FixationFormer: Direct Utilization of Expert Gaze Trajectories for Chest X-Ray Classification
Daniel Beckmann, Benjamin Risse · 2026
Read FixationFormer: Direct Utilization of Expert Gaze Trajectories for Chest X-Ray Classification by Daniel Beckmann, Benjamin Risse (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-2603-22939.txt.
Description
Beckmann and Risse present FixationFormer, a transformer that uses expert gaze paths to classify chest X-rays. Eye movements hold rich diagnostic knowledge, but they are noisy and sparse in space. Most models flatten gaze into heatmaps, losing the time order. FixationFormer turns gaze paths into token sequences instead. Image and gaze tokens meet through cross-attention. Tests on three public chest X-ray datasets reach state-of-the-art accuracy. The work shows gaze sequences can guide medical image models directly.
Catalog record
| CREATOR | Daniel Beckmann, Benjamin Risse |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 32815 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2603.22939