AI in Gastrointestinal Endoscopy: Enhancing Accuracy and Speed in Disease Detection

Authors : Masrufa Akter Muni, Mustafizur Rahman, Rahat Hasan, Md Mohaimin Rashid, Rakibul Islam, Khandakar Rabbi Ahmed

Publication date : 2025/11/20

Conference : 2025 9th International Conference on Computational System and Information Technology for Sustainable Solutions (CSITSS)

Pages : 1-6

Description :  The use of artificial intelligence (AI) in gastrointestinal (GI) endoscopy has demonstrated enormous promise for increasing procedural efficiency and diagnostic precision. In order to increase polyp segmentation performance, this research suggests PolypTransNet, a deep learning (DL) technique that combines the power of Transformer-based attention mechanisms and convolutional neural networks (CNNs). With the Kvasir-SEG dataset, the model achieves a Dice accuracy of 91.42%, an IoU of 84.88%, and an inference speed of 22 FPS, surpassing current benchmarks such as U-Net, ResUNet++, and PraNet. The network incorporates EfficientN etBO as a lightweight encoder and self-attention blocks for global context modeling, which facilitates both precise boundary definition and real-time deployment through ONNX optimization. Experimental outcomes validate that PolypTransNet presents a practical and …