Vision-language zero-shot models for radiographic image classification: a systematic review

Vista sencilla de ítem
dc.contributor.authorGuerrero Tamayo, Ana
dc.contributor.authorOleagordia Ruiz, Ibon
dc.contributor.authorGarcía-Zapirain, Begoña
dc.date.accessioned2026-03-03T18:47:48Z
dc.date.available2026-03-03T18:47:48Z
dc.date.issued2026-03
dc.date.updated2026-03-03T18:47:48Z
dc.description.abstractZero-shot Vision-Language Models (VLMs) link visual and textual features, enabling generalization to unseen domains, making them promising for radiographic diagnosis, though clinical adoption is limited. This systematic review examines zero-shot VLMs applied to radiographic image classification, following the PRISMA methodology. Articles were identified from IEEE, PubMed, Scopus, and Web of Science, with 16 selected after exhaustive screening. The analysis addressed five research questions (RQ1–RQ5) covering dataset characteristics, model attributes, natural language integration, reported limitations, and hyperparameter tuning. Geographically, China (37%) and the United States (38%) contributed 75% of the reviewed studies, with no EU-led research identified, highlighting the need for increased European engagement in this field. Architecturally (RQ2), high heterogeneity exists, with dual-encoder (43.75%) and attention-based fusion models most common. Most models (81.25%) employ a Joint Embedding Space for multimodal alignment. Regarding datasets and natural language use (RQ1, RQ3), VLMs rely on few large but semantically narrow datasets, limiting generalizability and amplifying bias. Real clinical reports (direct supervision) and implicit pretrained textual embeddings each represent 37.5% of strategies, yet unstructured clinical text is underutilized. Limited vision-language integration negatively affects performance and explainability (RQ4). Hyperparameter tuning (RQ5) is rarely reported, with 9 of 16 studies not specifying methods, compromising reproducibility. There is an urgent need for open, multilingual, multimodal datasets reflecting clinical and geographic diversity. Clinically useful zero-shot VLMs require transparent evaluation, including explainability metrics. Future models should adopt a multidisciplinary approach, combining technical innovation with usability, data representativeness, and methodological transparency to ensure diagnostic robustness.en
dc.description.sponsorshipThis work has been supported by the Basque Government through the Hazitek 2024 program, Spain , within the framework of the IRUD-IA project: “Medical Image Analysis Technologies with Artificial Intelligence for the Development of Medical Devices”, project code ZE-2024/00030en
dc.identifier.citationGuerrero-Tamayo, A., Oleagordia-Ruiz, I., & Garcia-Zapirain, B. (2026). Vision-language zero-shot models for radiographic image classification: a systematic review. Machine Learning with Applications, 23. https://doi.org/10.1016/J.MLWA.2025.100826
dc.identifier.doi10.1016/J.MLWA.2025.100826
dc.identifier.eissn2666-8270
dc.identifier.urihttps://hdl.handle.net/20.500.14454/5321
dc.language.isoeng
dc.publisherElsevier Ltd
dc.rights© 2025 The Authors
dc.subject.otherImage classification
dc.subject.otherRadiographic
dc.subject.otherSurvey
dc.subject.otherSystematic review
dc.subject.otherVision-language models
dc.subject.otherX-ray
dc.subject.otherZero-shot
dc.titleVision-language zero-shot models for radiographic image classification: a systematic reviewen
dc.typereview article
dcterms.accessRightsopen access
oaire.citation.titleMachine Learning with Applications
oaire.citation.volume23
oaire.licenseConditionhttps://creativecommons.org/licenses/by-nc-nd/4.0/
oaire.versionVoR
Archivos
Bloque original
Mostrando 1 - 1 de 1
Miniatura
Nombre:
guerrero_vision_2026.pdf
Tamaño:
2.5 MB
Formato:
Adobe Portable Document Format
Descargar
Colecciones
Artículos