TopoImages: Incorporating Local Topology Encoding into Deep Learning Models for Medical Image Classification Conference: Proceedings of the 33rd ACM International Conference on Multimedia; October 27–31, 2025; Dublin, Ireland Proceedings of the 33rd ACM International Conference on Multimedia (MM ’25
Pengfei Gu, Hongxiao Wang, Yejia Zhang, Huimin Li, Chaoli Wang, Danny Chen · 2025
Read TopoImages: Incorporating Local Topology Encoding into Deep Learning Models for Medical Image Classification Conference: Proceedings of the 33rd ACM International Conference on Multimedia; October 27–31, 2025; Dublin, Ireland Proceedings of the 33rd ACM International Conference on Multimedia (MM ’25 by Pengfei Gu, Hongxiao Wang, Yejia Zhang, Huimin Li, Chaoli Wang, Danny Chen (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-2508-01574.txt.
Description
Pengfei Gu and five co-authors feed deep networks, in a 2025 preprint, a new kind of image that encodes the topology of the original. Loops, holes, and connected pieces in a scan carry diagnostic meaning that pixel patterns can miss. Standard networks are nearly blind to such structure. The authors compute, for each small patch of an image, a persistence diagram summarizing its topological features. Those diagrams are flattened and laid back over the patch as extra channels. The result is a TopoImage, a picture of the picture's shape.
Several viewing filters produce several TopoImages, which are fused with the scan itself. The add-on drops into common network designs unchanged. On three public medical classification sets, accuracy rose clearly over leading methods. The authors pitch topology as a missing sense for medical AI. Scans of tissue, where loops and gaps are the story, stand to gain most.
Catalog record
| CREATOR | Pengfei Gu, Hongxiao Wang, Yejia Zhang, Huimin Li, Chaoli Wang, Danny Chen |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 45 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-01574