Semiautomatic estimation of breast density with DM-Scan software

Area of Science:

• Radiology
• Medical Imaging Analysis

Background:

• Accurate breast density assessment is crucial for mammography interpretation and breast cancer risk stratification.
• Visual estimation of breast density can be subjective and lead to inter-observer variability.

Purpose of the Study:

• To evaluate the reproducibility of breast density calculation using DM-Scan software.
• To compare the reproducibility of DM-Scan with traditional visual inspection methods.

Main Methods:

• 655 direct digital mammograms were analyzed by three radiologists using DM-Scan software.
• Inter- and intra-observer agreement for DM-Scan and visual inspection (Boyd and BI-RADS® scales) were calculated using intraclass correlation coefficient and Kappa index.

Main Results:

• DM-Scan demonstrated higher mean inter- and intra-observer agreement (0.92) compared to visual inspection.
• Visual inspection showed mean interobserver agreement of 0.876 (Boyd) and 0.823 (BI-RADS®).
• Visual inspection showed mean intraobserver agreement of 0.813 (Boyd) and 0.770 (BI-RADS®).

Conclusions:

• Semiautomatic breast density calculation with DM-Scan software is more reproducible than visual estimation.
• DM-Scan software reduces subjectivity and variability in breast density determination.