Classification of current density vector map using transformer hybrid residual network
Lihui Zhu, Yunfeng Yang, Wenyue Yu, Qingxin Zeng, Zhan Zheng, Qiang Lin, Xiaohu Zhao, Zhenghui Hu, Agnese Sbrollini, Agnese Sbrollini, Agnese Sbrollini · 2025
Read Classification of current density vector map using transformer hybrid residual network by Lihui Zhu, Yunfeng Yang, Wenyue Yu, Qingxin Zeng, Zhan Zheng, Qiang Lin, Xiaohu Zhao, Zhenghui Hu, Agnese Sbrollini, Agnese Sbrollini, Agnese Sbrollini (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/pmc-12707687.txt.
Description
Lihui Zhu and co-authors teach a hybrid network to read heart-current maps that doctors find scarce and hard to judge, in this 2025 methods paper. The maps, called current density vector maps, are rebuilt from magnetocardiograms, recordings of the heart's magnetic field. Their patterns flag cardiac problems across five classes.
Real data are rare, so the team inflated their set with noise, model-based prediction, and interpolation. Their classifier merges a transformer, whose attention picks out telling features, with residual network blocks. Transfer learning set its starting weights. The system reached 97.52 percent accuracy, beating earlier deep-learning attempts. Because the design scales, the authors expect it to keep pace as clinical data sets grow.
Catalog record
| CREATOR | Lihui Zhu, Yunfeng Yang, Wenyue Yu, Qingxin Zeng, Zhan Zheng, Qiang Lin, Xiaohu Zhao, Zhenghui Hu, Agnese Sbrollini, Agnese Sbrollini, Agnese Sbrollini |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 63 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |