A probabilistic approach for breast boundary extraction in mammograms

Area of Science:

• Medical Imaging
• Computer Vision
• Biomedical Engineering

Background:

• Accurate breast boundary extraction is essential for mammogram analysis.
• Existing methods, including thresholding and active contour models, have limitations in automation and accuracy.

Purpose of the Study:

• To develop a robust probabilistic approach for breast boundary extraction.
• To overcome the limitations of existing threshold-based and active contour model methods.

Main Methods:

• Utilized local binary patterns (LBPs) to characterize pixel texture.
• Introduced a novel probability model to ensure boundary smoothness.
• Developed a probabilistic framework for automated breast boundary detection.

Main Results:

• Achieved 38% improvement over active contour models.
• Demonstrated a 50% improvement compared to threshold-based methods.
• Increased the stability of breast boundary extraction by up to 86%.

Conclusions:

• The proposed probabilistic method offers superior performance for breast boundary extraction.
• This approach enhances accuracy and stability in mammogram analysis.
• The method effectively addresses limitations of conventional techniques.