Colorectal polyp detection: Elevation of convolutional neural network-based models

Colorectal polyp detection plays a crucial role in the early diagnosis and prevention of colorectal cancer. This study evaluates multiple convolutional neural network (CNN) models for binary classification of polyp images using a public dataset of 3,000 images (1,500 polyp and 1,500 non-polyp). We compare deep learning architectures based on ResNet101, ResNet50, VGG16, VGG19, Xception, R-CNN, and training time based on accuracy, recall, precision, F1-score, and training time. The results indicate that both ResNet101 and our CNN model achieved the highest performance metrics, with ResNet101 reaching an accuracy of 99.96% and our CNN model achieving 100% accuracy. However, considering computational efficiency, our CNN model demonstrated the shortest training time (1,302 MS), making it the optimal solution for balancing performance and time efficiency. These findings highlight the potential of CNN-based models for real-time polyp detection, which can enhance early diagnosis and improve clinical outcomes.