Intelligent Alzheimer's Disease Identification using Hybrid Inception V3 with MLP models

Alzheimer's Disease (AD), an increasing neurodegenerative disorder, presents a significant diagnostic challenge. The proposed method utilizes deep learning (DL) to identify crucial features and machine learning (ML) to classify diseases. The aim is to improve diagnostic accuracy. This study presents a smart system for identifying Alzheimer's disease (AD). It uses a hybrid model that combines Inception V3 (IV3) DL features with Multi-Layer Perceptron (MLP). The proposed experiment utilizes IV3 for feature extraction and classification using MLP models with various hidden layer (HL) configurations. It uses metrics such as AUC, CA, Precision, F1-score, and recall for evaluation. The IV3 + MLP (100x100) model has an AUC of 0.967. Its accuracy is 86.3%. The 50x50 configuration shows good performance. It has an AUC of 0.964. The accuracy is 86.0%. This indicates a good balance between model generalization and complexity. The study finds that larger hidden layers in MLP architectures improve learning. They also enhance accuracy. The study compares MLP models using Inception V3 features. It finds that larger NN's architecture leads to better performance. The model IV3 + NN (70 70) performed well. It achieved an AUC of 0.962 with a CA, F1-score, Precision, and Recall all at 0.854. The IV3 + NN (50 50) model is smaller in size.