❌

Normal view

Received — 19 November 2025 ⏭ Favourites

Foundation Models in Medical Imaging: A Review and Outlook

✇cs.AI, q-bio.NC updates on arXiv.org
By: Vivien van Veldhuizen · Vanessa Botha · Chunyao Lu · Melis Erdal Cesur · Kevin Groot Lipman · Edwin D. de Jong · Hugo Horlings · Cl\'arisa I. Sanchez · Cees G. M. Snoek · Lodewyk Wessels · Ritse Mann · Eric Marcus · Jonas Teuwen
19 November 2025 at 13:00
arXiv:2506.09095v4 Announce Type: replace-cross Abstract: Foundation models (FMs) are changing the way medical images are analyzed by learning from large collections of unlabeled data. Instead of relying on manually annotated examples, FMs are pre-trained to learn general-purpose visual features that can later be adapted to specific clinical tasks with little additional supervision. In this review, we examine how FMs are being developed and applied in pathology, radiology, and ophthalmology, drawing on evidence from over 150 studies. We explain the core components of FM pipelines, including model architectures, self-supervised learning methods, and strategies for downstream adaptation. We also review how FMs are being used in each imaging domain and compare design choices across applications. Finally, we discuss key challenges and open questions to guide future research.
Received — 3 November 2025 ⏭ Favourites

A Multi-Stage Framework with Taxonomy-Guided Reasoning for Occupation Classification Using Large Language Models

✇cs.AI, q-bio.NC updates on arXiv.org
By: Palakorn Achananuparp · Ee-Peng Lim · Yao Lu
3 November 2025 at 13:00
arXiv:2503.12989v3 Announce Type: replace-cross Abstract: Automatically annotating job data with standardized occupations from taxonomies, known as occupation classification, is crucial for labor market analysis. However, this task is often hindered by data scarcity and the challenges of manual annotations. While large language models (LLMs) hold promise due to their extensive world knowledge and in-context learning capabilities, their effectiveness depends on their knowledge of occupational taxonomies, which remains unclear. In this study, we assess the ability of LLMs to generate precise taxonomic entities from taxonomy, highlighting their limitations, especially for smaller models. To address these challenges, we propose a multi-stage framework consisting of inference, retrieval, and reranking stages, which integrates taxonomy-guided reasoning examples to enhance performance by aligning outputs with taxonomic knowledge. Evaluations on a large-scale dataset show that our framework not only enhances occupation and skill classification tasks, but also provides a cost-effective alternative to frontier models like GPT-4o, significantly reducing computational costs while maintaining strong performance. This makes it a practical and scalable solution for occupation classification and related tasks across LLMs.
Received — 13 August 2025 ⏭ Favourites

Development and validation of an integrative 54 biomarker-based risk identification model for multi-cancer in 42,666 individuals: a population-based prospective study to guide advanced screening strategies

✇Omics in Gastric
By: Renjia Zhao · Huangbo Yuan · Yanfeng Jiang · Zhenqiu Liu · Ruilin Chen · Shuo Wang · Linyao Lu · Ziyu Yuan · Zhixi Su · Qiye He · Kelin Xu · Tiejun Zhang · Li Jin · Ming Lu · Weimin Ye · Rui Liu · Chen Suo · Xingdong Chen
12 August 2025 at 18:00

Biomark Res. 2025 Aug 11;13(1):101. doi: 10.1186/s40364-025-00812-z.

ABSTRACT

BACKGROUND: Early identification of high-risk individuals is crucial for optimizing cancer screening, particularly when considering expensive and invasive methods such as multi-omics technologies and endoscopic procedures. However, developing a robust, practical multi-cancer risk prediction model that integrates diverse, multi-scale data and with proper validation remains a significant challenge.

METHODS: We initialized the FuSion study by recruiting 42,666 participants from Taizhou, China, with a discovery cohort (n = 16,340) and an independent validation cohort (n = 26,308) after exclusion criteria. We integrated multi-scale data from 54 blood-derived biomarkers and 26 epidemiological exposures to develop a risk prediction model for five common cancers, including lung, esophageal, liver, gastric, and colorectal cancer. Employing five supervised machine learning approaches, we used a LASSO-based feature selection strategy to identify the most informative predictors. The model was trained and internally validated in the discovery cohort, externally applied in the validation cohort, and further evaluated through a prospective clinical follow-up to assess cancer events via clinical examinations.

RESULTS: The final model comprising four key biomarkers along with age, sex, and smoking intensity, achieving an AUROC of 0.767 (95% CI: 0.723-0.814) for five-year risk prediction. High-risk individuals (17.19% of the cohort) accounted for 50.42% of incident cancer cases, with a 15.19-fold increased risk compared to the low-risk group. During follow-up of 2,863 high-risk subjects, 9.64% were newly diagnosed with cancer or precancerous lesions. Notably, cancer detection in the high-risk group was 5.02 times higher than in the low-risk group and 1.74 times higher than in the intermediate-risk group. In particular, the incidence of esophageal cancers in the high-risk group was 16.84 times that of the low-risk group.

CONCLUSIONS: This is the first population-based prospective study in a large Chinese cohort that leverage multi-scale data including biomarkers for multi-cancer risk prediction. Our effective risk stratification model not only enhances early cancer detection but also lays the foundation for the targeted application of advanced screening methods, including but not limited to multi-omics technologies and endoscopy. These findings support precision prevention strategies and the optimal allocation of healthcare resources.

PMID:40790537 | PMC:PMC12341305 | DOI:10.1186/s40364-025-00812-z

There are no more articles
❌