TransRUPNet for Improved Polyp Segmentation

TL;DR

The paper tackles real-time polyp segmentation with robust out-of-distribution generalization for colonoscopy. It introduces TransRUPNet, an encoder–decoder network that uses a Pyramid Vision Transformer encoder and residual upsampling to produce precise segmentation masks at $256\times256$ resolution with real-time performance (~$47.07$ FPS). On in-distribution data (Kvasir-SEG) it achieves high metrics (e.g., Dice $=0.9005$, mIoU $=0.8445$) and outperforms SOTA methods on several OOD datasets (e.g., PolypGen, BKAI-IGH, CVC-ClinicDB) with substantial gains in mIoU and mDSC. The study demonstrates strong cross-domain generalization and real-time applicability for clinical CAD systems, and provides public code to facilitate adoption and further research.

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.

Figures (2)

• Figure 1: Overall architecture of the TransRUPNet.
• Figure 2: Qualitative example showing polyp segmentation