DM-Density

DM-Density is a software application that provides objective, adjunctive information for breast tissue assessment by automatically calculating percent breast density and BI-RADS density categories from digital mammograms using proprietary area-based image analysis algorithms. Its methodology involves processing DICOM images from compatible mammography systems, and its performance has been established through non-clinical validation studies focusing on reliability, accuracy against expert radiologist assessments, and reproducibility.

DM-Density is a standalone software application designed to assist interpreting physicians in assessing breast tissue composition from full field digital mammograms. Its primary purpose is to automatically calculate percent breast density, defined as the ratio of fibroglandular tissue area to total breast area estimates, using proprietary algorithms to process "for presentation" digital mammograms from compatible Hologic systems (Selenia Dimensions and Lorad Selenia). The software generates numerical values for these areas and a BI-RADS breast density grade (supporting both 4th and 5th Editions), providing adjunctive information rather than being a diagnostic aid, with outputs available as DICOM Structured Reports or Secondary Captures for viewing on mammography workstations, PACS, or transmission to a RIS. The scientific basis of DM-Density involves area-based quantification of breast tissue components from mammographic images, with its performance evaluated through non-clinical testing against expert radiologist visual assessments, not human clinical trials. Validation studies confirmed its reliability using Pearson’s Correlation Coefficient for consistency in density, dense area, and total area measurements across different views and breasts. Accuracy was established for percent density and total area measurements, with dense breast area accuracy implied, and breast density grade assessments were validated against known grades using the Kappa statistic. The software also demonstrated reproducibility via Pearson's Correlation for images taken up to two years apart and confirmed the expected inverse correlation between breast density and patient age.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. How do the proprietary algorithms used by DM-Density differentiate between fibroglandular tissue and other breast tissues, and what is their robustness across varying image qualities?; Given the validation on specific Hologic systems (Selenia Dimensions and Lorad Selenia), what is the expected performance or process for validating DM-Density on other manufacturers' full field digital mammography systems?; What was the specific methodology, dataset size, and inter-rater reliability of the expert radiologist visual assessments against which DM-Density's accuracy (Kappa statistic for BI-RADS grade) was benchmarked?; Since DM-Density provides outputs based on both BI-RADS 4th and 5th editions, are there any studies comparing the clinical impact or radiologist concordance when using outputs from these different editions provided by the software?; The document states 'dense breast area accuracy follows by implication' from total area and percent density accuracy; what specific evidence or statistical reasoning supports this implication beyond direct validation of dense area measurement?