Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

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

Fiseha B. Tesema, Alejandro Guerra Manzanares, Tianxiang Cui, Qian Zhang, Moses Solomon, Sean He

TL;DR

LGPS introduces a lightweight GAN-based polyp segmentation framework that combines a MobileNetV2–based generator with ReSE blocks, a ConvCRF-refined discriminator, and a hybrid BCE+WIoU+Dice loss to address boundary accuracy and class imbalance. With only 1.07 million parameters, LGPS achieves competitive Dice and IoU across multiple public datasets and demonstrates exceptional generalization on unseen data, including the PolypGen set. The approach emphasizes real-time clinical applicability through efficiency, boundary refinement, and robust generalization, marking a notable step toward practical CAD systems for CRC screening. Future work may explore integration with transformer backbones or broader medical imaging tasks to further enhance performance and deployment feasibility.

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/.

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

TL;DR

LGPS introduces a lightweight GAN-based polyp segmentation framework that combines a MobileNetV2–based generator with ReSE blocks, a ConvCRF-refined discriminator, and a hybrid BCE+WIoU+Dice loss to address boundary accuracy and class imbalance. With only 1.07 million parameters, LGPS achieves competitive Dice and IoU across multiple public datasets and demonstrates exceptional generalization on unseen data, including the PolypGen set. The approach emphasizes real-time clinical applicability through efficiency, boundary refinement, and robust generalization, marking a notable step toward practical CAD systems for CRC screening. Future work may explore integration with transformer backbones or broader medical imaging tasks to further enhance performance and deployment feasibility.

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/.

Paper Structure

This paper contains 33 sections, 21 equations, 6 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. U-Net Architectures
  4. Attention Mechanisms and Feature Aggregation
  5. Transformer-Based Approaches
  6. Lightweight Models for Real-Time Deployment
  7. Proposed Model Architecture
  8. Generator Architecture
  9. Encoder
  10. Modified Residual with Squeeze-and-Excitation block (ReSE)
  11. Decoder
  12. Discriminator Architecture
  13. Adversarial Training and Loss Functions
  14. Binary Crossentropy Loss (BCE)
  15. Weighted IoU(WIoU) Loss
  16. ...and 18 more sections

Figures (6)

  • Figure 1: Comparison of model size and performance. The area of the circles relates to the size of the model in terms of the number of parameters, while the left axis reports the IoU value of each model on the CVC-ClinicDB dataset. The proposed model, LGPS, outperforms all state-of-the-art models with 17 times fewer parameters.
  • Figure 2: a) Overview of the LGPS architecture, showing the generator and discriminator components.
  • Figure 3: a) ReSE block b) SE block
  • Figure 4: Visualized heat maps (a) with ConvCRF and ReSE and (b) without ConvCRF and ReSE
  • Figure 5: Ablation experiment on different loss functions.
  • ...and 1 more figures