Kalman Filtered MR Temperature Imaging

David Fuentes¹, Joshua Yung¹, Andrew Elliott¹, John D. Hazle¹, Roger Jason Stafford¹

The presented work critically evaluates the ability of a Kalman Filtered MR thermal image acquisition scheme to accurately monitor a LITT procedure in the presence of corrupt or missing data. Details of the finite element-based stochastic form of the Pennes bioheat transfer model needed to achieve real-time performance within the Kalman framework are discussed. The ability to provide a robust temperature estimate in presence of data corruption was quantitatively evaluated in terms of an L2 (RMS) norm of the error. Results indicate the developed algorithm may provide a useful model-based estimate of the temperature state during a LITT procedure.