An Automated Ptosis Screening and Postoperative Prediction System

This study proposes a system for screening ptosis, predicting upper eyelid muscle strength, and forecasting postoperative outcomes based on facial images. In the screening of ptosis and prediction of upper eyelid muscle strength, The system can quantify eye-related features into feature values, which can serve as diagnostic aids for doctors. In clinical practice, we found that some patients with ptosis unconsciously use the frontalis muscle to lift the upper eyelid due to the obstruction of their visual field, which is manifested as raising the eyebrows. Some patients are accustomed to looking down or up. The existing ptosis screening models all default to input photos taken by patients in a natural state with relaxed eye muscles and eyes looking straight ahead, but this is not the case in clinical practice. To solve this problem, we added a pre-detection module to the system, which will issue a warning when detecting that the patient's eyes are not looking straight ahead or have a raised eyebrow action, thereby improving the accuracy of the system's ptosis screening and muscle strength prediction in clinical applications.