Health AI Central
Peer-Reviewed Publication
Eur Radiol2026;36(7):6107-6118.July 1, 2026Journal Article

Hallucination filtering in radiology vision-language models using discrete semantic entropy.

Patrick Wienholt1,2, Sophie Caselitz3,4, Robert Siepmann3,4, Philipp Bruners4, Keno Bressem5,6, Christiane Kuhl4, Jakob Nikolas Kather7,8,9,10, Sven Nebelung3,4, Daniel Truhn3,4
1Lab for Artificial Intelligence in Medicine, Department of Diagnostic and Interventional Radiology, University Hospital RWTH Aachen, Aachen, Germany. pwienholt@ukaachen.de.
2Department of Diagnostic and Interventional Radiology, University Hospital RWTH Aachen, Aachen, Germany. pwienholt@ukaachen.de.
3Lab for Artificial Intelligence in Medicine, Department of Diagnostic and Interventional Radiology, University Hospital RWTH Aachen, Aachen, Germany.
4Department of Diagnostic and Interventional Radiology, University Hospital RWTH Aachen, Aachen, Germany.
5Department of Diagnostic and Interventional Radiology, Technical University of Munich, School of Medicine and Health, Klinikum rechts der Isar, TUM University Hospital, Munich, Germany.
6Department of Cardiovascular Radiology and Nuclear Medicine, Technical University of Munich, School of Medicine and Health, German Heart Center, TUM University Hospital, Munich, Germany.
7Else Kroener Fresenius Center for Digital Health, Faculty of Medicine and University Hospital Carl Gustav Carus, TUD Dresden University of Technology, Dresden, Germany.
8Department of Medicine I, Faculty of Medicine and University Hospital Carl Gustav Carus, TUD Dresden University of Technology, Dresden, Germany.
9Pathology & Data Analytics, Leeds Institute of Medical Research at St James's, University of Leeds, Leeds, United Kingdom.
10Medical Oncology, National Center for Tumor Diseases (NCT), University Hospital Heidelberg, Heidelberg, Germany.
PubMed PMC DOI PDF

Abstract

OBJECTIVE: To determine whether using discrete semantic entropy (DSE) to reject questions likely to generate hallucinations can improve the accuracy of black-box vision-language models (VLMs) in radiologic image-based visual question answering (VQA). MATERIALS AND METHODS: This retrospective study evaluated DSE using two publicly available, de-identified datasets: the VQA-Med 2019 benchmark (500…

Create a free account to keep reading

Free members get 10 full research views every month across publications, clinical trials, FDA clearances, adverse events, and NIH grants. No credit card required.

Create Free AccountSign In

Want unlimited research access? See Pro plans

Data Accuracy Notice: Research intelligence on Health AI Central is aggregated from public sources (PubMed, ClinicalTrials.gov, FDA, NIH, CMS, and others) and refreshed nightly. Classifications and derived metrics are produced by automated methods described in our Methodology. We recommend verifying critical data points against the primary sources before making decisions.