Adaptive model-predictive controller for magnetic resonance guided focused ultrasound therapy

Area of Science:

• Medical Physics
• Oncology
• Biomedical Engineering

Background:

• Magnetic resonance image-guided focused ultrasound (MRgFUS) therapies face challenges in balancing treatment time with healthy tissue protection.
• Optimizing treatment delivery is crucial for clinical practicality and patient outcomes.

Purpose of the Study:

• To develop and validate an adaptive model-predictive controller (AMPC) for MRgFUS.
• To reduce MRgFUS treatment time while maintaining safety and efficacy.
• To enhance flexibility in MRgFUS treatment setup.

Main Methods:

• The AMPC models future temperatures and thermal dose accumulation to optimize switching between treatment cells.
• A configurable temperature constraint safeguards selected tissues.
• In vivo experiments in rabbit thighs were conducted to assess controller effectiveness and reliability.

Main Results:

• The AMPC successfully delivered the target thermal dose (≥240 CEM43) throughout treatment volumes in all in vivo experiments.
• Protected tissues were reliably maintained below 9 CEM43 by the safety temperature limit.
• Simulations showed path independence and significant time savings with a hottest-neighbor path strategy, reducing normal tissue heating and eliminating cooling periods.

Conclusions:

• Adaptive model-predictive control enables automated, safe, and effective MRgFUS treatments.
• The AMPC significantly reduces overall treatment duration for MRgFUS therapies.