[Support vector machine based high intensity focused ultrasound beam lesion degree classification and recognition]

Area of Science:

• Medical imaging
• Acoustic physics
• Biomedical engineering

Context:

• Non-invasive detection of thermal lesions is crucial for high-intensity focused ultrasound (HIFU) clinical applications.
• Current methods for monitoring HIFU-induced tissue changes can be limited.
• Accurate real-time feedback is needed to optimize HIFU therapies.

Purpose:

• To develop and validate a sub-pixel method for quantifying ultrasound image changes caused by HIFU.
• To utilize correlation distance as a metric for detecting thermal lesions.
• To employ a support vector machine (SVM) for classifying lesion characteristics.

Summary:

• A novel sub-pixel cross-correlation method was developed to measure ultrasound image alterations resulting from HIFU.
• Correlation distance was used as the primary feature to train a support vector machine (SVM) classifier.
• The study demonstrated that the sub-pixel method effectively identifies the position of HIFU-induced ablation lesions.

Impact:

• This technique offers a non-invasive approach for monitoring HIFU treatments.
• The SVM-based classification accurately determines the degree of HIFU beam lesions.
• Potential to improve the safety and efficacy of HIFU therapies through precise lesion monitoring.