LGPS: A Lightweight GAN-Based Approach for Polyp Segmentation in Colonoscopy Images

2025-03-24

AI Summary

To address clinical challenges in colonoscopy—namely, low segmentation accuracy for small or low-contrast polyps, high computational overhead, and poor cross-dataset generalization—this paper proposes a lightweight, efficient, and robust polyp segmentation framework. Methodologically, we design a MobileNetV2+ResE backbone that integrates an improved residual block with Squeeze-and-Excitation (SE) modules; incorporate ConvCRF for boundary refinement; and formulate a triple hybrid loss comprising Binary Cross-Entropy, Weighted IoU, and Dice losses. Evaluated on the PolypGen test set, our model achieves a Dice score of 0.7299 and IoU of 0.7867 using only 1.07M parameters—just 1/17 the size of the current smallest state-of-the-art (SOTA) model. It consistently outperforms existing methods, significantly enhancing both early detection utility and real-world deployment feasibility.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Learning on the Edge & Model Compression
• Search and Optimization: Learning to Search

Application Category

• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Responsible Web: Consent frameworks and practices on the web

Abstract

Colorectal cancer (CRC) is a major global cause of cancer-related deaths, with early polyp detection and removal during colonoscopy being crucial for prevention. While deep learning methods have shown promise in polyp segmentation, challenges such as high computational costs, difficulty in segmenting small or low-contrast polyps, and limited generalizability across datasets persist. To address these issues, we propose LGPS, a lightweight GAN-based framework for polyp segmentation. LGPS incorporates three key innovations: (1) a MobileNetV2 backbone enhanced with modified residual blocks and Squeeze-and-Excitation (ResE) modules for efficient feature extraction; (2) Convolutional Conditional Random Fields (ConvCRF) for precise boundary refinement; and (3) a hybrid loss function combining Binary Cross-Entropy, Weighted IoU Loss, and Dice Loss to address class imbalance and enhance segmentation accuracy. LGPS is validated on five benchmark datasets and compared with state-of-the-art(SOTA) methods. On the largest and challenging PolypGen test dataset, LGPS achieves a Dice of 0.7299 and an IoU of 0.7867, outperformed all SOTA works and demonstrating robust generalization. With only 1.07 million parameters, LGPS is 17 times smaller than the smallest existing model, making it highly suitable for real-time clinical applications. Its lightweight design and strong performance underscore its potential for improving early CRC diagnosis. Code is available at https://github.com/Falmi/LGPS/.

Problem

• Lightweight GAN for polyp segmentation in colonoscopy images
• Improves small and low-contrast polyp detection accuracy
• Reduces computational cost for real-time clinical use

Innovation

• MobileNetV2 backbone with ResE modules
• Convolutional Conditional Random Fields refinement
• Hybrid loss function for class imbalance