A World Model of Radiologist Reading for Medical Image Representation Learning
GazeWorld is a medical imaging world model that treats the image as the world and radiologist fixation sequences as trajectories. It autoregressively predicts latent representations of fixated patches while using a spatial-completion branch for unvisited regions. At inference, it generates patch representations from the image alone without real gaze data. Frozen GazeWorld features achieve state-of-the-art diagnostic accuracy on all nine supervised settings across CheXpert, RSNA Pneumonia, and SIIM-ACR Pneumothorax, and highest zero-shot accuracy on all three benchmarks. On GazeSearch, a generic decoder trained on the same frozen features outperforms the purpose-built LogitGaze-Med by over 16% in ScanMatch and 22% in SED. The work demonstrates that modeling how experts read, not just their conclusions, offers a promising pretraining paradigm for medical imaging AI.
Article intelligence
Key points
- GazeWorld leverages radiologist eye-tracking data as reading trajectories for autoregressive prediction and spatial completion.
- It requires no real gaze data at inference, generating patch sequences from images alone.
- Frozen features achieve top diagnostic and zero-shot accuracy on multiple medical imaging benchmarks.
- On GazeSearch, a generic decoder significantly outperforms the specialized LogitGaze-Med model.
Why it matters
This matters because gazeWorld leverages radiologist eye-tracking data as reading trajectories for autoregressive prediction and spatial completion.
Technical impact
May affect model selection, inference cost, product capability, and evaluation benchmarks.
[2605.23992] A World Model of Radiologist Reading for Medical Image Representation Learning
[Submitted on 17 May 2026]
Title:A World Model of Radiologist Reading for Medical Image Representation Learning
View a PDF of the paper titled A World Model of Radiologist Reading for Medical Image Representation Learning, by Yiwei Li and 7 other authors
View PDF HTML (experimental)
Abstract:Radiologist eye-tracking data provide a rich record of how experts search, compare, and accumulate evidence during image reading; yet, existing methods exploit this signal only partially, either as a static spatial prior or as an auxiliary prediction target decoupled from diagnosis. We propose GazeWorld, a medical imaging world model that treats the image as the world and the radiologist's fixation sequence as a trajectory through it. GazeWorld autoregressively predicts the latent representation of the next fixated patch from all previously visited ones, while a spatial-completion branch covers unvisited regions. At inference, GazeWorld generates a sequence of patch representations from the image alone without requiring real gaze data. Frozen GazeWorld features achieve state-of-the-art diagnostic accuracy across all nine supervised settings on CheXpert, RSNA Pneumonia, and SIIM-ACR Pneumothorax, as well as the highest zero-shot accuracy on all three benchmarks. On the GazeSearch benchmark, a generic decoder trained on the same frozen features outperforms the purpose-built LogitGaze-Med by over 16\% in ScanMatch and 22\% in SED, despite not being explicitly trained to predict gaze. GazeWorld demonstrates that modeling how experts read, not just what they conclude, offers a promising pretraining paradigm for medical imaging AI.
Subjects:
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
Cite as: arXiv:2605.23992 [cs.CV]
(or arXiv:2605.23992v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2605.23992
arXiv-issued DOI via DataCite (pending registration)
Submission history
From: Yiwei Li [view email] [v1] Sun, 17 May 2026 22:30:29 UTC (8,920 KB)
Full-text links:
Access Paper:
View a PDF of the paper titled A World Model of Radiologist Reading for Medical Image Representation Learning, by Yiwei Li and 7 other authors
View PDF
HTML (experimental)
TeX Source
view license
Current browse context:
cs.CV
new | recent | 2026-05
Change to browse by:
cs cs.AI
References & Citations
NASA ADS
Google Scholar
Semantic Scholar
Loading...
Data provided by:
Bibliographic Tools
Bibliographic and Citation Tools
Bibliographic Explorer Toggle
Bibliographic Explorer (What is the Explorer?)
Connected Papers Toggle
Connected Papers (What is Connected Papers?)
Litmaps Toggle
Litmaps (What is Litmaps?)
scite.ai Toggle
scite Smart Citations (What are Smart Citations?)
Code, Data, Media
Code, Data and Media Associated with this Article
alphaXiv Toggle
alphaXiv (What is alphaXiv?)
Links to Code Toggle
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub Toggle
DagsHub (What is DagsHub?)
GotitPub Toggle
Gotit.pub (What is GotitPub?)
Huggingface Toggle
Hugging Face (What is Huggingface?)
ScienceCast Toggle
ScienceCast (What is ScienceCast?)
Demos
Demos
Replicate Toggle
Replicate (What is Replicate?)
Spaces Toggle
Hugging Face Spaces (What is Spaces?)
Spaces Toggle
TXYZ.AI (What is TXYZ.AI?)
Related Papers
Recommenders and Search Tools
Link to Influence Flower
Influence Flower (What are Influence Flowers?)
Core recommender toggle
CORE Recommender (What is CORE?)
Author
Venue
Institution
Topic
About arXivLabs
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)