Endoscopic colorectal polyp detection based on improved YOLOv8

Jincai Huang, Jianyuan Zeng, Jinfeng Peng, Mengye Lyu, Shaojun Liu

Abstract: Colorectal tumors typically arise from colorectal polyps and are commonly identified through colonoscopy during clinical examinations. Manual detection of polyps, crucial for distinguishing between benign and malignant cases, suffers from a limited detection rate of approximately 25%, influenced by subjective factors and other variables. This study leverages advancements in deep learning and computer detection technology to explore the application of artificial intelligence in polyp detection. Specifically, we employ an improved YOLOv8 model for polyp detection to boost the detection capabilities. The proposed method demonstrates superior accuracy and efficiency compared to existing artificial intelligence approaches.

External IDs:dblp:conf/embc/0002ZPLL24