---
title: "Micro-Ultrasound Foundation Model (muFM) Enables Data-Efficient Prostate Image Analysis"
id: "medrxiv-9-data-efficient-prostate-micro-ultrasound-image-analysis-using-a-modality"
canonical_url: "https://medichelpline.com/clinical-feed/medrxiv-9-data-efficient-prostate-micro-ultrasound-image-analysis-using-a-modality"
content_type: "clinical_feed_article"
specialty: "Radiology"
source_name: "medRxiv (Clinical Preprints)"
source_url: "https://www.medrxiv.org/content/10.64898/2026.09.29.26364330v1?rss=1"
published_at: "2026-10-05T12:00:00.000Z"
evidence_level: "Verified Feed"
license: "CC-BY-NC-4.0 / Informational Use"
---
# Micro-Ultrasound Foundation Model (muFM) Enables Data-Efficient Prostate Image Analysis
## Provenance & Clinical Metadata
- **Canonical URL:** https://medichelpline.com/clinical-feed/medrxiv-9-data-efficient-prostate-micro-ultrasound-image-analysis-using-a-modality
- **Specialty:** [Radiology](https://medichelpline.com/clinical-feed/radiology.md)
- **Primary Source:** medRxiv (Clinical Preprints)
- **Source URL:** [Original Journal Publication](https://www.medrxiv.org/content/10.64898/2026.09.29.26364330v1?rss=1)
- **Published At:** 2026-10-05T12:00:00.000Z
- **Evidence Rating:** Verified Feed
## Executive GIST (TL;DR)
- The authors developed **muFM**, a Micro-ultrasound (MicroUS)–specific foundation model pretrained with DINOv2-style self-supervised learning on 1,736,391 unlabeled MicroUS images from 780 patients. - muFM was fine-tuned and evaluated on three downstream prostate tasks: prostate capsule segmentation (public cohort, 75 patients), urethra segmentation (institutional cohort, 105 patients), and prostate cancer lesion localization (institutional cohort, 190 patients). - Under limited annotation, muFM produced the largest improvements compared with randomly initialized models and foundation models pretrained on natural, medical, or other ultrasound images. - With only 10% of training scans, muFM achieved Dice scores of **93.8%** for prostate capsule segmentation and **58.4%** for urethra segmentation, versus 87.9% and 47.0% for the strongest comparator models; corresponding HD95 values were 2.0 mm and 4.3 mm. - For lesion localization, muFM reached about **0.75 sensitivity at two false-positive detections per 3D volume**, outperforming other pretrained models. - The pretraining was retrospective and single-institution; the approach did not include explicit 3D modeling. Data used are not publicly available but may be requested from the corresponding author with IRB permission. - Ethical approval was provided by the UCLA Institutional Review Board with waiver of informed consent. Relevant competing interests include affiliations and financial relationships between some authors and Exact Imaging and Siemens Healthineers. - The authors conclude that MicroUS-specific self-supervised pretraining can reduce annotation requirements and improve data-efficient automated tools for MicroUS image analysis.
## Clinical Analysis & Structured Key Points
Data-Efficient Prostate Micro-Ultrasound Image Analysis using a Modality-specific Foundation Model | medRxiv Skip to main content Data-Efficient Prostate Micro-Ultrasound Image Analysis using a Modality-specific Foundation Model View ORCID Profile Kaifeng Pang , Junxin Fan , View ORCID Profile Alex Ling Yu Hung , View ORCID Profile Qi Miao , View ORCID Profile Changsuk Oh , View ORCID Profile Qiudi He , View ORCID Profile Brian Wodlinger , View ORCID Profile Adam Kinnaird , View ORCID Profile Wayne Brisbane , View ORCID Profile Kyunghyun Sung doi: https://doi.org/10.64898/2026.09.29.26364330 Kaifeng Pang 1 UCLA; Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Kaifeng Pang For correspondence: kaifengpang{at}mednet.ucla.edu Junxin Fan 1 UCLA; Find this author on Google Scholar Find this author on PubMed Search for this author on this site Alex Ling Yu Hung 1 UCLA; Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Alex Ling Yu Hung Qi Miao 1 UCLA; Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Qi Miao Changsuk Oh 1 UCLA; Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Changsuk Oh Qiudi He 1 UCLA; Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Qiudi He Brian Wodlinger 2 Exact Imaging; Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Brian Wodlinger Adam Kinnaird 3 University of Alberta Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Adam Kinnaird Wayne Brisbane 1 UCLA; Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Wayne Brisbane Kyunghyun Sung 1 UCLA; Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Kyunghyun Sung Abstract Info/History Metrics Supplementary material Data/Code Preview PDF Abstract Background and Objective: Micro-ultrasound (MicroUS) is an emerging high-resolution imaging modality for prostate cancer diagnosis and real-time biopsy guidance, but its automated analysis is limited by labor-intensive expert annotation. We aimed to develop and evaluate a foundation model for data-efficient MicroUS image analysis. Methods: We developed muFM, a MicroUS-specific foundation model pretrained using DINOv2-style self-supervised learning on 1,736,391 unlabeled images from 780 patients. The model was fine-tuned and evaluated in three downstream cohorts for prostate capsule segmentation (public, 75 patients), urethra segmentation (institutional, 105 patients), and prostate cancer lesion localization (institutional, 190 patients). Under limited annotation, muFM was compared with randomly initialized models and foundation models pretrained on natural, medical, and ultrasound images. Evaluation used the Dice scores and 95th-percentile Hausdorff distance (HD95) for segmentation and free-response receiver operating characteristic analysis for lesion localization. Key Findings and Limitations: muFM improved performance across all three tasks, with the largest gains under limited annotation. With 10% of training scans, muFM achieved Dice scores of 93.8% for prostate capsule segmentation and 58.4% for urethra segmentation, versus 87.9% and 47.0% for the strongest comparators. Corresponding HD95 values were 2.0 and 4.3 mm. For lesion localization, muFM outperformed other models with approximately 0.75 sensitivity at two false-positive detections per 3D volume. Limitations include retrospective single-institution pretraining and lack of explicit 3D modeling. Conclusions and Clinical Implications: MicroUS-specific self-supervised pretraining improved data-efficient image analysis across prostate and urethra segmentation, and lesion localization. This approach may reduce annotation requirements for developing automated tools in MicroUS image analysis. Competing Interest Statement Dr. Kyunghyun Sung has a financial interest in Siemens Healthineers. Dr. Adam Kinnaird has received research funding and honoraria from Exact Imaging. Dr. Wayne Brisbane is a consultant for Exact Imaging. Brian Wodlinger is affiliated with Exact Imaging, the manufacturer of the ExactVu MicroUS system used in this study. The remaining authors declare no competing interests. Author Declarations I confirm all relevant ethical guidelines have been followed, and any necessary IRB and/or ethics committee approvals have been obtained. Yes The details of the IRB/oversight body that provided approval or exemption for the research described are given below: The Institutional Review Board of the University of California, Los Angeles gave ethical approval for this work and waived the requirement for informed consent. I confirm that all necessary patient/participant consent has been obtained and the appropriate institutional forms have been archived, and that any patient/participant/sample identifiers included were not known to anyone (e.g., hospital staff, patients or participants themselves) outside the research group so cannot be used to identify individuals. Yes I understand that all clinical trials and any other prospective interventional studies must be registered with an ICMJE-approved registry, such as ClinicalTrials.gov. I confirm that any such study reported in the manuscript has been registered and the trial registration ID is provided (note: if posting a prospective study registered retrospectively, please provide a statement in the trial ID field explaining why the study was not registered in advance). Yes I have followed all appropriate research reporting guidelines, such as any relevant EQUATOR Network research reporting checklist(s) and other pertinent material, if applicable. Yes Data Availability The data used in this study are not publicly available due to patient privacy and institutional restrictions, but may be available from the corresponding author upon reasonable request and with permission from the Institutional Review Board (IRB). Funder Information Declared National Institutes of Health , R01-CA248506 , R01-CA272702 Copyright The copyright holder for this preprint is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license . Back to top Previous Next Posted October 05, 2026. Download PDF Supplementary Material Data/Code Email Thank you for your interest in spreading the word about medRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Data-Efficient Prostate Micro-Ultrasound Image Analysis using a Modality-specific Foundation Model Message Subject (Your Name) has forwarded a page to you from medRxiv Message Body (Your Name) thought you would like to see this page from the medRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Data-Efficient Prostate Micro-Ultrasound Image Analysis using a Modality-specific Foundation Model Kaifeng Pang , Junxin Fan , Alex Ling Yu Hung , Qi Miao , Changsuk Oh , Qiudi He , Brian Wodlinger , Adam Kinnaird , Wayne Brisbane , Kyunghyun Sung medRxiv 2026.09.29.26364330; doi: https://doi.org/10.64898/2026.09.29.26364330 Share This Article: Copy Citation Tools Data-Efficient Prostate Micro-Ultrasound Image Analysis using a Modality-specific Foundation Model Kaifeng Pang , Junxin Fan , Alex Ling Yu Hung , Qi Miao , Changsuk Oh , Qiudi He , Brian Wodlinger , Adam Kinnaird , Wayne Brisbane , Kyunghyun Sung medRxiv 2026.09.29.26364330; doi: https://doi.org/10.64898/2026.09.29.26364330 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Areas All Articles Addiction Medicine (626) Allergy and Immunology (914) Anesthesia (340) Cardiovascular Medicine (4884) Dentistry and Oral Medicine (486) Dermatology (421) Emergency Medicine (659) Endocrinology (including Diabetes Mellitus and Metabolic Disease) (1643) Epidemiology (16062) Forensic Medicine (34) Gastroenterology (1218) Genetic and Genomic Medicine (7128) Geriatric Medicine (736) Health Economics (1072) Health Informatics (5122) Health Policy (1447) Health Systems and Quality Improvement (1792) Hematology (589) HIV/AIDS (1355) Infectious Diseases (except HIV/AIDS) (16355) Intensive Care and Critical Care Medicine (1184) Medical Education (678) Medical Ethics (156) Nephrology (729) Neurology (7384) Nursing (370) Nutrition (1102) Obstetrics and Gynecology (1263) Occupational and Environmental Health (1010) Oncology (3656) Ophthalmology (1066) Orthopedics (399) Otolaryngology (459) Pain Medicine (483) Palliative Medicine (142) Pathology (711) Pediatrics (1822) Pharmacology and Therapeutics (741) Primary Care Research (776) Psychiatry and Clinical Psychology (5986) Public and Global Health (9833) Radiology and Imaging (2463) Rehabilitation Medicine and Physical Therapy (1478) Respiratory Medicine (1256) Rheumatology (651) Sexual and Reproductive Health (785) Sports Medicine (588) Surgery (786) Toxicology (107) Transplantation (310) Urology (298)
## Related Clinical Research

- [Gated-Attention Multiple Instance Learning for Multi-Center Cervical Cytology Slide Triage](https://medichelpline.com/clinical-feed/medrxiv-1-weakly-supervised-gated-attention-multiple-instance-learning-for-multi-center.md)
- [Accelerated Whole-Body DWI with Deep Learning Reconstruction in Metastatic Prostate Cancer: Image](https://medichelpline.com/clinical-feed/medrxiv-11-evaluation-of-accelerated-whole-body-diffusion-weighted-imaging-with-deep.md)
- [Tailored Treatment Approaches for Breast Cancer in Older Women](https://medichelpline.com/clinical-feed/stat-news-5-breast-cancer-can-behave-differently-in-older-women-this-physician-is-making.md)
- [Cancer prevalence and characteristics in primary care patients with type 2 diabetes mellitus](https://medichelpline.com/clinical-feed/pubmed-42817592.md) (DOI: 10.1017/S1463423626101935)
- [Sintilimab Plus ADT for Synchronous Prostate Adenocarcinoma and Lung Squamous Cell Carcinoma in a](https://medichelpline.com/clinical-feed/frontiers-in-immunology-7-successful-management-of-synchronous-prostate-adenocarcinoma-and-lung-squamous.md)

## Navigation
- [← Back to Radiology Feed](https://medichelpline.com/clinical-feed/radiology.md)
- [← All Clinical Specialties](https://medichelpline.com/clinical-feed.md)
## Medical & Regulatory Disclaimer

> [!CAUTION]
> MedicHelpline content is structured for research, educational, and professional discovery purposes. It does not constitute individual medical advice, clinical diagnosis, or treatment recommendations.
> Always verify dosing, contraindications, and regulatory alerts against official product labeling and primary regulatory sources before clinical decision-making.