GAURI relates to a method and a system that uses adaptive pattern recognition and learning schemes for predictive single or differential diagnosis, designed with magnetic resonance imaging modalities and neuropsychological batteries. This invention employs principles of pattern recognition and machine learning to analyze variations of neurochemicals under the controlled and diseased conditions. The integrated features in this embodiment comprises of processed proton (1H), Phosphorus (31P) or other multi-nuclei Magnetic Resonance Spectroscopy (MRS) data with interest metabolite peak areas, quantitative metabolite concentrations, anatomical features from processed and segmented MRI, neuropsychological evaluation scores for statistical analysis and metabolite pattern recognition with results in order to generate reliable multi-parametric diagnostic maps. The system and methodology developed under this claim integrates uni- or multi-modal non-invasive magnetic resonance imaging data from various sources for statistical analysis to filter out good biomarkers, which can also be used with other imaging modalities. This claimed methodology and system also performs statistical analysis and receiver operator characteristic (ROC) analysis for determining the diagnostic effectiveness of the characteristic neurochemicals as disease specific biomarkers. Data under analysis can be controlled for outlier detection and elimination preferably by determining cut-off thresholds. This methodology can be used to carry out cross-sectional, longitudinal studies and disease prognosis analysis.