In the Department of Biomedical Engineering, the master’s student Najla Suhail Muzhir defended her thesis on Sunday, August 24, 2025, entitled:

“A Multimodal Deep Learning Based Scheme for Diagnosis”

The examining committee consisted of:

• Chairman: Prof. Dr. Nabeel Kazem Abd Al-Sahib

• Member: Prof. Dr. Ali Hussein Miri

• Member: Asst. Prof. Dr. Iman Ghadhban Khalil

The thesis was supervised by:

• Asst. Prof. Dr. Ahmed Faiq Hussein – College of Engineering / Al-Nahrain University

• Asst. Prof. Dr. Sufian Munther Saleh – College of Business Economics / Al-Nahrain University

The thesis was scientifically evaluated by:

• Scientific Evaluator 1: Asst. Prof. Dr. Mohammed Saadoun Hatheel – College of Engineering / University of Baghdad

• Scientific Evaluator 2: Asst. Prof. Dr. Israa Badr Nasser – College of Engineering / Al-Nahrain University

The thesis was linguistically evaluated by:

• Lecturer Dr. Noor Ali Sadiq

Thesis Summary

This study aims to develop a multimodal framework based on deep learning (DL) techniques to improve the accuracy of Alzheimer’s disease diagnosis through multimodal integration of Magnetic Resonance Imaging (MRI) and Electroencephalography (EEG) signals. The goal is to address the limitations of single-source models, enhance accuracy and reliability, and provide a more comprehensive understanding of Alzheimer’s disease.

The proposed system is based on three main models:

1. MRI Model: Based on VGG16 for extracting deep features from brain images.

2. EEG Model: Based on Convolutional Neural Networks (CNNs) for processing signals and extracting brain-related features.

3. Fusion Model: Employs feature-level fusion by combining the extracted feature vectors from both modalities into a single layer, which is then passed to a fully connected classification network for Alzheimer’s diagnosis.

Results

• The MRI-based classification system achieved an accuracy of 69%.

• The EEG model achieved a higher accuracy of 94%.

• The proposed fusion framework achieved the best performance with an accuracy of 95%, significantly outperforming single-modality models, thereby enhancing reliability and precision in Alzheimer’s disease diagnosis.

The thesis was accepted as it fulfilled the requirements for obtaining a Master’s degree in Biomedical Engineering.