Computer-assisted bone tumour ablation using sparse radiographs

Abstract

Percutaneous radiofrequency ablation (RFA) is a minimally invasive technique increasingly employed in the treatment of unresectable bone tumours due to its safety, less invasiveness, reduced complications, effectiveness, and predictability. This paper investigates a new pipeline of computer-assisted interventional system with navigation and planning functions for percutaneous RFA as an useful tool to improve clinicians’ performance. Unlike traditional open procedures, percutaneous RFA requires image guidance because a direct visual field of surgical targets is unavailable. Radiographic imaging can be used intra-operatively for navigation, but it generates radiation exposure to patients and clinicians. To reduce the risk of radiation exposure and enhance the performance in computer-assisted bone tumor RFA, this paper addressed two key problems: (1) statistical atlas is employed to construct a global navigation map by deformable registration using a minimal number of intra-operative fluoroscopic images to localize the tumor and (2) a multiple-objective optimization for treatment planning of large tumor RFA is proposed to obtain optimal trajectories and ablation coverage by considering clinical constraints. The proposed system was validated by experiments, showing that the construction error of patient-specific model was 0.84 mm in terms of surface-to-surface distance measure and the optimal planning achieved 100% tumor coverage under multiple clinical constraints.