Compressed Sensing 5D Spectroscopic Imaging of Perinatally HIV-Infected Youth

University Of California Los Angeles, Los Angeles CA

Investigators

Linked publications, trials & patents

Abstract

DESCRIPTION (provided by applicant): Brain abnormalities occur in HIV-infected individuals (adults and children) despite highly active antiretroviral treatment. Hence, noninvasive neuroimaging tools could assess the effectiveness of these treatments in preserving brain health and possibly lead to improvements in therapy. One-dimensional (1D) MR Spectroscopy (MRS) usually enables study of only six cerebral metabolites due to limited spectral dispersion even with 3 Tesla MRI scanners. Recently, more than fifteen cerebral metabolites have been quantified non-invasively in the prefrontal dorsolateral white matter region of perinatally HIV-infected youth and healthy children including novel metabolites such as glutathione (GSH), aspartate (Asp) and scyllo-inositol using the home-developed two-dimensional (2D) localized correlated spectroscopy (L-COSY) sequence combined with the prior-knowledge fitting (ProFit) algorithm. However, the requirement of a bigger voxel (27ml) and longer acquisition times were major limitations. Hence, two novel four-dimensional (4D) multi-voxel based 2D MRS sequences, namely echo-planar J-resolved spectroscopic imaging (EP-JRESI) and echo-planar correlated spectroscopic imaging (EP-COSI) were recently implemented by our group where two spectroscopic dimensions were combined with two spatial dimensions. Previously, due to the required number of encoding steps for one of the spatial and the spectral dimensions, a total duration of approximately 30 minutes was necessary. The feasibility of enhancing the speed of MRI/MRSI and diffusion tensor imaging (DTI) using non uniform sampling (NUS) or sparse sampling along selected spatial and spectral dimensions using compressed sensing (CS) based reconstruction has recently been demonstrated. Hence, the proposed study will test the following hypotheses: 1) The NUS-based five-dimensional (5D) EP-JRESI data acquisition and CS-based reconstruction will shorten the total acquisition duration by at least 8 fold, improving clinical applicability, ad will offer similar spectral and spatial resolution. 2) DTI will show higher radial and axial diffuson (reflecting neuroinflammation) and lower fractional anisotropy (reflecting neuronal injury) in multiple brain regions as markers of microstructural abnormalities in the brains of perinatally HIV-infected youths (a population not yet studied using DTI). Three specific goals are proposed: 1) to optimize the 5D EP-JRESI sequence on a 3T MRI scanner in which NUS will be incorporated into 2 spatial and 1 spectral dimensions, and to further optimize the CS-based reconstruction of the 5D MRSI data using home developed MATLAB code; 2) to acquire multi-voxel 2D J-resolved spectra and 3D spectroscopic images of cerebral metabolites in HIV-infected and age/sex matched healthy youths; 3) to record DTI of the brain and to correlate DTI measures with metabolite ratios and other HIV disease variables. The ongoing brain injury will be assessed in perinatally HIV- infected youth compared to healthy youths. This novel technology may also be extended to adult HIV and will also permit improved brain assessments for other degenerative neurologic diseases.