Model Predictive Filtering MR Temperature Imaging for Laser-Induced Interstitial Thermotherapy.

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Thermal Therapy

Background:

• Laser interstitial thermal therapy (LITT) requires precise temperature monitoring.
• Current magnetic resonance temperature imaging (MRTI) methods may lack sufficient temporal resolution.
• Improving MRTI temporal resolution is crucial for effective LITT ablation guidance.

Purpose of the Study:

• To evaluate the Model Predictive Filtering (MPF) method for enhancing MRTI temporal resolution.
• To assess MPF's utility in monitoring LITT ablations.
• To determine if MPF can improve volumetric temperature imaging during LITT.

Main Methods:

• Developed an MPF method using a Green's function approach and the Pennes bioheat equation.
• Modeled LITT heating and supplemented subsampled k-space data.
• Evaluated MPF with phantom and clinical LITT data, assessing reconstruction accuracy via thermal dose and hottest voxel analysis.

Main Results:

• MPF achieved high temporal resolution in phantom data, with good hottest voxel accuracy and thermal dose agreement (DSC ≥ 0.7).
• In vivo data showed excellent thermal dose agreement (DSC > 0.9) for orthogonal slices.
• MPF demonstrated high hottest voxel accuracy (RMSE ≤ 1°C) up to a subsampling factor of 3, maintaining accuracy up to 5.

Conclusions:

• The MPF algorithm enables large field-of-view volumetric temperature imaging without sacrificing temporal resolution.
• MPF shows good accuracy for in vivo LITT monitoring using bi-planar clinical data.
• MPF is a promising technique for improving MRTI during LITT procedures.