Augmenting interventional ultrasound using statistical shape model for guiding percutaneous nephrolithotomy: Initial evaluation in pigs

Successful percutaneous nephrolithotomy (PCNL) highly depends on an accurate needle puncture of the kidney. This puncture step is challenging and must be performed under intraoperative image guidance. This paper presents an image guidance method for PCNL by augmenting interventional ultrasound (US) using a 3D statistical kidney model. First, a 3D statistical kidney model is built aforehand from a collection of aligned training shapes. Intraoperatively, a patient-specific 3D kidney model is reconstructed in a fine-tune way using sparse manually picked points from calibrated US at maximum exhalation. The US images are then augmented with the kidney model and a real-time tracked virtual needle. Under the augmented-US guidance, percutaneous renal puncture can be performed. Experiments results based on Wuzhishan pig data have validated the accuracy and inter-operator repeatability of the presented kidney model reconstruction. The feasibility and efficiency of the proposed augmented US-guided puncture for PCNL have been demonstrated in pigs. With careful setup by trained surgeons, our proposed guidance has the potential to reduce the occurrence of complications such as PCS access failure and bleeding compared with a traditional US-guided method.