MTSCL-Net: Multi-level temporal spatial contrastive learning for robust breast tumor segmentation in DCE-MRI

Key Points

• Robust breast tumor segmentation is achieved with a novel model using multi-level temporal spatial contrastive learning,
• Segmentation accuracy improved by 15% compared to previous methods, showing the potential of advanced modeling techniques.
• Evaluation was conducted using DCE-MRI across multiple patient datasets, ensuring a comprehensive analysis of model performance.
• The findings support the need for more sophisticated algorithms in medical imaging for better diagnosing breast tumors.