Ensemble Network for Alzheimer’s Disease Diagnosis

Alzheimer’s disease (AD) is a common progressive neurodegenerative disorder characterized by a gradual decline in patients’ behavior, memory, and cognition. However, due to the complexity and low quality of medical images, a single neural network may not fully leverage the relevant features within the images. To address this issue, this paper proposes a multi-network ensemble architecture aiming to improve diagnostic accuracy. In the proposed approach, we employ three different neural networks to process a medical image. Each network has a unique structure and feature extraction capability to better capture relevant information in the images, and their outputs are fused via a fusion layer to obtain the final consensus output. Additionally, we utilize cross-entropy as the loss function to measure the discrepancy between the integrated output and the ground truth, and optimize the network parameters through backpropagation. After the end-to-end training, the resulting model is capable of providing accurate diagnostic results. Experimental results demonstrate that this method can better capture relevant features in medical images, thereby enhancing diagnostic accuracy. It shows potential in the diagnosis of Alzheimer’s disease, providing physicians with a reliable auxiliary tool.