Tutorials

The first steps to build complex medical image classification pipelines can be difficult.

Here, several Jupyter Notebooks that aim to give an introduction into the core functionality of AUCMEDI, are presented. Even though deep learning models are easier to setup with AUCMEDI compared to TensorFlow, AUCMEDI is not less flexible. For this reason, rather than creating one huge “demonstration Notebook” several small Notebooks were created, so that the reader can pick the sections of interest.

Overview

Topic	Link
Three Pillars of AUCMEDI	tutorial01.ipynb
Data Loading from CSV	tutorial02.ipynb
Data Loading from Directories	tutorial03.ipynb
Architecture Selection	tutorial04.ipynb
Custom Image Preprocessing	tutorial05.ipynb
Data Exploration and Performance Evaluation	tutorial06.ipynb
Modern Image Augmentation	tutorial07.ipynb
Transfer Learning	tutorial08.ipynb
Early Stopping and Transfer Learning	tutorial09.ipynb
Explainable Artificial Intelligence	tutorial10.ipynb
Integrating Metadata into a Model	tutorial11.ipynb