Monday, December 15, 2014
1:00PM (334 JFB)
Title: Improved MR thermometry for transcranial magnetic resonance guided focused ultrasound
Magnetic resonance guided focused ultrasound (MRgFUS) is a totally non-invasive thermal treatment modality that is being investigated for various diseases and disorders. It has gained popularity because it combines the non-invasive and very precise delivery of thermal energy from FUS with the anatomical- and thermal-imaging capabilities of MR imaging. This dissertation presents original research that aim at improving MR temperature imaging (MRTI) in general, and specifically real-time large field-of-view (FOV) 3D MRTI for transcranial MRgFUS applications.
First, five different data sampling schemes utilizing variable density subsampling in the spatial frequency space (k-space) were developed and implemented in a fast MRI pulse sequence (segmented echo planar imaging - EPI). The combination of these subsampling schemes with a previously described temporally constrained reconstruction (TCR) method allows for fast and large FOV MRTI with improved temperature measurement accuracy, compared to standard subsampling schemes.
Secondly, a proof-of-concept study was performed to evaluate the transcranial treatment envelope (defined as intracranial areas where therapeutic levels of FUS can be delivered) utilizing 3D MRTI. It was shown that by using 3D MRTI to monitor the fully insonified FOV, improved safety can be achieved by monitoring of unintended power depositions in near-field tissue-bone interfaces.
Lastly, the accuracy of the model predictive filtering (MPF) method to reconstruct MRTI data, which utilizes a thermal model to predict missing data in k-space, was evaluated. The evaluation was performed through simulations and /ex-vivo /experiments in terms of model parameter accuracy, amount of k-space subsampling, and rate of the FUS induced temperature rise. It was shown that the MPF method can accurately (temperature root-mean-square-error < 0.7 °C) and in real-time (reconstruction time ~1 s) monitor a FOV large enough (192x162x96 mm) to cover the fully insonified volume in transcranial MRgFUS applications.