Unscented Kalman Filtering for Real Time Thermometry During Laser Ablation Interventions

Area of Science:

• Biomedical Engineering
• Computational Biology
• Oncology

Background:

• Laser ablation is a minimally invasive technique for tumor treatment.
• Accurate temperature monitoring is crucial for effective and safe laser ablation.
• Real-time prediction of thermal effects is challenging due to limited sensor data.

Purpose of the Study:

• To develop and evaluate a data-assimilation Bayesian framework for laser ablation cancer therapy.
• To assess the performance of the Unscented Kalman Filter (UKF) in predicting tissue temperature.
• To analyze the impact of filter settings on the framework's predictive capabilities.

Main Methods:

• Implementation of a data-assimilation Bayesian framework.
• Utilizing the Unscented Kalman Filter (UKF) for nonlinear estimation of tissue temperature.
• Analysis of UKF performance based on varying time resolution, number, and location of temperature observations.

Main Results:

• The UKF-based prediction model demonstrates effectiveness in estimating tissue temperature during laser ablation.
• Framework performance is influenced by the time resolution of the filter.
• The number and location of temperature observations significantly affect prediction accuracy.

Conclusions:

• A data-assimilation Bayesian framework with UKF can effectively monitor and predict thermal effects in real-time during laser therapy.
• Optimizing filter settings is essential for enhancing the accuracy and reliability of thermal monitoring.
• This approach holds clinical relevance for improving tumor treatment outcomes.