Artificial Bee Colony Algorithm based Segnet Model for Structural Magnetic Resonance Image Segmentation in Alzheimer’s Disease Classification

Alzheimer’s disease is a chronic and incurable brain disorder. Alzheimer’s disease is preceded clinically by mild cognitive impairment. A prominent indicator of Alzheimer’s disease that is most commonly recognized is memory loss. Alzheimer’s disease can be identified using structural magnetic resonance imaging, which analyses the hippocampus’s volume and shape. These characteristics encode a small volume of data and are likely to cause segmentation mistakes. Hence, deep learning based feature extraction and segmentation are essential to improve the performance classification of Alzheimer’s disease. This paper proposes an Artificial Bee Colony based SegNet (ABC-SN) model for image segmentation in Alzheimer’s disease classification. The proposed framework uses the OASIS Alzheimer’s image dataset database for training, testing, and validation. The proposed ABC-SN model produces improved results of 91.2% specificity, 95.6% accuracy, 95% sensitivity, and 94% F1-score. The performance analysis shows convincingly that the proposed ABC-SN performs better segmentation compared with the existing threshold based segmentation, edge based segmentation, region based segmentation, and clustering based segmentation methods.