Attention GhostUNet++: Enhanced Segmentation of Adipose Tissue and Liver in CT Images

2025-04-14

AI Summary

This study addresses the challenges of simultaneous segmentation of subcutaneous adipose tissue (SAT), visceral adipose tissue (VAT), and liver in abdominal CT images—namely, ambiguous tissue boundaries, inter-organ anatomical coupling, and excessive computational overhead. To this end, we propose Ghost UNet++, a lightweight yet high-performance architecture. Its core innovation is a novel Ghost bottleneck module integrating channel-, spatial-, and depth-wise attention mechanisms, enabling significant parameter reduction while enhancing discriminative feature learning and multi-scale contextual modeling. Coupled with UNet++’s dense skip connections, the design optimizes cross-level feature reuse and propagation. Evaluated on the AATTCT-IDS and LiTS datasets, Ghost UNet++ achieves Dice scores of 0.9639 (SAT), 0.9430 (VAT), and 0.9652 (liver), consistently outperforming state-of-the-art baselines. The method delivers both high accuracy and low inference cost, offering a practical solution for body composition analysis and metabolic disease risk assessment.

Technology Category

Computer Vision: SegmentationMachine Learning: Large Multimodal Models (LMMs)Constraint Satisfaction and Optimization: Satisfiability Modulo Theories

Application Category

Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphsWeb Mining and Content Analysis: Large pretrained models with web dataSystems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Accurate segmentation of abdominal adipose tissue, including subcutaneous (SAT) and visceral adipose tissue (VAT), along with liver segmentation, is essential for understanding body composition and associated health risks such as type 2 diabetes and cardiovascular disease. This study proposes Attention GhostUNet++, a novel deep learning model incorporating Channel, Spatial, and Depth Attention mechanisms into the Ghost UNet++ bottleneck for automated, precise segmentation. Evaluated on the AATTCT-IDS and LiTS datasets, the model achieved Dice coefficients of 0.9430 for VAT, 0.9639 for SAT, and 0.9652 for liver segmentation, surpassing baseline models. Despite minor limitations in boundary detail segmentation, the proposed model significantly enhances feature refinement, contextual understanding, and computational efficiency, offering a robust solution for body composition analysis. The implementation of the proposed Attention GhostUNet++ model is available at:https://github.com/MansoorHayat777/Attention-GhostUNetPlusPlus.

Problem

• Accurate segmentation of abdominal adipose tissue and liver in CT images
• Improved feature refinement and contextual understanding for body composition analysis
• Enhanced computational efficiency in deep learning-based medical image segmentation

Innovation

• Channel, Spatial, Depth Attention mechanisms
• Ghost UNet++ bottleneck enhancement
• High Dice coefficients for segmentation