Diagnostic Performance of AI-Assisted Handheld Breast Ultrasound for Early Cancer Detection in Resource-Limited Screening Clinics

Description

This prospective diagnostic accuracy study was conducted across three peripheral women’s health clinics in Sudan to assess the feasibility of AI-assisted handheld breast ultrasound (HHUS) as a point-of-care triage tool for early cancer detection. A total of 232 women aged 30–65 years presenting with breast symptoms were examined by trained non-radiologist clinicians using portable HHUS devices equipped with an offline artificial intelligence algorithm for lesion classification. All scans were anonymized and independently reviewed by two blinded breast radiologists, and suspicious lesions underwent biopsy for histopathological confirmation. The AI system achieved high diagnostic concordance with radiologists while reducing the median time-to-triage from 72 hours to 10 minutes, highlighting its potential to enhance breast cancer detection in resource-limited Sudanese settings.