Longitudinal in silico imaging study comparing digital mammography and digital breast tomosynthesis systems

Area of Science:

• Medical Imaging
• Computational Modeling
• Radiology

Background:

• In silico clinical trials offer sophisticated, realistic assessments of medical imaging systems.
• Fully computational models provide fast, affordable trial designs mirroring real-world clinical trends.

Purpose of the Study:

• Evaluate three breast imaging system models for digital mammography (DM) and digital breast tomosynthesis (DBT) using a fully in silico longitudinal study.
• Compare the performance of different detector technologies in breast imaging.

Main Methods:

• Developed in silico models for three breast imaging systems, incorporating detector technology, pixel size, and projection parameters.
• Utilized a computational reader to assess mass and calcification detectability across systems.
• Calculated the Area Under the ROC Curve (AUC) for various lesion types, sizes, breast densities, and imaging systems (DM vs. DBT).

Main Results:

• Mass detectability (AUC) increased proportionally with mass size and decreased with higher breast density.
• Significant performance differences were observed between device models for mass detection, with digital breast tomosynthesis (DBT) showing a higher average AUC (0.109).
• Digital mammography (DM) outperformed DBT in calcification detection, particularly in dense breasts (AUC difference of -0.055 favoring DM).

Conclusions:

• In silico comparison revealed that breast imaging system technology influences the visibility of masses and calcifications.
• The study highlights the advantages of in silico methodology for resource-efficient device development, optimization, and regulatory evaluation.
• Longitudinal, multi-device in silico studies are feasible and valuable for advancing breast imaging technology.