% sensitivity for hemorrhage larger than 0.5ml. Results establish the capability to accurately identify stroke and enable treatment with a compact autonomous system, outside the hospital environment.