Model Assisted Monitoring

Patient Specific Calibration

PRF based temperature imaging
Canine prostate model

Calibrated Vs Uncalibrated FEM Model prediction
Non-damaging laser test pulses may simultaneously be used for patient safety to verify the application position as well as used for calibration the parabolic bioheat transfer model on a patient specific basis. PDE constrained optimization problem techniques may be used to recover the underlying tissue biothermal and optical parameters. The adjoint based techniques typically involve O(1e3-1e4) parameters for the inverse problem. Calibrated bioheat transfer models are seen to reproduce the MR thermal imaging within lest than 2degC accuracy pointwise in space and time.

Kalman Filtered MRTI

Kalman parameter study

Kalman Uncertainty

Kalman Filter theory provides a rigorous mathematical framework for combining computer models of the bioheat transfer with MR thermal imaging to provide a robust estimate of the temperature state. Both uncertainty in the bioheat transfer model and MR temperature imaging acquisition may be exploited. Under this probabilistic framework, the mean and variance of temperature isotherms and resulting thermal dose may be estimated. The 95% confidence interval around the 57degC isotherm is seen to grow and shrink in the bottom right given a simulated data corruption.