Supervisors: Dieter Schweizer, Prof. Orçun Göksel
Every year, thousands of people are diagnosed with cancer in brain and central nervous system. When treating it using high intensity focused ultrasound, an algorithm for navigation and targeting is required. To aid in navigation, this thesis investigates the use of markers on transcranial implants to estimate a coordinate transform from ultrasound transducer to marker reference frame. To this end, a simulator was implemented to provide ultrasound-like image data. Subsequently, a marker phantom was made, imaged using a Fukuda Denshi ultrasound imaging system (UF-760AG + FUT-CVA403-6P probe) and a semantic segmentation network trained on the resulting images to detect points, which lie on the marker. Finally, a pattern detection algorithm was implemented, which identi es the marker pattern from a point cloud. Subsequently, the performance of each of these steps was assessed and documented.