Model based deep learning method for focused ultrasound pathway scanning.

Area of Science:

• Medical Physics
• Biomedical Engineering
• Computational Medicine

Background:

• High-intensity focused ultrasound (HIFU) is a non-invasive therapy for tumor ablation.
• Treating large tumors requires multiple ablations, necessitating precise planning to avoid side effects.
• Effective preoperative planning is crucial for optimizing HIFU efficacy and patient safety.

Purpose of the Study:

• To develop a machine learning-based approach for designing HIFU treatment plans.
• To integrate patient-specific material properties and thermal simulations for accurate treatment design.
• To enhance the precision of HIFU tumor ablation through improved preoperative planning.

Main Methods:

• A machine learning model was developed for HIFU treatment planning.
• A numerical model simulated HIFU absorption, heat transfer, and temperature rise.
• Ex vivo bovine liver experiments validated the numerical model's accuracy.

Main Results:

• The model accurately represents variations in tumor geometry.
• High-quality treatment plans were generated with minimal over- or under-treatment (<0.06%).
• The machine learning and simulation approach accurately predicted HIFU heating processes.

Conclusions:

• The proposed machine learning strategy enhances HIFU treatment planning.
• Accurate thermal simulation combined with machine learning improves presurgical planning.
• This approach has the potential to significantly improve HIFU therapy outcomes.