Hybrid Model based on ANFIS with Gray Level Co-occurrence Matrix for Dementia Prediction

Abstract

Dementia is a brain condition in which cognitive abilities decline more quickly than expected from the usual consequences of biological aging. It impacts memory and a person's physical and mental health. Early stages of dementia are challenging to anticipate, and there is presently no treatment for this condition. Therefore, a precise and prompt diagnosis of dementia is strongly advised in order to give the patient the best possible treatment. This study provides a hybrid model for the automatic diagnosis of dementia from T1-weighted magnetic resonance imaging (MRIs). The proposed model consists of two stages: the first step implements gray level co-occurrence matrix (GLCM) to extract texture features from imaging data, and the second step applies an adaptive neuro-fuzzy inference system (ANFIS) for the prediction of dementia from these extracted texture features. The proposed framework has been evaluated on the benchmark Dementia dataset comprising 5154 2D T1w MRI scans. In order to assess the model's performance, the proposed model is also compared with a neural network, fuzzy logic, and other machine learning (ML) techniques using the same dataset. The accuracy of the proposed model is recorded as 82.5%, which is greater than that attained by existing ML methods.