Human immunodeficiency virus (HIV) affects more than 1 million individuals in the United States (US) and over 40 million people worldwide (1). Central nervous system (CNS) is commonly involved in the early stage of HIV infection. Magnetic resonance spectroscopy (MRS) offers a valuable method for monitoring HIV-associated neuropathologic changes (2-4). MRS measures may detect subtle early changes associated with HIV infection, and concentrations or ratios of cerebral metabolites measured by MRS could be used as a quantitative indicator of the CNS involvement (5, 6). Echo planar spectroscopic imaging (EPSI) enables acceleration of MR Spectroscopic Imaging (MRSI) using conventional phase-encoding gradients for spatial encoding. However, one of the drawbacks of the EPSI technique is due to lower sensitivity compared to conventional MRSI. Reduced MRS sensitivity of low concentrated brain metabolites limit the applicability of parallel imaging. Conversely, Compressed Sensing (CS) is a technique for accelerating the inherently slow data acquisition process, and is well suited for MRSI due to its intrinsic denoising effect. In this study, semi LASER based non-uniform undersampling (NUS) three dimensional (3D) EPSI was used to quantitate changes in brain metabolites, N-acetyl aspartate (NAA), total creatine (Cr), total-choline (Cho), glutamate and glutamine (Glx), and myo-inositol (mI) in a group of HIV-infected adults in comparison to a matched control group using compressed sensing reconstruction by minimizing total variation.The NUS based 3D EPSI sequence was evaluated cross-sectionally in 15 HIV-infected adults (mean age 45.1 years) and five healthy controls (mean age 51.0 years) using 3T MRI scanner (Siemens Medical Systems, Germany) with 16-channel head ‘receive’ coil. Reliability was assessed by test –retest studies of brain phantom (10 measurements). CS reconstruction was then performed by solving the total variation minimization problem using the linearized Bregman iteration. The 3D EPSI parameters were: FOV = 240x240x120 mm³, image matrix = 32x32x8, spectral width = 1190 Hz, number of spectral points = 256, TE = 41ms, TR = 1.5s, Avg=12. Data acquisition was 20 minutes long including water-suppressed (WS) and non-water-suppressed (NWS) scans. The NWS scan was used to perform eddy current and spectral phase correction. Acceleration by a factor of 4 (4X NUS) scheme was imposed along the two spatial dimensions (k_y and k_z). Extractable individual voxel volume in human brain was 0.84ml. The phantom, HIV and healthy brain 3D EPSI data were extracted and post-processed using homebuilt MATLAB-based (The Mathworks, Natick, MA, USA) library of programs. The full width at half maximum (FWHM) of HIV adults and healthy controls were approximately 18Hz. A p-value of <0.05 was considered significant.Fig.1a shows an axial MRI and semi LASER NUS based 3D EPSI voxel placement in the brain of a 50 year old HIV patient. Fig.1b.shows extracted spectra from the selected voxels marked in Fig.1a (0.84ml). Fig.2. shows the right frontal and right parietal metabolites ratios of HIV patients and healthy controls. Fig.3 shows the metabolites ratios of the right and left basal ganglia regions of HIV patients and healthy controls. Metabolite ratios with respect to Cr of semi-LASER NUS based 3D EPSI 4X compared with phantom concentration are also shown in Fig 4. Creatine and phosphocreatine are widely considered to be unchanged in numerous brain pathologies. However previous reports indicated changes in this metabolite levels. Metabolites were quantified in the following locations: right frontal, left frontal, right parietal, left parietal, right basal ganglia, left basal ganglia, right occipital, and left occipital. Significant metabolite changes were observed in HIV infected adults in the following locations: Declined NAA/Cr in the right frontal and right parietal elevated Cho/Cr in the right basal ganglia and elevated Glx/Cr and mI/Cr in the left basal ganglia. Similar changes were observed with the choline ratios also. By using NUS based 3D EPSI, we observed decreased NAA and increased mI, which provides evidence for both chronic neuroinflammation and neuronal injury, which is consistent with earlier findings (7). A decrease in NAA is detected in many diseases characterized by neuronal damage, such as stroke and multiple sclerosis (8, 9). Myoinositol is present mainly in glial cells and considered to be an important brain osmolyte. Increased Cho levels may be due to increased cell membrane turnover, which may result from cellular injury and increased glial proliferation. Glx, a major brain excitatory neurotransmitter, is associated with HIV-induced neurotoxicity where excessive activation of N-methyl-D-aspartate receptors results in increased extracellular Glx and neuronal cell death. This pilot study confirms evidence that there are neurometabolic differences between HIV adults and healthy controls using non-uniformly undersampled compressed sensing based EPSI.