Ensemble Deep Learning Denoising (EDLD) model and optimized OTSU segmentation for Alzheimer's disease diagnosis using MRI images.
Authors
Affiliations (1)
Affiliations (1)
- School of Computer Science Engineering and Information Systems, Vellore Institute of Technology, Vellore, India.
Abstract
Diagnosing Alzheimer's disease (AD) is necessary to determine treatment options. AD categorization using machine learning (ML) relies on difficult, manually specified features. The most important stage in AD diagnosis is denoising to restore image stability and quality. An ensemble image denoising technique that combines Attention Guided Convolutional Neural Network (AGCNN), Adaptive Denoising Autoencoder (ADAE), and Gaussian Deep Belief Network (GDBN) improves image denoising performance. The hybrid AGCNN reduces noise and aligns along the global route by combining global and local characteristics. In ADAE, the encoder learns picture representations using convolutional layers (CLs) while the decoder uses deconvolutional layers. In addition, the GDBN extends the standard Deep Belief Network (DBN) to Gaussian Restricted Boltzmann Machines (RBMs). Ensemble learning selects the approach with the greatest Peak Signal-to-Noise Ratio (PSNR) to integrate learning outcomes. After separating the background from the foreground by calculating the variances within the two groups, OTSU determines the threshold that minimizes the weighted sum of the variances. Levy Grasshopper Optimization Algorithm (LGOA) optimizes threshold selection by mimicking grasshopper swarming. VGG16, the DCNN model, is pre-trained for Alzheimer's datasets. The results are Sensitivity (SEN -95.86%), specificity (SPC - 94.93%), precision (PPV - 94.55%), F1-score (F1 - 95.21%), accuracy (ACC -95.87%), and Area Under the Receiver Operating Characteristic Curve (AUC - 96.45%) assess system and method performance.