780 breast ultrasound images categorized into normal, benign, and malignant classes. The dataset includes corresponding ground truth segmentation masks for each image to facilitate automated lesion detection and boundary delineation.
Use Cases
- Train a multi-class classifier to distinguish between normal, benign, and malignant tissue using folder-based labels
- Develop image segmentation architectures like U-Net using the ground truth mask files to delineate tumor boundaries
- Benchmark object detection models for localized breast cancer screening on 500x500 pixel ultrasound frames
Strengths
- 780 ultrasound images across three diagnostic categories
- Includes binary ground truth masks for lesion segmentation
- Images provided in PNG format at 500x500 pixel resolution