Spinocerebellar ataxia type 12 (SCA12) is a neurodegenerative genetic disorder triggered by abnormal CAG repeat expansion at locus 5q32. MRI recognises dissimilarities in affected areas of SCA12 patients and healthy subjects. But manual diagnosis is time-consuming and prone to subjective errors. This has motivated us in developing a systematic and authentic decision model for computer-aided diagnosis (CAD) of SCA12. Four different feature extraction techniques are examined in this research work, such as First Order Statistics, GLRLM, GLCM, and GLGCM, to obtain distinguishable characteristics for differentiating SCA12 patients from healthy subjects. The model’s performance is measured using sensitivity, specificity, accuracy and F1-score with leave-one-out cross-validation scheme. Our experimental results show that features based on the GLRLM can distinguish SCA12 from healthy subjects with a maximum classification accuracy of 85% among all the different function extraction techniques used.