Data augmentation using Diffusion Models: Case of Medical Imaging

Artificial Intelligence and particularly machine learning and deep learning have taken the world by storm. Nowadays it is used in many fields, ranging from manufacturing with vision-based defect inspection platforms (e.g., LandingLens by LandingAI), to voice assistants (e.g., Apple's Siri, Amazon's Alexa …), passing by medical diagnosis (e.g., brain tumor detection). Nevertheless, one important aspect to consider when using machine learning is data: how much (quality) data do you have? That is one of the issues faced particularly in medical imaging. Because annotated data is scarce, it makes it difficult to build good machine learning models.