Mammogram synthesis using a 3D simulation. II. Evaluation of synthetic mammogram texture

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Computational Anatomy

Background:

• Mammography is crucial for breast cancer screening.
• Accurate simulation of mammogram texture is essential for developing and validating imaging algorithms.
• Understanding breast tissue composition influences image texture.

Purpose of the Study:

• To evaluate a novel method for synthesizing realistic mammograms.
• To compare the texture characteristics of synthetic and clinical mammograms.
• To determine the effectiveness of simulated adipose tissue compartment projections in replicating clinical mammogram texture.

Main Methods:

• Developed a synthesis algorithm based on breast tissue and mammographic imaging process simulations.
• Synthesized mammogram texture using projections of simulated adipose tissue compartments.
• Analyzed texture energy and fractal dimension using mathematical morphology and compared distributions between clinical and synthetic images.

Main Results:

• Synthetic mammogram texture closely simulated the mean features of clinical mammograms across 95 samples.
• The range of texture features observed in a large group of clinical mammograms was effectively replicated by synthetic images.
• Optimal simulation accuracy was achieved with specific compartment radii in different breast tissue regions (adipose, retroareolar, dense fibroglandular).

Conclusions:

• The evaluated synthesis method successfully simulates the texture of clinical mammograms.
• This technique holds potential for generating realistic synthetic mammograms for research and training.
• The findings support the hypothesis that mammogram texture reflects underlying 3D tissue distribution.