Quantitative breast mass classification based on the integration of B-mode features and strain features in...

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Diagnostic Systems

Background:

• Elastography is an advanced sonographic technique that captures tissue strain information, transforming it into images.
• Radiologists interpret elastographic images, where gray-scale distribution relates to Young's modulus, to detect pathological changes like scirrhous carcinoma.
• Current methods can be operator-dependent, necessitating automated procedures for reliable breast mass classification.

Purpose of the Study:

• To develop a computer-aided diagnosis (CAD) system for breast mass classification using elastographic images.
• To extract quantitative strain features from elastographic data to reduce operator dependence.
• To automatically classify breast masses by integrating strain and B-mode features.

Main Methods:

• A database of 45 malignant and 45 benign breast masses was utilized.
• Tumor segmentation on B-mode images defined tumor areas, which were mapped to elastographic images.
• Fuzzy c-means clustering classified gray-scale pixels (white, gray, black) to identify stiff tissues; quantitative strain features were extracted from the dark cluster.

Main Results:

• The proposed strain features alone achieved 80% accuracy, 80% sensitivity, and 80% specificity (Az=0.84).
• Combining strain features with B-mode features significantly improved classification performance (Az=0.93, p<0.05).

Conclusions:

• Quantified strain features from elastography show promise for breast mass analysis.
• Integrating strain features with B-mode imaging offers a robust approach for distinguishing malignant from benign tumors.