[Paper Review] Leveraging Deep Learning and Xception Architecture for High-Accuracy MRI Classification in Alzheimer Diagnosis

Shaojie Li, Haichen Qu+4 more|arXiv (Cornell University)|Mar 24, 2024

Brain Tumor Detection and Classification9 citations

TL;DR

Uses a pre-trained Xception-based CNN with MRI data to achieve 99.6% accuracy in four-class Alzheimer’s MRI grading. Demonstrates high precision, recall, and F1 across categories.

ABSTRACT

Exploring the application of deep learning technologies in the field of medical diagnostics, Magnetic Resonance Imaging (MRI) provides a unique perspective for observing and diagnosing complex neurodegenerative diseases such as Alzheimer Disease (AD). With advancements in deep learning, particularly in Convolutional Neural Networks (CNNs) and the Xception network architecture, we are now able to analyze and classify vast amounts of MRI data with unprecedented accuracy. The progress of this technology not only enhances our understanding of brain structural changes but also opens up new avenues for monitoring disease progression through non-invasive means and potentially allows for precise diagnosis in the early stages of the disease. This study aims to classify MRI images using deep learning models to identify different stages of Alzheimer Disease through a series of innovative data processing and model construction steps. Our experimental results show that the deep learning framework based on the Xception model achieved a 99.6% accuracy rate in the multi-class MRI image classification task, demonstrating its potential application value in assistive diagnosis. Future research will focus on expanding the dataset, improving model interpretability, and clinical validation to further promote the application of deep learning technology in the medical field, with the hope of bringing earlier diagnosis and more personalized treatment plans to Alzheimer Disease patients.

Motivation & Objective

• Investigate whether a deep learning framework can accurately classify MRI images into Alzheimer’s disease stage categories.
• Leverage the Xception architecture as a feature extractor to improve classification performance.
• Assess model robustness through data augmentation, regularization, and transfer learning techniques.

Proposed method

• Construct a custom CNN using pre-trained Xception as a feature extractor (include_top=False, pooling='max').
• Add Flatten, two Dropout layers (0.3 and 0.25), a 128-neuron Dense layer with ReLU, and a softmax output for four classes.
• Train with data augmentation, Adam optimizer, and categorical_crossentropy loss.
• Use early stopping and model checkpoints; evaluate with precision, recall, and F1-score.

Experimental results

Research questions

• RQ1 Can Xception-based architectures achieve high accuracy in multi-class MRI Alzheimer’s staging?
• RQ2 What is the impact of transfer learning and data augmentation on MRI-based AD classification performance?
• RQ3 How well does the model balance precision and recall across dementia stages?

Key findings

• The model achieves an overall accuracy of 99.6% on the test set.
• Per-class precision and recall are 1.00 for Mild and Moderate Demented classes and 0.99 for Non Demented and Very Mild Demented classes.
• F1-scores are 1.00 for Mild and Moderate Demented, and 0.99 for Non Demented and Very Mild Demented.
• Table 1 shows class-wise precision, recall, F1-score, and support values (Mild Demented: 2693; Moderate Demented: 1977; Non Demented: 2811; Very Mild Demented: 2715).
• The macro and weighted averages of metrics approach 1, indicating balanced performance across classes.