Semi-Automatic Radiofrequency Ablation Planning Based on Constrained Clustering Process for Hepatic Tumors

Radiofrequency ablation (RFA) is currently one of the most effective methods for minimally invasive treatment of hepatic tumors. Planning the probe placements is an essential and challenging step in RFA treatment. To completely destroy the tumor with minimum amount of affected native tissue, a new RFA planning system is proposed in this paper. In the proposed planning system, the minimum number of ablations and a conical insertion region for each ablation session are determined automatically. Based on the geometric character of the tumor, a novel clustering algorithm is developed to allow a better layout of the overlapping ablations. For each case, we force the clustering process under the constraint of a manually defined puncture scope, such that all of the needle trajectories are gathered in a reasonable region. Moreover, the proposed planning system enables the clinician to manually choose a proper insertion path inside the conical insertion region to avoid penetrating large vessels or ribs, which is critical in RFA treatment. The proposed planning system was evaluated on 18 CT scan images and two clinical cases. Results implied that the planning system could provide feasible and accurate RFA treatment plans for hepatic tumors.