Breast density classification in mammograms: An investigation of encoding techniques in binary-based local patterns.

Area of Science:

• Medical Imaging
• Computer Vision
• Machine Learning

Background:

• Accurate breast density classification in mammograms is crucial for breast cancer risk assessment.
• Traditional channel encoding techniques have limitations in capturing complex textural patterns.

Purpose of the Study:

• To introduce and evaluate a novel local septenary pattern operator for breast density classification.
• To compare the performance of the proposed operator against existing local binary, ternary, and quinary encoding methods.

Main Methods:

• Development of a new seven-encoding technique, the local septenary pattern operator.
• Application and evaluation of local binary, ternary, quinary, and septenary operators on mammographic datasets.
• Performance assessment using classification accuracy metrics.

Main Results:

• The local septenary pattern operator demonstrated superior performance compared to other encoding techniques.
• Maximum classification accuracies of 83.3% (MIAS) and 80.5% (InBreast) were achieved with the proposed operator.
• The local quinary operator showed the closest performance, with accuracies of 82.1% (MIAS) and 80.1% (InBreast).

Conclusions:

• The proposed local septenary pattern operator is a robust and effective method for breast density classification.
• This new operator offers improved accuracy over existing methods, potentially enhancing mammographic analysis.