Automatic radiofrequency ablation planning for liver tumors

Abstract

Radiofrequency ablation (RFA) is widely used in the treatment of liver tumors. Computer-aided planning is needed to preoperatively provide reliable paths for puncturing electrode needles into the treatment zone with multiple clinical constraints. Under the constraints, a genetic algorithm (GA)-based method was proposed to plan the optimal needle paths without passing the import tissues, and the produced ablation zone completely and conformably cover the tumor. In the proposed method, the appropriate paths between the treatment zone and the skin were first filtered in accordance with the constraints. Then the ablation zone model was determined for each appropriate path. Each point in the treatment zone was treated as a gene, and all genes were grouped as a chromosome. The path planning optimization could be regarded as the gene expression in a chromosome. On the basis of the filtered appropriate paths and the determined ablation zone, the optimal paths were obtained through GA. In the experiment, 32 tumors from nine patients were used to evaluate the proposed method. The resultant paths ensured no import tissues were passed, and the number of used electrode needles and damaged healthy tissues by the ablation zone was minimum. Therefore, the proposed method provides reliable electrode needle paths for physicians.