RC 1205 - Artificial intelligence (AI) trustworthiness: from black boxes to glass boxes

1. To understand the basics of GAN and how they have been applied to radiology.
2. To discover how GAN can be used in data augmentation for training radiology AI models.
3. To learn about potential clinical applications of GAN in clinical routine workflows.

1. To understand the importance of high-quality annotation when developing AI models.
2. To assimilate how AI algorithms should report results and the associated confidence.
3. To discover the main limitations of narrow AI and the challenges towards general AI.

1. To understand the difference between supervised and non-supervised AI model creation.
2. To become familiar with manifold learning as a way to solve weak annotation in radiology.
3. To review examples of clinical applications using manifold learning.

1. To understand how to improve trustworthiness in AI models through interpretability and explainability.
2. To learn about the challenges of current AI models with regards to interpretability and explainability.
3. To discover the existing techniques to make AI more interpretable and understandable.