MTFN: multi-temporal feature fusing network with co-attention for DCE-MRI synthesis

Area of Science:

• Medical Imaging
• Artificial Intelligence in Healthcare
• Radiology

Background:

• Dynamic Contrast Enhanced Magnetic Resonance Imaging (DCE-MRI) is crucial for breast cancer diagnosis and treatment.
• Extended scanning times for DCE-MRI cause patient discomfort.
• Current methods require complete temporal image acquisition, leading to prolonged procedures.

Purpose of the Study:

• To develop a method for generating DCE-MRI images without complete scanning.
• To reduce patient discomfort associated with long MRI scan times.
• To enable faster acquisition of essential DCE-MRI data.

Main Methods:

• Proposed a Multi-Temporal Feature Fusing Neural Network with Co-attention (MTFN).
• Employed a Co-attention module to fuse features from early temporal images (1st and 3rd).
• Focused on capturing long-range dependencies for enhanced feature fusion.

Main Results:

• MTFN successfully generated realistic DCE-MRI images.
• Experiments conducted on private breast DCE-MRI and BraTs2018 datasets.
• Achieved a 3% improvement in breast cancer molecular typing classification accuracy using synthetic images (92.46% vs. 89.53%).

Conclusions:

• The MTFN method is effective in generating comprehensive and useful DCE-MRI information.
• Generated synthetic images are practical and improve diagnostic accuracy.
• The approach offers a viable solution for faster DCE-MRI acquisition.