TransRUPNet for Improved Polyp Segmentation

TL;DR

TransRUPNet is a novel deep learning architecture that enables real-time, accurate polyp segmentation for early colorectal cancer detection, demonstrating high speed and generalizability across datasets.

Contribution

We introduce TransRUPNet, a transformer-based residual upsampling network that improves polyp segmentation accuracy and speed, especially on out-of-distribution datasets.

Findings

Achieves 47.07 fps with high accuracy on polyp datasets.

Outperforms existing methods on out-of-distribution data.

Provides real-time feedback for clinical applications.

Abstract

Colorectal cancer is among the most common cause of cancer worldwide. Removal of precancerous polyps through early detection is essential to prevent them from progressing to colon cancer. We develop an advanced deep learning-based architecture, Transformer based Residual Upsampling Network (TransRUPNet) for automatic and real-time polyp segmentation. The proposed architecture, TransRUPNet, is an encoder-decoder network consisting of three encoder and decoder blocks with additional upsampling blocks at the end of the network. With the image size of $256\times256$, the proposed method achieves an excellent real-time operation speed of 47.07 frames per second with an average mean dice coefficient score of 0.7786 and mean Intersection over Union of 0.7210 on the out-of-distribution polyp datasets. The results on the publicly available PolypGen dataset suggest that TransRUPNet can give real-time feedback while retaining high accuracy for in-distribution datasets. Furthermore, we demonstrate the generalizability of the proposed method by showing that it significantly improves performance on out-of-distribution datasets compared to the existing methods. The source code of our network is available at https://github.com/DebeshJha/TransRUPNet.