An approach to analyze the breast tissues in infrared images using nonlinear adaptive level sets and Riesz transform...

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Computational Pathology

Background:

• Breast thermography offers potential for early breast cancer detection by analyzing temperature variations.
• Infrared images present challenges like low contrast and noise, complicating accurate breast tissue delineation.

Purpose of the Study:

• To develop an automated segmentation technique for breast tissues in thermal images.
• To extract characteristic features for analyzing thermal variations in normal and abnormal breast tissues.

Main Methods:

• Utilized nonlinear adaptive level sets and Riesz transform for automated breast thermal image analysis.
• Applied wavelet denoising and contrast-limited adaptive histogram equalization for image enhancement.
• Integrated phase map into level set framework for boundary estimation and Riesz transform for feature extraction.

Main Results:

• Achieved a 38% improvement in signal-to-noise ratio and 6% increase in image sharpness.
• Demonstrated high segmentation accuracy (98%) and regional overlap (99%) with ground truth.
• Identified significant differences (11%) in directionality between normal and abnormal breast tissues.

Conclusions:

• The proposed framework, with preprocessing, effectively aids in the early diagnosis of breast abnormalities using thermal imaging.
• The automated analysis enhances diagnostic capabilities for breast cancer detection.