Real-Time Polyp Detection in Colonoscopy Using YOLOv8: A Fast and Accurate Deep Learning Approach

Abstract

Polyps that may develop on the inner surfaces of the intestines or rectum are considered the primarily cause of colorectal cancer (CRC). To enhance survival rates, it is essential to focus on early detection, accurate prognosis, and timely treatment, typically involving surgical removal of polyps. The employment of advanced computer-aided diagnosis systems (CADx) that utilize appropriate machine learning techniques, particularly deep learning methods, aids physicians in achieving a highly relevant detection of abnormalities during internal examinations of the human body. In this context, this paper discusses a deep learning framework for automated polyp detection utilizing the you only look once (YOLO) model. This paper introduces a detection system based on the YOLOv8n model, designed for simplicity, effectiveness, cost-efficiency, and potential significant support for healthcare providers and patients in the realm of polyp detection. The results achieved are compared with those obtained using the YOLOv7 model and demonstrate enhanced performance.