Title: Faster Magnetic Resonance Imaging (MRI) While Improving Signal-to-Noise Ratio (SNR): The Best of Both Worlds
MRI scanning is well known to for its superior tissue image contrast, excellent spatial resolution and lack of ionizing radiation. It is the diagnostic imaging modality of choice for brain imaging and paediatric imaging. One of the notable problems with MRI, however, is the length of time for scanning. Long scans (often minutes long) are required due to the relatively low thermal equilibrium spin polarization (~ 0.0003% for 1H per Tesla, at 37oC). Long scan times increase image unsharpness, and thus reduce image quality, simply because of patient motion over the scan time. To improve polarization, and thus improve SNR, larger magnetic field strength MRIs have been developed. There are now over 75 seven Tesla scanners globally and a handful of 9T and 11T systems also exist. However, improved SNR through this approach is expensive (approximately $1million per Tesla). Furthermore, an increase in field strength also results in increased tissue heating (specific absorption rate, SAR) due to RF, excessive magnetic field (Bo and B1+) inhomogeneities and prolonged NMR T1 relaxation times (which necessitate increased scan times). Thus, although higher magnetic fields appear attractive they can result in even more difficulties. Alternatively, new approaches using routine clinical strength MRI scanners (i.e. mostly 1.5T and 3T systems) have been developed that are now being applied. The first approach, compressed sensing, takes into consideration the sparsity of images and involves pseudo-stochastic undersampling of acquisition data and subsequent iterative image reconstruction. The approach results in faster image acquisition with improved SNR. The second approach to improving speed and SNR focuses on increasing spin polarization of nuclei within endogenous molecules (generally referred to as ‘hyperpolarization’). In human MRI two techniques are currently used: dynamic nuclear polarization (DNP) and spin exchange optical pumping (SEOP). The DNP approach is used predominantly for 13C enriched molecules while SEOP is used for 3He and 129Xe lung imaging. In this talk I will present and overview of compressed sensing and hyperpolarzation approaches we are currently using to improve speed and SNR in MRI scanning.
Host: Dr.Soo Hyun Byun