Characterizing the clustered microcalcifications on mammograms to predict the pathological classification and...

Area of Science:

• Medical Imaging
• Radiology
• Oncology

Background:

• Mammography is crucial for detecting breast cancer, often relying on microcalcification analysis.
• Accurate pathological classification and grading of microcalcifications are vital for effective treatment planning.
• Existing methods for microcalcification assessment can be subjective and require further refinement.

Purpose of the Study:

• To develop and validate a mathematical model for characterizing clustered microcalcifications on mammograms.
• To predict the pathological classification and grading of microcalcifications using a novel feature parameter.
• To assess the utility of this model in computer-aided diagnosis for breast cancer.

Main Methods:

• A database of 109 mammograms with pathologically diagnosed microcalcification clusters was analyzed.
• A feature parameter, the pattern form factor of microcalcification cluster (θ), was defined and calculated.
• A mathematical model (G = 6.438 + 1.186 × Ln <θ>) was derived and validated using retrospective and prospective data.

Main Results:

• A positive relationship was found between the pattern form factor (θ) and pathological grading (G) in retrospective cases.
• The model achieved an approximate evaluation accuracy of 77.42% in the prospective study.
• Binary prediction for benignancy and malignancy showed promising results with a receiver operating characteristic (ROC) value between 0.74351 and 0.79891.

Conclusions:

• The pattern form factor (θ) of clustered microcalcifications is a potential feature for predicting pathological grading and classification.
• This mathematical model offers a quantitative approach to microcalcification analysis on mammograms.
• The findings suggest the pattern form factor θ could be a valuable addition to computer-aided diagnosis systems for breast cancer.