Phase Sensitive MRI and its Application to Cardiovascular Diseases

Abstract

Inversion recovery phase sensitive (IRPS) imaging is a T1 weighted magnetic resonance imaging(MRI) method and its T1 contrast is introduced by a $180^{o}$ radio frequency(RF) pulse. IRPS image keeps the polarity map and has strong T1 contrast with flexible inversion time(TI). IRPS is used in a number of clinical applications such as myocardial scar detection and atherosclerosis evaluation. To reconstruct the IRPS images, however, a fully sampled reference scan is usually needed to help remove the background phase. This extra reference scan doubles the scan time and restrict its application to 3D imaging, especially the 3D myocardial late gadolinium enhancemen(LGE) imaging. Besides, this reference scan may introduce artifacts to the reconstructed phase sensitive images. Motivated by this issue, in the first part of this dissertation, 3D true phase polarity recovery with independent phase estimation using the three-layer stacks based region-growing (3D-TRIPS) was proposed to directly reconstruct the inversion recovery phase sensitive images without the reference scan. Supported by theoretical simulation, a structure containing three-layer stacks is designed to optimize the region-growing track based on the signal intensity. 3D-TRIPS was evaluated on 15 patients and was compared with phase sensitive inversion recovery reconstruction(PSIR). The results show good polarity consistency between two reconstructions and 3D-TRIPS images have similar or better image quality relative to PSIR images. IRPS images provide two types of contrast: intensity contrast and polarity contrast. The binary polarity contrast of IRPS images has not been used to directly identify the tissue boundary. In the second part of this thesis, carotid lumen identified by the negative polarity map of simultaneous noncontrast angiography and intraplaque hemorrhage(SNAP) magnetic resonance angiography(MRA) was evaluated against 3D time of flight(TOF) and contrast enhanced MRA(CE-MRA). SNAP showed a good agreement with CE-MRA in measuring both lumen area and percent stenosis. And the binary polarity map of SNAP MRA is less sensitive to flow artifacts relative to the gradient map, which was used to identify the lumen boundary of TOF. This study shows the potential of IRPS imaging on automatic segmentation and quantification of tissues of interest assisted by its binary polarity map. Since the negative polarity map of SNAP MRA is proved to be reliable to identify the lumen boundary, in the last section of the dissertation, the lumen of SNAP MRA was firstly transferred to the magnitude part of the reference image to produce the grey-blood proton density(PD) weighted images which can be used to detect the calcified nodules near the thin fibrous cap. Besides, SNAP-motion sensitized driven equilibrium(MSDE) sequence was proposed to directly produce the grey blood PD weighted images on intracranial arteries. Furthermore, to separate the arteries from veins on SNAP, improved SNAP(iSNAP) was developed, and a binary polarity mask transferred from the difference between two different SNAP scan was used to suppress the venous lumen signal. In conclusion, phase of complex signal plays an important role in MRI, in terms of improving tissue contrast and differentiating moving and stationary tissues. But its application has been difficult because of background phase and complex structure and SNR distributions and therefore, extra scan is usually in need to help remove the background phase. This thesis works to improve the efficiency of phase sensitive reconstruction and acquisition and therefore allows more application scenarios of phase sensitive imaging.