Attention GhostUNet++: Enhanced Segmentation of Adipose Tissue and Liver in CT Images
Mansoor Hayat, Supavadee Aramvith, Subrata Bhattacharjee, Nouman Ahmad · 2025
Read Attention GhostUNet++: Enhanced Segmentation of Adipose Tissue and Liver in CT Images by Mansoor Hayat, Supavadee Aramvith, Subrata Bhattacharjee, Nouman Ahmad (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-11491.txt.
Description
Doctors read body fat and liver outlines on CT scans to judge risks such as diabetes and heart disease, but tracing those outlines by hand is slow. Mansoor Hayat, Supavadee Aramvith, Subrata Bhattacharjee, and Nouman Ahmad present a deep learning model for that tracing task in this 2025 paper. Their model, Attention GhostUNet++, adds attention steps that help it focus on the right features.
The model was tested on two public scan sets, one for belly fat and one for liver tumors. Its overlap scores reached about 0.94 for fat under the skin, 0.96 for fat around the organs, and 0.97 for the liver, beating the base models it was compared with. The authors note weaker detail at borders, but stress the gains in accuracy and speed. Code for the model is shared in public.
Catalog record
| CREATOR | Mansoor Hayat, Supavadee Aramvith, Subrata Bhattacharjee, Nouman Ahmad |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 13 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-11491