rehabib/PRECISE-Africa-Breast-Ultrasound-Segmentation-and-Classification-Challenge-PACE-

# PRECISE-Africa-Breast-Ultrasound-Segmentation-and-Classification-Challenge-PACE- The PRECISE-ABreast Challenge utilizes a curated, multi-institutional dataset of breast ultrasound (BUS) images to support the development and evaluation of algorithms for lesion detection, classification, and segmentation. The dataset integrates images from four distinct sources: the BUSI dataset by Al-Dhabyani et al. (2020)[1], the BrEaST dataset by Pawłowska et al. (2024)[2], the BUS-BRA dataset by Gómez-Flores et al. (2024)[3], and the ABreast point-of-care dataset, an unpublished prospective collection of breast handheld ultrasound scans from community-dwelling women in sub-Saharan Africa. All images are paired with expert-verified reference annotations, including lesion boundaries and diagnostic labels (normal, benign, malignant), enabling robust supervised learning and quantitative evaluation. Each ultrasound study includes a single 2D grayscale image or a Doppler image, with resolutions varying based on the equipment used.