Classification of Microcalcification Clusters in Digital Mammograms Using a Stack Generalization Based Classifier.

Area of Science:

• Medical Imaging
• Machine Learning
• Biomedical Engineering

Background:

• Microcalcification (MC) clusters in digital mammograms are crucial indicators for breast cancer detection.
• Accurate differentiation between malignant and benign MCs is essential for effective diagnosis and treatment planning.

Purpose of the Study:

• To develop and evaluate a machine learning-based approach for discriminating malignant and benign microcalcification clusters.
• To extract and select relevant features for improved classification accuracy.

Main Methods:

• Applied morphological operations for feature extraction from segmented microcalcifications.
• Extracted morphological, texture, and distribution features from individual MC components and clusters.
• Utilized a correlation-based feature selection technique and stack generalization classifier.

Main Results:

• Achieved a best classification accuracy of 95.00 ± 0.57% on the OPTIMAM database.
• Obtained an Area Under the Curve (A z) value of 0.97 ± 0.01.
• Validated the method on three diverse mammography databases (OMI-DB, DDSM, MIAS).

Conclusions:

• The proposed machine learning approach demonstrates high efficacy in classifying microcalcification clusters.
• The selected features and classification method hold clinical relevance for improving mammogram analysis.
• This technique offers a promising tool for enhancing the accuracy of breast cancer detection.