Downloads · 30 days
15
3% of all-time downloads
hab200/breast-cancer-late-fusion
breast-cancer-late-fusion is a machine learning model from hab200. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for keras.
--- Multi-modal breast cancer classification system that combines mammogram and ultrasound images using Late Fusion to improve diagnostic accuracy.
Downloads · 30 days
15
3% of all-time downloads
All-time downloads
589
Public
Repo size
793 MB
Likes
0
Public
Click a slice to open those files.
.keras793 MB · 100%
From the Hugging Face model README
Multi-modal breast cancer classification system that combines mammogram and ultrasound images using Late Fusion to improve diagnostic accuracy.
| Model | AUC | Sensitivity | Specificity |
|---|---|---|---|
| Mammogram Only (EfficientNetV2L) | 0.7515 | 86.75% | 52.57% |
| Ultrasound Only (ResNet50) | 0.8945 | 79.52% | 78.86% |
| Late Fusion (Ours) | 0.9199 | 84.34% | 84.57% |
The fusion achieves an optimal balance between sensitivity and specificity, outperforming single-modality approaches and significantly reducing false positives while maintaining high sensitivity.
Mammogram Image (456×456)
↓ EfficientNetV2L (fine-tuned on CBIS-DDSM)
↓ PS_mammo = P(malignant)
↓
Logistic Regression (C=0.001)
(Late Fusion Meta-Classifier)
↓
Ultrasound Image (224×224) Prediction
↓ ResNet50 (fine-tuned on BUS-BRA) (benign / malignant)
↓ PS_us = P(malignant)