Speaker: Junyu Guo

Title: Improvement in Dynamic Magnetic Resonance Imaging Thermometry

Abstract:

Temperature monitoring using MRI becomes more crucial as thermal therapy is developing. The spin-lattice relaxation constant T1 diffusion coefficient D, the proton frequency shift (PRF) and other physical properties can be used as the temperature indicator in MRI. PRF method is the most efficient one among all these methods. But it is very sensitive to motion. Reducing scan time is very important to suppress the motion artifact and reach high temporal resolution in dynamic MRI. Parallel imaging can reduce scan time significantly using multiple coils.

This dissertation is focused on improving MRIT techniques. The application of the spin-lattice relaxation constant is investigated in which T1 is used as indicator to measure the temperature of flowing fluid such as blood. Problems associated with this technique are evaluated and a new method to improve the consistency and repeatability of T1 measurements is presented. A novel method called K-space Inherited Parallel Acquisition (KIPA) is developed to achieve faster dynamic temperature measurements. Artifacts in KIPA images are significantly reduced in comparison with those in GRAPPA images. The Root-Mean-Square (RMS) error of temperature for GRAPPA is 2 to 5 times larger than that for KIPA. The coefficient of variation for GRAPPA is 4 to 5 times larger than that for KIPA.

Finally, the accuracy and comparison of the effects of motion on three parallel imaging methods: SENSE, VSENSE and KIPA are investigated. According to the investigation, KIPA is the most accurate and robust method among all three methods for studies with or without motion. KIPA is less affected by the motion. The ratio of the normalized RMS (NRMS) error for SENSE to that for KIPA is within the range from 1 to 3.7. The ratio of the NRMS error for VSENSE to that for KIPA is about 1 to 2. These factors change with the reduction factor, motion and subject.