An AI-enabled comprehensive breast ultrasound diagnostic system for low-resource settings without a sonographer or a radiologist

Sep 29, 2026

Uploaded by Jake

Breast cancer is the leading cause of cancer death in women, yet early detection via mammography is constrained in low-resource settings by cost and technical complexity. Ultrasound offers an alternative, but its impact is limited by the scarcity of trained sonographers and radiologists. This study proposes an automated AI-enabled Breast VSI (VSI-B) system combining segmentation (Attention U-Net 3D) and lesion classification to support non-expert environments. Evaluated on 98 patients, it uses key-frame selection and majority voting to improve mass identification and classification performance.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages18

Reading Time45 min

AccessFree to download

Background and clinical need

Proposed AI-enabled VSI-B system

Segmentation and classification pipeline

Key-frame selection and majority voting

Study dataset and evaluation metrics

Results and diagnostic performance

Implications for low-resource settings

[www. nature.com/scientificreports](www. nature.com/scientificreports) OPEN AnAI-enabled comprehensive breast ultrasound diagnostic system for low-resource settings without a sonographer or a radiologist Emilio J. Ochoa1, Luis C. Revilla2, Stefano E. Romero2, Giancarlo A. Guarnizo2, Thomas J. Marini3, Kevin J. Parker4, YuT. Zhao3, Galen Brennan3, Jonah Kan5, Steven Meng6, Ann Dozier7, Anna Weiss8,9 & Benjamin Castaneda1􀀍 Breast cancer is the most common non-skin related malignancy and the leading cause of cancer death in women. Mammography remains the gold standard for early detection; however, its accessibility is limited in low-resource settings due to cost and technical complexity. Ultrasound (US) is a viable alternative, but its implementation is hindered by the scarcity of trained radiologists and sonographers. Volume sweep imaging (VSI) has addressed the issue of US acquisition by enabling non-specialists to perform standardized scans. However, these still require expert interpretation, limiting their impact. To overcome this barrier, we propose a fully automated Breast VSI (VSI-B) system integrating artificial intelligence (AI) for segmentation and classification of breast lesions, aiming to provide an accessible diagnostic tool for low-resource environments. This study developed an AI-driven diagnostic system for VSI-B, combining a segmentation model (Attention U-Net 3D) with a classification model for lesion detection. A total of 98 patients with palpable breast lumps were included in the study. The dataset consisted of 392 VSI-B US videos and 2,100 classified frames. Anew method was implemented to enhance mass identification by selecting key frames for analysis. A majority voting algorithm was used to optimize lesion classification. The system’s performance was assessed based on sensitivity, specificity, and accuracy. Following a detection step that achieved 100% sensitivity and 93.6% specificity for cancer and no cancer patients, as well as 95.0% sensitivity and 63.0% specificity for mass and no mass patients, the DenseNet classification model reached 87% accuracy, 100% sensitivity, and 83% specificity. A majority voting algorithm optimized classification, yielding anAUC of 0.91. This study highlights the potential of anAI-enabled VSI-B system as a reliable diagnostic tool in low-resource settings. By integrating multi-modal segmentation and classification, the system automates breast lesion detection and stratification, reducing reliance on radiologists. The results suggest that this approach could enhance early breast cancer diagnosis and guide clinical decision-making, particularly in underserved regions. Keywords VSI, Breast ultrasound, Breast cancer research, Artificial intelligence, Telemedicine Approximately 2.1 million women are diagnosed with breast cancer worldwide each year, and approximately 685,000 deaths occur annually1. Breast cancer incidence and mortality are expected to increase, and it is 1Department of Biomedical Engineering, University of Rochester, Rochester, NY, USA. 2Departamento de Ingeniería, Pontificia Universidad Católica del Perú, Lima, Peru. 3Department of Imaging Sciences, University of Rochester Medical Center, Rochester, NY, USA. 4Department of Electrical Engineering, University of Rochester, New York, USA. 5Department of Internal Medicine, Mayo Clinic, Rochester, MN, USA. 6Department of Radiology, UT Southwestern Medical Center, Dallas, TX, USA. 7Department of Public Health, University of Rochester Medica Center, Rochester, NY, USA. 8Division of Surgical Oncology, Department of Surgery, University of Rochester Medical Center, Rochester, NY, USA. 9Wilmot Cancer Center, University of Rochester Medical Center, Rochester, NY, USA. 􀀍 email: [bcastane@ur.rochester.edu](bcastane@ur.rochester.edu) [www. nature.com/scientificreports/](www. nature.com/scientificreports/) projected to reach more than 3.2 million new cases by 20302, with at least 60% of deaths in low-and m

FAQs

Why is mammography limited in low-resource settings?

Mammography is constrained by cost and technical complexity, reducing accessibility in low-resource environments.

What does the proposed VSI-B system do?

The system is fully automated and integrates AI for segmentation and classification of breast lesions to support diagnosis without reliance on expert interpretation.

How was diagnostic performance evaluated?

Performance was assessed using sensitivity, specificity, accuracy, and AUC, including a detection step followed by DenseNet-based classification optimized by majority voting.

Sep 29, 2026

Uploaded by Jake