main package

main.main_DL module

Image binary classification using a custom-built Convolutional Neural Network. This python script trains and evaluates the performance of a custom-built Convolutional Neural Network. The Neural Network is trained using different types of Diffusion Tensor Images centered on the hippocampus region. The original dataset is augmented by applying random rotations, zooms and random contrast enhancements to the images by means of a custom-built function that uses Keras layers. This model uses the color channels as a way of avoiding the use of a 3D convolutional layer. As an option, the user can use pre-trained weights for the convolutional layers (previously trained on the same dataset). The subjects are divided in two groups: AD and CN, corresponding to subjects suffering from Alzheimer’s disease and control subjects respectively. The CNN is evaluated by means the following parameters: Accuracy and AUC (both on validation and test sets).

main.main_ML module

Image binary classification using Machine Learning. This python script evaluates the performance of a Random Forest classifier and a Support Vector Machine classifier on the feature extracted from the Diffusion Tensor Images. The subjects are divided in two groups: AD and CN, corresponding to subjects suffering from Alzheimer’s disease and control subjects respectively. As an option, the user can choose to perform a Principal Component Analysis (PCA) on the Random Forest classifier and a Recursive Feature Elimination (RFE) on the Support Vector Machine classifier. The binary classifiers are evaluated by means the following parameters: Accuracy, Precision, Recall and AUC.