Multimodal Breast Parenchymal Patterns Correlation Using a Patient-Specific Biomechanical Model.

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Radiology

Background:

• Accurate breast density assessment is crucial for mammography interpretation and breast cancer risk evaluation.
• Bridging the gap between 3D Magnetic Resonance Imaging (MRI) and 2D mammography data presents a significant challenge in breast imaging analysis.

Purpose of the Study:

• To develop an automated framework for generating realistic 2D projections of breast parenchymal distribution from 3D MRI data.
• To validate the accuracy of the generated 2D projections by comparing them with density maps from full-field digital mammograms.

Main Methods:

• A subject-specific biomechanical breast model was developed for registering 3D MRI volumes to 2D X-ray mammograms.
• An optimization process adjusted the breast model's parameters (position, orientation, elasticity) for accurate image alignment.
• A novel ray-casting approach was employed to project MRI glandular tissue onto the mammogram plane without resampling, minimizing information loss.

Main Results:

• The developed framework achieved high structural agreement between simulated and real breast density maps across varying glandularity.
• Increased similarity in glandular tissue distribution and correlation were observed between the synthetic and real density maps for denser breasts.
• The generated synthetic images demonstrated continuity with large structures present in the mammographic density maps.

Conclusions:

• The automated framework provides a robust method for creating accurate 2D breast density projections from 3D MRI.
• The approach shows promise for improving the correlation and comparison between MRI and mammography data, particularly in dense breast tissue.
• This technique could enhance the interpretation of breast density and potentially aid in breast cancer risk assessment and screening.