Enregistré dans:
| Auteurs principaux: | , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | https://arxiv.org/abs/2605.16572 |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866916017961172992 |
|---|---|
| author | Elbatel, Marawan Ghonim, Mohamed Mao, Jiaji Lin, Zhuosheng Eckstein, Katharina Mora, Andrés Martínez Deissler, Jonathan Rokuss, Maximilian Ulrich, Constantin Marinov, Zdravko Deng, Wenhui Li, Baoxun Hu, Huijun Shen, Jun Ghonim, Mohanad Nassar, Khadiga Omar Elbakry, Mariam Dyab, Menna Salem, Amr Muhammad Abdo Elghitany, Nouran Elghitany, Noha Qin, Yi Huang, Xuanqi Wang, Haonan Yen, Shao-Woo Saba, Ahmed Elghamry Ahmad, Salma Fang, Xinyan Zhang, Jiahao Wang, Xiaodi Ma, Xinghua Luo, Gongning Delmoral, Jessica C. Tavares, João Manuel R. S. Deria, Ankan Dukre, Adinath Xie, Yutong Razzak, Imran Kim, Dongwook Choi, Matthew Zhang, Hanxiao Zhang, Minghui You, Xin Qayyum, Abdul Niederer, Steven A. Mazher, Moona Hamadache, Rachika E. Montoya-del-Angel, Ricardo Martí, Robert Lladó, Xavier Musah, Toufiq Ayivor, Livingstone Eli Almar-Munoz, Enrique Mayr, Agnes Mouheb, Kaouther Bron, Esther E. Klein, Stefan Abouelhoda, Ahmed Adel, Amira Ali, Susan Adil Stiefelhagen, Rainer Maier-Hein, Klaus H. Isensee, Fabian Yassin, Aya Li, Xiaomeng |
| author_facet | Elbatel, Marawan Ghonim, Mohamed Mao, Jiaji Lin, Zhuosheng Eckstein, Katharina Mora, Andrés Martínez Deissler, Jonathan Rokuss, Maximilian Ulrich, Constantin Marinov, Zdravko Deng, Wenhui Li, Baoxun Hu, Huijun Shen, Jun Ghonim, Mohanad Nassar, Khadiga Omar Elbakry, Mariam Dyab, Menna Salem, Amr Muhammad Abdo Elghitany, Nouran Elghitany, Noha Qin, Yi Huang, Xuanqi Wang, Haonan Yen, Shao-Woo Saba, Ahmed Elghamry Ahmad, Salma Fang, Xinyan Zhang, Jiahao Wang, Xiaodi Ma, Xinghua Luo, Gongning Delmoral, Jessica C. Tavares, João Manuel R. S. Deria, Ankan Dukre, Adinath Xie, Yutong Razzak, Imran Kim, Dongwook Choi, Matthew Zhang, Hanxiao Zhang, Minghui You, Xin Qayyum, Abdul Niederer, Steven A. Mazher, Moona Hamadache, Rachika E. Montoya-del-Angel, Ricardo Martí, Robert Lladó, Xavier Musah, Toufiq Ayivor, Livingstone Eli Almar-Munoz, Enrique Mayr, Agnes Mouheb, Kaouther Bron, Esther E. Klein, Stefan Abouelhoda, Ahmed Adel, Amira Ali, Susan Adil Stiefelhagen, Rainer Maier-Hein, Klaus H. Isensee, Fabian Yassin, Aya Li, Xiaomeng |
| contents | Automated segmentation of liver lesions on non-contrast computed tomography (NCCT) is clinically important but fundamentally challenging, particularly in low-resource settings across Africa and Asia where contrast agents are frequently unavailable. Progress has been limited by the absence of annotated NCCT benchmarks. Here we describe the TriALS challenge for automated liver lesion segmentation under contrast-limited conditions, supported by a multi-centre dataset of 150 cases with four-phase CT acquisitions (600 volumes) from Egyptian and Chinese institutions. Algorithms were evaluated on 70 cases from three institutions, including an independent external cohort. The top-performing method achieved a mean venous-phase Dice of 0.754, consistent with human-level performance, yet dropped to 0.57 on NCCT. On external validation, the leading method outperformed off-the-shelf models by up to 28% in Dice on NCCT. Algorithm performance was most strongly predicted by training data scale and pre-training strategy. A cross-year comparison exposed a persistent perceptual barrier on NCCT that scaling pre-training alone cannot overcome. Data, annotations, and code are available at https://github.com/xmed-lab/TriALS. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_16572 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | TriALS: Triphasic-Aided Liver Lesion Segmentation Benchmark in Non-Contrast CT Elbatel, Marawan Ghonim, Mohamed Mao, Jiaji Lin, Zhuosheng Eckstein, Katharina Mora, Andrés Martínez Deissler, Jonathan Rokuss, Maximilian Ulrich, Constantin Marinov, Zdravko Deng, Wenhui Li, Baoxun Hu, Huijun Shen, Jun Ghonim, Mohanad Nassar, Khadiga Omar Elbakry, Mariam Dyab, Menna Salem, Amr Muhammad Abdo Elghitany, Nouran Elghitany, Noha Qin, Yi Huang, Xuanqi Wang, Haonan Yen, Shao-Woo Saba, Ahmed Elghamry Ahmad, Salma Fang, Xinyan Zhang, Jiahao Wang, Xiaodi Ma, Xinghua Luo, Gongning Delmoral, Jessica C. Tavares, João Manuel R. S. Deria, Ankan Dukre, Adinath Xie, Yutong Razzak, Imran Kim, Dongwook Choi, Matthew Zhang, Hanxiao Zhang, Minghui You, Xin Qayyum, Abdul Niederer, Steven A. Mazher, Moona Hamadache, Rachika E. Montoya-del-Angel, Ricardo Martí, Robert Lladó, Xavier Musah, Toufiq Ayivor, Livingstone Eli Almar-Munoz, Enrique Mayr, Agnes Mouheb, Kaouther Bron, Esther E. Klein, Stefan Abouelhoda, Ahmed Adel, Amira Ali, Susan Adil Stiefelhagen, Rainer Maier-Hein, Klaus H. Isensee, Fabian Yassin, Aya Li, Xiaomeng Computer Vision and Pattern Recognition Automated segmentation of liver lesions on non-contrast computed tomography (NCCT) is clinically important but fundamentally challenging, particularly in low-resource settings across Africa and Asia where contrast agents are frequently unavailable. Progress has been limited by the absence of annotated NCCT benchmarks. Here we describe the TriALS challenge for automated liver lesion segmentation under contrast-limited conditions, supported by a multi-centre dataset of 150 cases with four-phase CT acquisitions (600 volumes) from Egyptian and Chinese institutions. Algorithms were evaluated on 70 cases from three institutions, including an independent external cohort. The top-performing method achieved a mean venous-phase Dice of 0.754, consistent with human-level performance, yet dropped to 0.57 on NCCT. On external validation, the leading method outperformed off-the-shelf models by up to 28% in Dice on NCCT. Algorithm performance was most strongly predicted by training data scale and pre-training strategy. A cross-year comparison exposed a persistent perceptual barrier on NCCT that scaling pre-training alone cannot overcome. Data, annotations, and code are available at https://github.com/xmed-lab/TriALS. |
| title | TriALS: Triphasic-Aided Liver Lesion Segmentation Benchmark in Non-Contrast CT |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2605.16572 |