Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This study investigates the impact of dataset-origin bias on the performance of deep learning models for breast MRI tumor classification, revealing that traditional training methods lead to near-chance accuracy on external datasets. By employing a mixed training set that incorporates samples from multiple sources while maintaining patient-level controls, the authors significantly improve classification accuracy and F1 scores on the independent MAMA-MIA cohort. The findings underscore the necessity of addressing domain shifts in medical imaging to enhance the generalizability of AI models across different institutions.
Dataset-origin bias can reduce external accuracy to near chance, but mixing training data from multiple sources boosts classification performance to over 88%.
Breast MRI is highly sensitive for detecting breast tumors, but exams contain many slices and require substantial reading time. Deep learning models often perform well on internal splits but can fail across institutions because of domain shift and dataset-origin bias. We study this failure mode for binary breast MRI tumor classification. EfficientNet-B3 and WaveViT-Small are trained using Duke Breast Cancer MRI and fastMRI, and evaluated only on the independent multi-center MAMA-MIA cohort. In a deliberately confounded setup, where label is perfectly correlated with dataset origin, external accuracy is near chance (0.5048--0.5265), despite very high recall. We then construct a mixed training set in which each class contains samples from both Duke and fastMRI, while preserving patient-level splitting, augmentation, and leakage controls. On MAMA-MIA, dataset mixing improves accuracy/F1 to 0.8463/0.8625 for WaveViT-Small and 0.8884/0.8994 for EfficientNet-B3. These results show that controlling dataset-origin bias is important for reliable breast MRI classification.