Alzheimer's and Parkinson's Detection with Video-Based Hybrid Deep Learning from Brain MRI

Area of Science:

• Neuroimaging and computational neuroscience
• Medical artificial intelligence
• Neurological disorder diagnostics

Background:

• Dementia, encompassing Alzheimer's Disease (AD) and Parkinson's Disease (PD), significantly impacts cognitive and physical functions.
• Accurate and early detection of AD and PD is crucial for effective patient management.
• Current diagnostic methods can be invasive or lack the sensitivity for early-stage detection.

Purpose of the Study:

• To develop and evaluate advanced deep learning models for detecting Alzheimer's Disease (AD), Parkinson's Disease (PD), and control subjects using brain MRI videos.
• To assess the efficacy of Convolutional Neural Network (CNN) combined with Long Short-Term Memory (LSTM) and Gated Recurrent Units (GRU) architectures for video-based neurological disorder classification.
• To create a user-friendly Graphical User Interface (GUI) for practical application of the developed diagnostic models.

Main Methods:

• Utilized a public dataset of brain Magnetic Resonance Imaging (MRI) scans for AD, PD, and control groups.
• Preprocessed MRI data to generate class-specific videos, from which short video clips were extracted.
• Implemented and trained deep learning models including LSTM, LSTM+GRU, and Deeper LSTM architectures on extracted features using a 50% training/50% validation split.
• Developed a Graphical User Interface (GUI) integrating the trained models for detection.

Main Results:

• The video-based deep learning models demonstrated high performance in classifying AD, PD, and control subjects.
• Achieved maximum accuracy of 99.67% and maximum specificity of 99.83%.
• The study highlighted the effectiveness of these architectures in achieving high performance even with limited training data (50%) and short video clips.

Conclusions:

• Video-based analysis of brain MRI using advanced deep learning architectures (CNN-LSTM, CNN-LSTM+GRU) is a highly effective method for diagnosing AD and PD.
• The developed models offer a sensitive and specific tool for early detection of these neurodegenerative diseases.
• The integrated GUI application provides a practical and user-friendly solution for clinical settings.