Particle Swarm Optimization Algorithm Based U-Net Model for MRI Segmentation in Alzheimer's Disease

Alzheimer's disease is becoming more common over time. It is a long-term brain disease that gradually impairs people's memory, thinking skills, and finally their ability to perform even the most basic tasks. Early detection can help reduce the progression of this condition because it grows severe before any obvious symptoms show up and produces a brain problem that cannot be treated with drugs or other treatments. This paper proposes a Particle Swarm Optimization Algorithm based U-Net model for image segmentation in Alzheimer's disease classification. The proposed framework uses the Kaggle Alzheimer's image dataset database for training, testing, and validation and shows improved results of 88% Dice Coefficient, 83% Jaccard Index, and 90% F1 Score when compared with Threshold-based-segmentation, Region-Based-segmentation and Clustering-based-segmentation methodologies.