Polyp Detection in Colonoscopy Videos via Boosted Multi-Scale Feature Fusion

Computer-aided detection (CAD) systems are vital for advancing colorectal cancer screening and prevention through effective polyp detection approaches. Recent breakthroughs in deep learning technology have greatly enhanced medical image analysis and CAD systems for polyp detection. This paper proposes to use a boosted multi-scale feature fusion method for polyp detection in colonoscopy videos. The method is evaluated on a publicly available colonoscopy video dataset, in terms of metrics mAP, precision, recall and F1-score. In comparison with other state-of-the-art polyp detection methods, our method achieves better performance in both accuracy and efficiency, positioning it as a robust choice for real-time polyp detection in colonoscopy videos.