@inproceedings{sim-etal-2026-learning,
title = "Learning Pathology-Localized Medical Vision{--}Language Representations with Atomic and Grounded Disease Supervision",
author = "Sim, Mong Yuan and
Zhang, Wei Emma and
Dai, Xiang and
Fang, Byron and
Liu, Miao",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.2309/",
pages = "49550--49564",
ISBN = "979-8-89176-462-0",
abstract = "Medical vision-language models are typically trained by aligning whole images with entire radiology reports, assuming that global correspondence is sufficient for learning clinically meaningful representations. We argue that effective medical representation learning requires pathology-localized supervision, which explicitly aligns specific disease observations with their corresponding anatomical image regions. To enable this at scale, we propose a fully automated pipeline that decomposes radiology reports into atomic disease observations, identifies relevant regions, and pairs them with corresponding image regions without manual annotation, resulting in a fine-grained dataset, LocalizedCXR, with over 6 million region{--}caption pairs. Through extensive experiments, we demonstrate that our pathology-localized semantic supervision significantly improve performance in cross-modal retrieval, image classification, and phrase grounding of radiology findings. Our results suggest that moving beyond global alignment to disease-aware region-level alignment is essential for developing effective medical vision-language representation learning."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="sim-etal-2026-learning">
<titleInfo>
<title>Learning Pathology-Localized Medical Vision–Language Representations with Atomic and Grounded Disease Supervision</title>
</titleInfo>
<name type="personal">
<namePart type="given">Mong</namePart>
<namePart type="given">Yuan</namePart>
<namePart type="family">Sim</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wei</namePart>
<namePart type="given">Emma</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Xiang</namePart>
<namePart type="family">Dai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Byron</namePart>
<namePart type="family">Fang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Miao</namePart>
<namePart type="family">Liu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Findings of the Association for Computational Linguistics: EMNLP 2026</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>Medical vision-language models are typically trained by aligning whole images with entire radiology reports, assuming that global correspondence is sufficient for learning clinically meaningful representations. We argue that effective medical representation learning requires pathology-localized supervision, which explicitly aligns specific disease observations with their corresponding anatomical image regions. To enable this at scale, we propose a fully automated pipeline that decomposes radiology reports into atomic disease observations, identifies relevant regions, and pairs them with corresponding image regions without manual annotation, resulting in a fine-grained dataset, LocalizedCXR, with over 6 million region–caption pairs. Through extensive experiments, we demonstrate that our pathology-localized semantic supervision significantly improve performance in cross-modal retrieval, image classification, and phrase grounding of radiology findings. Our results suggest that moving beyond global alignment to disease-aware region-level alignment is essential for developing effective medical vision-language representation learning.</abstract>
<identifier type="citekey">sim-etal-2026-learning</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.2309/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>49550</start>
<end>49564</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Learning Pathology-Localized Medical Vision–Language Representations with Atomic and Grounded Disease Supervision
%A Sim, Mong Yuan
%A Zhang, Wei Emma
%A Dai, Xiang
%A Fang, Byron
%A Liu, Miao
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F sim-etal-2026-learning
%X Medical vision-language models are typically trained by aligning whole images with entire radiology reports, assuming that global correspondence is sufficient for learning clinically meaningful representations. We argue that effective medical representation learning requires pathology-localized supervision, which explicitly aligns specific disease observations with their corresponding anatomical image regions. To enable this at scale, we propose a fully automated pipeline that decomposes radiology reports into atomic disease observations, identifies relevant regions, and pairs them with corresponding image regions without manual annotation, resulting in a fine-grained dataset, LocalizedCXR, with over 6 million region–caption pairs. Through extensive experiments, we demonstrate that our pathology-localized semantic supervision significantly improve performance in cross-modal retrieval, image classification, and phrase grounding of radiology findings. Our results suggest that moving beyond global alignment to disease-aware region-level alignment is essential for developing effective medical vision-language representation learning.
%U https://aclanthology.org/2026.findings-emnlp.2309/
%P 49550-49564
Markdown (Informal)
[Learning Pathology-Localized Medical Vision–Language Representations with Atomic and Grounded Disease Supervision](https://aclanthology.org/2026.findings-emnlp.2309/) (Sim et al., Findings 2026)
ACL