Prototype-Based Knowledge Guidance for Fine-Grained Structured Radiology Reporting

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Pellegrini, Chantal, Delchev, Adrian, Özsoy, Ege, Navab, Nassir, Keicher, Matthias
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908881948508160
author Pellegrini, Chantal
Delchev, Adrian
Özsoy, Ege
Navab, Nassir
Keicher, Matthias
author_facet Pellegrini, Chantal
Delchev, Adrian
Özsoy, Ege
Navab, Nassir
Keicher, Matthias
contents Structured radiology reporting promises faster, more consistent communication than free text, but automation remains difficult as models must make many fine-grained, discrete decisions about rare findings and attributes from limited structured supervision. In contrast, free-text reports are produced at scale in routine care and implicitly encode fine-grained, image-linked information through detailed descriptions. To leverage this unstructured knowledge, we propose ProtoSR, an approach for injecting free-text information into structured report population. First, we introduce an automatic extraction pipeline that uses an instruction-tuned LLM to mine 80k+ MIMIC-CXR studies and build a multimodal knowledge base aligned with a structured reporting template, representing each answer option with a visual prototype. Using this knowledge base, ProtoSR is trained to retrieve prototypes relevant for the current image-question pair and augment the model predictions through a prototype-conditioned residual, providing a data-driven second opinion that selectively corrects predictions. On the Rad-ReStruct benchmark, ProtoSR achieves state-of-the-art results, with the largest improvements on detailed attribute questions, demonstrating the value of integrating free-text derived signal for fine-grained image understanding.
format Preprint
id arxiv_https___arxiv_org_abs_2603_11938
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Prototype-Based Knowledge Guidance for Fine-Grained Structured Radiology Reporting
Pellegrini, Chantal
Delchev, Adrian
Özsoy, Ege
Navab, Nassir
Keicher, Matthias
Artificial Intelligence
Computer Vision and Pattern Recognition
Machine Learning
Structured radiology reporting promises faster, more consistent communication than free text, but automation remains difficult as models must make many fine-grained, discrete decisions about rare findings and attributes from limited structured supervision. In contrast, free-text reports are produced at scale in routine care and implicitly encode fine-grained, image-linked information through detailed descriptions. To leverage this unstructured knowledge, we propose ProtoSR, an approach for injecting free-text information into structured report population. First, we introduce an automatic extraction pipeline that uses an instruction-tuned LLM to mine 80k+ MIMIC-CXR studies and build a multimodal knowledge base aligned with a structured reporting template, representing each answer option with a visual prototype. Using this knowledge base, ProtoSR is trained to retrieve prototypes relevant for the current image-question pair and augment the model predictions through a prototype-conditioned residual, providing a data-driven second opinion that selectively corrects predictions. On the Rad-ReStruct benchmark, ProtoSR achieves state-of-the-art results, with the largest improvements on detailed attribute questions, demonstrating the value of integrating free-text derived signal for fine-grained image understanding.
title Prototype-Based Knowledge Guidance for Fine-Grained Structured Radiology Reporting
topic Artificial Intelligence
Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2603.11938