Mild Cognitive Impairment Conversion Prediction

Mild Cognitive Impairment (MCI) is a formative stage of dementia. Its detection may help clinicians to delay the progression. The conversion prediction of patients suffering from MCI to dementia can be done using the structural Magnetic Resonance Imaging data, using the gray matter. This work carries out an extensive empirical analysis exploring various feature extraction methods including Local Binary Pattern, Discrete Wavelet Transform, Histogram of Oriented Gradients, and Gray Level Co-occurrence Matrix along with forward feature selection. The work presents the findings by validating the various pipelines using the ADNI dataset. In total, 32 experiments were carried out. The combination of the Local Binary Pattern with Fisher Discriminant Ratio gives the best result with the linear kernel of SVM. We also explored the effect of the cost parameter on the recital of the proposed pipeline. The work has been compared with the existing works, and the results are encouraging.