Purpose: Development of new needle insertion force feedback algorithms requires comparison with a gold standard method. A new evaluation framework was formulated and tested on needle punctures for percutaneous transhepatic catheter drainage (PTCD).
Methods: Needle insertion is an established procedure for minimally invasive interventions in the liver. Up-to-date, needle insertions are precisely planned using 2D axial CT slices from 3D data sets. To provide a 3D virtual reality and haptic training and planning environment, the full segmentation of patient data is often a mandatory step. To lessen the time required for manual segmentation, we propose direct haptic volume-rendering based on CT gray values and partially segmented patient data. The core contribution is a new force output evaluation method driven by a ray-casting technique that defines paths from the skin to target structures, i.e., the right hepatic duct near the juncture with the common hepatic duct. A ray-casting method computes insertion trajectories from the skin to the duct considering no-go structures and plausibility criteria. A rating system scores each trajectory. Finally, the best insertion trajectories are selected that reach the target. Along the selected paths, force output comparison between a reference system and the new haptic force output algorithm is carried out, quantified and visualized.
Results: The evaluation framework is presented along with an exemplary study of the liver using the atlas data set from a reference patient. In a comparison of our reference method to a newer algorithm, force outputs are found to be similar in 99% of the paths.
Conclusion: The proposed evaluation framework allows reliable detection of problematic PTCD trajectories and provides valuable hints to improve force feedback algorithm development.