Self-supervised Learning in Medical Imaging - Theory and Applications (IN2107)
More information will be provided during an introduction meeting scheduled at 12/02/2024 at 15h via Zoom:
- https://tinyurl.com/SSLMEDIMG , Meeting ID: 651 0606 0734 Passcode: 819595
Self-supervised learning (SSL), described as “the dark matter of intelligence, overcomes the limitations of most learning methods, which require high-quality labelled data. This paradigm leverages large amounts of unlabeled data to learn meaningful representations through pretext tasks and enables models to generalise better and perform well on downstream tasks involving only a few labelled data. In medical imaging, image labels are usually very limited. Therefore, SSL is particularly valuable in this context. In a first step, the seminar will focus on the integration of state-of-the-art self-supervised learning techniques in computer vision. Next, those techniques will then be discussed in the context of medical problem settings featuring segmentation, detection, and classification. Selected material on methods and applications from the field of medical imaging will be covered. Basic problem formulations to recent advances will be discussed.
This includes, but is not limited to:
- Theory of Self-Supervised-Learning
- Examples of SSL data in medical imaging
- Clinical applications
Please register at: https://matching.in.tum.de/ or write an e-mail to maxime.di-folco@tum.de
Check the intro slides here: