Ultrasonic multi-feature analysis procedure for computer-aided diagnosis of solid breast lesions

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Oncology

Background:

• Accurate noninvasive identification of cancerous breast lesions is crucial for timely treatment.
• Current methods may lack objectivity and rely on subjective interpretation.
• Quantitative descriptors offer a potential solution for objective lesion assessment.

Purpose of the Study:

• To develop and validate quantitative descriptors for objective, noninvasive identification of cancerous breast lesions.
• To assess the diagnostic performance of combined acoustic and morphometric features.
• To compare quantitative descriptors with existing breast imaging criteria.

Main Methods:

• Quantitative acoustic features derived from spectrum analysis of ultrasonic radiofrequency (rf) echo signals.
• Morphometric features quantified using geometric and fractal analysis of lesion boundaries.
• Sliding-window Fourier analysis applied to rf echo signals for spectral-parameter mapping.

Main Results:

• Multi-feature analysis demonstrated excellent discrimination between cancerous and noncancerous lesions.
• An area under the receiver-operating-characteristics (ROC) curve of 0.947 +/- 0.045 was achieved across 130 patients.
• Lesion-margin definition, spiculation, and border irregularity were identified as highly useful quantitative descriptors.

Conclusions:

• Quantitative descriptors, particularly those analyzing lesion margins and shape, provide objective and effective means for breast lesion classification.
• The developed multi-feature analysis aligns with established BI-RADS criteria, enhancing diagnostic consistency.
• This approach holds promise for improving the accuracy and objectivity of breast cancer diagnosis via ultrasound.