K-t GRAPPA Accelerated MR Temperature Mapping

High spatiotemporal temperature mapping with high accuracy is required for thermaltherapy modalities, such as, HIFU(High Intensity Focused Ultrasound) ablation.PRF-based MR thermometry is widely used at high magnetic field for its highspatiotemporal resolution, excellent linearity with temperature and its non-reliability totissue type. However, the spatiotemporal resolution at the moment is still not high enoughfor real-time temperature monitoring of HIFU ablations.Reducing the amount of acquired data by a given reduction factor in k space andreconstruct the image through special method is a typical means to accelerate dataacquisition without loss the image quality for MRI. This kind of methods involve the use ofspecial reconstruction algorithms to reconstruct alias-free images with partially acquireddata， like TCR and direct estimation using k-space data. These methods requirepre-determined parameters and iteration process, which are not suitable for clinicalapplication.A variable density acquisition and k-t GRAPPA reconstruction scheme areproposed in this thesis, which does not require training data, sensitivity map calculation,iteration, or selecting filters/parameters by exploiting local correlations in both spatial andtemporal dimensions.Simulation studies, phantom experiments and in vivo experiments were performed. Insimulation studies, when the reduction factor is4, the root mean square error(RMSE) of theheating area is smaller than0.1°C. Five phantom heating experiments and five in vivoexperiments were performed on Philips Sonalleve MR-HIFU system, the power level andheating duration is different for each experiment. When the reduction factor is4, the RMSEof the heating area is smaller than0.3°C for phantom experiments. For in vivo experiments,the RMSE is smaller than1.1°C. The reconstruction time is less than1s for each timeframe(6slices). In conclusion, k-t GRAPPA accelerated MR temperature mapping has thepotential clinical applications.