Real-time View Expansion and Visual Servoing for Image-guided Fetal Surgery
UCL was recently awarded £10million to develop better tools and imaging techniques that will improve the success of surgery and other therapies on unborn babies in collaboration with KU Leuven, Great Ormond Street Hospital and University College London Hospital. Key to this widely scoped endeavour is the interaction between the information provided by real-time imaging, notably fetoscopy, and the robotic actuators to be developed for fetal surgery. Although visual
servoing is a standard tool in traditional robotics, its application to robotic fetal surgery poses a number of challenges due in part to the large deformations happening in the surgical scene but also due to the small field of view of the fetoscope.
This PhD project will focus on the extraction of localisation information from the fetoscopic view. A first objective will be to register, in real-time, the consecutive images of the video stream so as to dynamically expand the field of view through the construction of a mosaic image. Visual servoing will then be built on top of the real-time registration. Applications include compensation of physician tremor and application of motion constraint to protect the uterine wall and fetal membrane. Close collaboration with the Mechanical Engineering department at KU Leuven, Belgium is to be expected. Candidates with a strong interest in the following areas are encouraged to apply: medical image registration, visual servoing, computer vision, translational applications of medical image computing, and real-time image processing.