Purpose: To describe approaches for determining optimal spatial and temporal resolutions for the proton resonance frequency shift method of quantitative magnetic resonance temperature imaging (MRTI) guidance of transurethral ultrasonic prostate ablation.
Materials and methods: Temperature distributions of two transurethral ultrasound applicators (90 degrees sectored tubular and planar arrays) for canine prostate ablation were measured via MRTI during in vivo sonication, and agree well with two-dimensional finite difference model simulations at various spatial resolutions. Measured temperature distributions establish the relevant signal-to-noise ratio (SNR) range for thermometry in an interventional MR scanner, and are reconstructed at different resolutions to compare resultant temperature measurements. Various temporal resolutions are calculated by averaging MRTI frames.
Results: When noise is added to simulated temperature distributions for tubular and planar applicators, the minimum root mean squared (RMS) error is achieved by reconstructing to pixel sizes of 1.9 and 1.7 mm, respectively. In in vivo measurements, low spatial resolution MRTI data are shown to reduce the noise without significantly affecting thermal dose calculations. Temporal resolution of 0.66 frames/minute leads to measurement errors of more than 12 degrees C during rapid heating.
Conclusion: Optimizing MRTI pixel size entails balancing large pixel SNR gain with accuracy in representing underlying temperature distributions.