matthewchung74/alzheimer-cnn-baseline

Alzheimer Disease MRI Classification (Baseline CNN)

This model is a baseline CNN for classifying Alzheimer's disease severity from MRI scans. It distinguishes between four classes: Mild Demented, Moderate Demented, Non Demented, and Very Mild Demented.

Model Architecture

Basic CNN using 2D convolutional layers, max pooling, and batch normalization.

Performance (Test Set)

• Accuracy: 0.9766
• Weighted Average F1-Score: 0.9764

Classification Report:

                    precision    recall  f1-score   support

     Mild Demented       1.00      0.92      0.96       172
 Moderate Demented       1.00      0.93      0.97        15
      Non Demented       0.97      1.00      0.98       634
Very Mild Demented       0.97      0.97      0.97       459

          accuracy                           0.98      1280
         macro avg       0.99      0.96      0.97      1280
      weighted avg       0.98      0.98      0.98      1280