Automatic path proposal computation for CT-guided percutaneous liver biopsy.

ABSTRACT

PURPOSE: To evaluate feasibility of automatic software-based path proposals for CT-guided percutaneous biopsies. METHODS: Thirty-three patients (60 [Formula see text] 12 years) referred for CT-guided biopsy of focal liver lesions were consecutively included. Pre-interventional CT and dedicated software (FraunhoferMeVis Pathfinder) were used for (semi)automatic segmentation of relevant structures. The software subsequently generated three path proposals in downward quality for CT-guided biopsy. Proposed needle paths were compared with consensus proposal of two experts (comparable, less suitable, not feasible). In case of comparable results, equivalent approach to software-based path proposal was used. Quality of segmentation process was evaluated (Likert scale, 1 [Formula see text] best, 6 [Formula see text] worst), and time for processing was registered. RESULTS: All biopsies were performed successfully without complications. In 91 % one of the three automatic path proposals was rated comparable to experts' proposal. None of the first proposals was rated not feasible, and 76 % were rated comparable to the experts' proposal. 7 % automatic path proposals were rated not feasible, all being second choice ([Formula see text]) or third choice ([Formula see text]). In 79 %, segmentation at least was good. Average total time for establishing automatic path proposal was 42 [Formula see text] 9 s. CONCLUSION: Automatic software-based path proposal for CT-guided liver biopsies in the majority provides path proposals that are easy to establish and comparable to experts' insertion trajectories.