Date of Graduation
Master of Natural and Applied Science in Computer Science
Deep Convolutional Neural Networks (CNNs) have become the go-to method for medical imaging classification on various imaging modalities for binary and multiclass problems. Deep CNNs extract spatial features from image data hierarchically, with deeper layers learning more relevant features for the classification application. The effectiveness of deep learning models are hampered by limited data sets, skewed class distributions, and the undesirable "black box" of neural networks, which decreases their understandability and usability in precision medicine applications. This thesis addresses the challenge of building an explainable deep learning model for a clinical application: predicting the severity of Alzheimer's disease (AD). AD is a progressive neurodegenerative disorder that affects the brain and could result in dementia. Early detection of AD is crucial for more precise treatment and enhanced patient outcomes. The diagnosis and prognosis of AD rely heavily on neuroimaging information, particularly Magnetic Resonance Imaging (MRI). The research developed a deep learning model framework that integrates a local data-driven interpretation method (SHapley Additive exPlanation values) to explain the relationship between the predicted AD severity from the neural network and the input MR brain image. This thesis addresses the skewed class distribution using the synthetic minority oversampling technique. To evaluate the performance of the proposed framework, the study performed a comparative analysis using three CNN models: DenseNet121, DenseNet169, and Inception-ResNet-v2. The framework shows high sensitivity and specificity in the test sample of subjects with varying levels of AD severity. To facilitate a better understanding of model performance, five key AD neurocognitive assessment outcome measures and the APOE genotype biomarker were correlated with model misclassifications.
deep learning, convolutional neural network, Alzheimer's disease, magnetic resonance imaging, transfer learning, data augmentation, explainability, classification, medical imaging.
Artificial Intelligence and Robotics | Biomedical Informatics | Computer Sciences | Data Science | Diagnosis | Investigative Techniques
© Godwin O. Ekuma
Ekuma, Godwin O., "An Explainable Deep Learning Prediction Model for Severity of Alzheimer's Disease From Brain Images" (2023). MSU Graduate Theses. 3886.