Brain MRI and CT images are widely used in radiation therapy. We process these image by computer vision and machine learning algorithms to accelerate the therapy planning process and make it less subjective.

Specific well studied biomarkers of cardiovascular diseases can be extracted from medical images. However, these manual measurements are time-consuming and rarely used in clinical settings. We aim to develop automatic CT processing methods and overall methodology to identify subjects at high risk for specific cardiovascular diseases.

State-of-the-art computer vision algorithms are believed to be a foundation for many AI applications in medicine and neuroscience. At the same time, a straightforward adaptation of 2D methods to 3D medical images such as CT and MRI doesn't work well in many scenarios. We work on two essential subproblems within this area of research: segmentation of 1D/2D objects in 3D medical images and estimation of models uncertainty.