Machine learning meets medical imaging: From signals to clinically useful information

Daniel Rueckert, Professor of Visual Information Processing, Imperial College London Presents... Machine learning meets medical imaging: From signals to clinically useful information Three-dimensional (3D) and four-dimensional (4D) imaging plays an increasingly important role in computer-assisted diagnosis, intervention and therapy. However, in many cases the interpretation of these images is heavily dependent on the subjective assessment of the imaging data by clinicians. Over the last decades image registration and segmentation techniques have transformed the clinical workflow in many areas of medical imaging. At the same time, advances in machine learning have transformed many of the classical problems in computer vision into machine learning problems. This talk will focus on the convergence medical imaging and machine learning techniques for the discovery and quantification of clinically useful information from medical images: The first part of the talk will describe machine learning techniques such a dictionary learning that can be used for image reconstruction, e.g. the acceleration of MR imaging. The second part will discuss various model-based approaches that employ statistical as well as probabilistic approaches for segmentation. In particular, we will focus on atlas-based segmentation approaches that employ advanced machine learning approaches such as manifold learning and classifier fusion to improve the accuracy and robustness of the segmentation approaches.