A computer-aided system for ocular myasthenia gravis diagnosis

The current mode of clinical aided diagnosis Ocular Myasthenia Gravis (OMG) is time-consuming and laborious, it lacks quantitative standards. An diagnostic system for OMG proposed to solve this problem. values calculated by the include three indicators: eyelid distance, sclera palpebra superior fatigability test time. For first two indicators, semantic segmentation method was used extract pathological features patient's eye image a model constructed. patient divided into regions: iris, sclera, background. indicators were based on position pixels in mask. last indicator, calculation Eyelid Aspect Ratio (EAR) proposed; can better reflect change distance overtime. evaluated collected data. results show that achieves mean Intersection-Over-Union (mIoU) value 86.05%. paired-sample T-test compare obtained doctors, p all greater than 0.05. Thus, reduce cost has high application value.