Quantitative assessment of breast density from digitized mammograms into Tabar's patterns

Area of Science:

• Radiology
• Medical Imaging Analysis
• Quantitative Breast Density Assessment

Background:

• Accurate breast density assessment is crucial for mammography interpretation and breast cancer risk stratification.
• Subjective classification methods, like Tabar patterns, can exhibit interobserver variability.

Purpose of the Study:

• To develop and validate a semi-automated technique for quantitative breast density assessment from digitized mammograms.
• To compare the quantitative results with subjective Tabar pattern classification and assess observer agreement.

Main Methods:

• A MATLAB-based graphical user interface application was developed using interactive thresholding.
• The technique calculates breast density as the percentage of fibroglandular tissue area to total breast area.
• Validation involved assessing correlation with subjective classification, classification performance, interobserver, and intraobserver agreement.

Main Results:

• The quantitative assessment showed a strong correlation (r² = 0.92) with subjective radiologist classification.
• The technique correctly classified 83% of digitized mammograms based on Tabar patterns.
• Moderate interobserver agreement (kappa = 0.63) and good intraobserver agreement (kappa = 0.75) were observed.

Conclusions:

• The semi-automated technique provides a reliable quantitative measure of breast density.
• This method can serve as a valuable supplement to radiologists' subjective assessments for breast cancer risk stratification.
• The tool demonstrates potential for improving consistency in mammogram classification according to established patterns.