Performance evaluation of deep learning based YOLOv5 and YOLOv8 models for real time breast cancer detection in mammographic images
Advanced deep learning detects breast cancer lesions in mammograms with near-perfect accuracy on test images.
Evaluation of YOLOv5 and YOLOv8 deep learning models on 2,620 DDSM mammographic images showed YOLOv8 achieved superior mAP (0.99) with strong precision and recall for breast cancer lesion detection. While these results support YOLOv8 for automated screening development, validation on prospective or clinical datasets beyond DDSM is needed before deployment.
What the study was
- Study design
- Retrospective model comparison
- Population
- DDSM dataset: 2,620 digitized mammography images (normal, benign, malignant)
- Sample size
- 2620
- Category
- Diagnostics
- Maturity
- Exploratory
- Journal
- Discovery Oncology
Why it surfaced
Standard YOLO comparison on public DDSM dataset; incremental contribution; DDSM is a historical dataset with limited clinical representativeness; no prospective validation.
A plain-language summary of published research — not medical advice. Talk to a clinician about your own care.