The research group of Prof. Jan Kybic is offering 2 PhD and 2 Postdoc positions in the field of medical image processing.
The project, Embryo quality estimation from time lapse images for assisted reproduction, supported by the Czech Health Research Council (AZV ČR), focuses on predicting the genetic quality of human embryos from time-lapse videos of embryo development. This includes, among other things, the problem of automatic classification of developmental stages, utilization of temporal coherence, compensation for differences between imaging systems, automatic segmentation of individual parts, creation of clinically motivated and automatically learned descriptors, and the combination of all relevant data for the final prediction. This project is primarily focused on the application and adaptation of existing methods to new problems.
The project, “Leveraging expert knowledge for medical image segmentation,” supported by the Czech Science Foundation (GAČR), will seek to improve the quality and efficiency of automatic segmentation using expert-defined constraints concerning invariance, topology, dimensions, position, number, shape, and other properties. New methods should be more robust and require a smaller volume of training data. We will attempt to connect deep learning-based segmentation methods with optimization methods under constraints, or alternatively, ensure that conditions are met through suitable parameterization or network architecture. This project is focused more on the development of new methods than on a specific application and is therefore suitable for candidates with deeper knowledge of mathematics and algorithmization.
Both projects require a good knowledge of programming and methods for image processing and machine learning, including deep learning. Start date as soon as possible.