G-LoG Bi-filtration for Medical Image Classification
Qingsong Wang, Jiaxing He, Bingzhe Hou, Tieru Wu, Yang Cao, Cailing Yao · 2026
Read G-LoG Bi-filtration for Medical Image Classification by Qingsong Wang, Jiaxing He, Bingzhe Hou, Tieru Wu, Yang Cao, Cailing Yao (2026) 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-2602-18329.txt.
Description
Qingsong Wang and five co-authors build a two-parameter topological filter for medical images. Their 2026 paper uses the Laplacian of Gaussian operator, which sharpens image boundaries, to define the G-LoG bi-filtration for multi-parameter persistence. Volumetric images are modeled as bounded functions, and the team proves the resulting persistence modules stay stable under the maximum norm.
On the MedMNIST dataset, the bi-filtration beats single-parameter filtration clearly. Strikingly, a simple multi-layer perceptron trained only on the topological features matches complex deep learning models trained on raw images, including ResNet, Google AutoML Vision, auto-sklearn, and AutoKeras. The work shows well-built topological features can rival heavy models in medical image classification.
Catalog record
| CREATOR | Qingsong Wang, Jiaxing He, Bingzhe Hou, Tieru Wu, Yang Cao, Cailing Yao |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 40 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-18329