Power-law Beta Calculations for Realistic Mammography Images from Egs_cbct Simulation

Background: Mammography is the most effective method for the early detection of breast cancer. The glandular tissue in a mammogram acts as anatomic noise and limit the detection of suspicious lesions. Previous studies found that that the power-law form can be used to characterize anatomic noise in mammograms with β-exponent values ranging from 1.5 to 4.0. Purpose: The goal is to determine if the egs_cbct simulation code can successfully simulate the mammography images by extracting the β-values from them. This is done using the power spectrum to quantitatively evaluate and compare the β exponent of the egs_cbct images with previously reported values from the clinical data sets. Materials and Methods: The in silico imaging software tools developed for the Virtual Clinical Trial for Regulatory Evaluation project was used to model breast phantoms. The egs_cbct code and a mammography setup were used to simulate the images of 400 phantoms. The resulting images were analyzed using 200 regions of interest per image, and the anatomical noise power spectrum was computed. The Mammographic Accreditation Phantom (American College of Radiology [ACR]) was also simulated to compare the image quality with that of a mammography unit. Results: The obtained values correspond to those found in earlier research using authentic mammography images. The mean β-value for the simulated images was 3.42, with a maximum value of 4.41 and a minimum value of 2.36. The simulated ACR phantom was comparable to that of the mammography unit. Conclusions: Based on the β values, the egs_cbct software can successfully replicate mammography images.