Medical diagnostics includes brain tumor and Tumor-like classification, which is a vital challenge, since conventional imaging modality such as Magnetic Resonance Imaging (MRI) is prone to issue in discriminate Tumor-like from Tumor. This study proposes a new methodology based on Magnetic Resonance Spectroscopy (MRS) for improving classification accuracy and reducing computational processing time. This methodology includes the preprocessing of MRS data, key MRS biomarkers features extraction and selection, as well as the implementation of rule-based classification (RBC) machine. This study is based on MRS data obtained from 45 patients et al.-Andalus Oncology Centre by using a 1.5 T scanner. Medical features were extracted, such as Lipid Lactate (LL), N-Acetyl Aspirate (NAA), Creatine, Choline and Myo-Inositol; and were ranked by their diagnostic index. The RBC approach which selected to reduce the processing time of the traditional MRI based approaches and the application of the proposed model resulted in a classification accuracy of 94.4%. Our results highlight the great potential of MRS in brain lesion classification that overcomes the limitations of MRI in the discrimination between Tumor and non-Tumor lesions. In future research they could validate the proposed approach on larger data sets and combine it with multimodal imaging modalities to further improve diagnostic precision and clinical utilization.