Minimally invasive navigated liver intervention : ultrasound-guided surgery and ablation validation

Navigated surgery is becoming widely used in applications such as neurosurgery and orthopedics, however it is still in the research phase for soft organs such as the liver. This is mainly because of the organ’s deformability which leads to substantial intraoperative anatomical changes. Particularly in minimally invasive interventions, where the surgeon has no direct access to the organ, image guided navigation can play a crucial role to guide the surgical instrument to the target tumor. In case of radiofrequency ablation, accurately targeting the tumor is not enough: the lack of quantitative feedback during the procedure is another problem which may lead to inefficient treatment. To maintain high navigation accuracy during minimally invasive liver interventions, we propose two new approaches based on tracked 2D ultrasound and focused on the tumor region. The aim is to compensate for intraoperative tumor shift, which can happen due to surgical manipulation or repositioning of the patient. The developed techniques combine segmentation and registration algorithms to estimate the actual tumor position during the surgical procedure. We tested the approaches in a needle navigation experiment on ex-vivo porcine liver. The results indicate that the methods can quickly and accurately compensate for changes in the tumor position, providing a mean needle insertion error of 1.4 ±0.8 mm using the first approach and 2.1 ±0.8 mm using the second. The developed techniques have the advantages of being radiationfree, based on the widely available 2D ultrasound, and repeatable throughout the surgery. So they present a practical solution to maintain high tumor targeting accuracy in navigated laparoscopic and radiofrequency ablation procedures. The second point we investigated is the ablation feedback. To enable quantitative assessment of ablated lesions based on monitoring images, we developed techniques to correlate ablation outcomes on different images (MRI and ultrasound) to actual necroses. Two approaches were developed: one is based on a geometrical phantom and was applied ex-vivo, the other is based on electromagnetic tracking and was tested in-vivo. Results of the experiments showed that the feedback given to the surgeon, upon which the extent of the ablation is determined, varies depending on the imaging modality. This emphasizes the importance of such validation schemes and the need to apply them systematically to obtain a larger dataset. The ultimate goal is to standardize the ablation monitoring procedure and eventually improve the radiofrequency ablation recurrence rate.