Research on Real-time Prostate Encapsulation Detection Method

Benign prostatic hyperplasia is a 1 common disease in middle-aged and elderly men. Its main histological manifestations include prostate stromal and glandular hyperplasia, bladder outlet obstruction, etc. Manual detection of prostate envelope can improve the success rate of surgery, ensure the safety of patients, reduce certain medical costs and promote the combination of artificial intelligence and medical treatment. For patient images, the target detection method, SSD method, FSSD method, FoveaBox method, TOOD method and Sparse R-CNN method are used for real-time detection, the real-time image of the endoscope of the host is calibrated and the position warning is framed, and the algorithm is evaluated by mAP accuracy evaluation index and FPS speed evaluation index.