Currently, the most common method for doctors to acquire surgical skills is to assist more experienced colleagues during procedures. Due to the safety of patients, high costs and efficiency, this form of training is the last one that is often questioned, and more and more emphasis is placed on training outside the operating room. Virtual reality (VR) surgical simulators are one of the methods of such education. They offer a fully customizable, predictable and repeatable training environment in which clinicians can acquire and improve their skills without endangering the health and lives of patients. However, due to the computing power of modern computers, such a simulation is still a compromise between its realism and real-time feasibility.
The project will investigate the application of Deep Learning in the context of real-time surgical simulation. Specifically, it will explore training of different kinds of deep neural networks, which will improve the simulation and visualization realism and computational performance of virtual surgical procedures such as minimally-invasive and robotic surgery, flexible endoscopy, cardiovascular interventions and/or ophthalmology. You will join a thriving AI research community with many ongoing projects in healthcare.
We are looking for candidates with an enthusiasm for research, multidisciplinary collaboration and tackling challenging problems through teamwork. We are targeting candidates with a PhD (recent or pending) in Computer Science and/or Machine Learning, but we would also consider exceptional STEM graduates providing that they have appropriate Computer Science skills. Ideal skills would include: programming (Python/C/C++) and ML frameworks (PyTorch/Tensorflow). Nice to have: 3D engines (Unity3D, Unreal Engine, Omniverse, Bullet Physics) and knowledge of version control tools such as Git.
You do not need to have a background in Surgical Simulation, but willingness to learn basic concepts and definitions will be required.