Rapid Methods for 2D and 3D T1 Estimation using Highly Undersampled Radial MRI Data

Abstract

Magnetic Resonance Imaging (MRI) is a medical imaging modality routinely used in the clinic to obtain images of the human body. Conventional MRI is qualitative in nature. It is often up to the radiologists to interpret the images by qualitatively comparing relative signal intensities in different tissues. Since contrast in MRI is governed by the underlying physical parameters of tissues, it is possible to generate quantitative MRI images. Measurement of these physical parameters is referred to as parameter mapping. Parameter maps can provide valuable information for the characterization of disease. However, despite being used in a handful of applications, parameter maps are still not used routinely in the clinic. This is due to the fact that parameter mapping requires the acquisition of images at various contrasts, leading to increased scan times and making parameter mapping impractical for routine clinical work. In this dissertation, 2D and 3D accelerated parameter mapping methods for measuring the spin-lattice relaxation time (T1 ) using inversion recovery (IR) radial gradient echo sequences are presented. The 2D T1 mapping technique is based on an IR radial balanced steady state free precession (bSSFP) pulse sequence. A principle component (PC)-based reconstruction algorithm is also proposed for reconstruction of high fidelity T1 maps from the acquired undersampled data. This novel T1 mapping technique can acquire data needed to produce high fidelity T1 maps with sub-millimeter in-plane spatial resolution in less than 3 seconds/slice. The 3D technique extends the 2D technique by using a radial stack-of-stars (SOS) trajectory. The proposed technique increases the 3D acquisition efficiency by using an interleaved multi-slab scheme and can generate volumetric T1 maps with 1 mm isotropic resolution in less than 5 minutes. Both techniques are validated using phantom and in vivo studies.