Colorectal polyp detection based on deep learning approaches

Category: AI

Exhibitor: MING CHI UNIVERSITY OF TECHNOLGY

Booth No: N432

Characteristic

According to recent studies, colorectal cancer is one of the leading causes of death these days. Colorectal polyps are irregular formation of tissues that lead to colorectal cancer. Conventional methods are not sufficient to properly detect colorectal cancer in the early stages. Therefore, an efficient method is inevitable to detect and classify colorectal polyps in the early stage. In recent years, deep learning-based methods have been proposed, but there is still a need for improvement to make these methods more efficient. In this study, we have compared the latest methods based on their performance to propose an efficient model for real-time polyp detection. The comparison is made between deep learning-based four computation algorithms, FasterRCNN, SSD, YOLOv3, and YOLOv4, using a custom colorectal dataset from a hospital. Similarly, data augmentation is also applied to analyze the performance of the mentioned techniques. We expect this study to provide insight into selecting an efficient method that would play a significant role in real-time polyp detection and classification to ensure invaluable medical aid.