Multimodal registration of intraoperative ultrasound and preoperative computed tomography imaging is the basis for percutaneous hepatic interventions. Currently a rigid registration is performed manually by the surgeon using vessel structures and other anatomical landmarks for visual guidance. In this work we are presenting our approach for automation of this intraoperative registration step. Our method relies on the vessel structures visible in contrast enhanced CT and ultrasound B-Mode volumes. Therefore, an important part of this work is the intra-operative segmentation of vessel structures from ultrasound. For the nine randomly chosen patients studied in this paper, our system achieves a 3.45 mm accuracy for the points used for registration (target registration error (TRE)) and 5.01 mm for other landmarks which were not used for the registration process (fiducial registration error (FRE)). Of a total number of 27 patients, our system worked successfully in 15 cases.
展开▼
