Referenceless thermometry using radial basis function interpolation

The Proton Resonance Frequency (PRF) shift provide a method for temperature change measurements during thermotherapy. Conventional PRF thermometry works subtracting one or multiple baseline images. The method leads to artifacts caused by tissue motion and frequency drift. Various works estimating the background phase from each acquired image phase are present in literature. These algorithms are called 'referenceless' because they don't require any subtraction of baseline images for calculating temperature increment. Conventional referenceless methods estimate baseline image by fitting the background phase outside the heated region through a polynomial approach. In this work a background phase reconstruction method using Radial Basis Function (RBF) interpolation is presented. The effectiveness of the proposed approach has been demonstrated with Magnetic Resonance guided Focused UltraSound (MRgFUS) sonications on ex-vivo porcine muscle. A significant improvement of the proposed RBF referenceless method, against classical PRF and polynomial referenceless methods, has been demonstrated.