Towards an image-guided navigation system for rectal cancer
Roeland Eppenga is a PhD student in the department Nanobiophysics. (Co)Supervisors are prof.dr. T.J.M. Ruers from the faculty of Science and Technology, dr. J.A. Nijkamp from the Aarhus University and dr. K.F. D. Kuhlmann from NKI/Antoni van Leeuwenhoek hospital.
The standard treatment for rectal cancer is the surgical resection of the entire rectum with its surrounding fat. Despite this radical procedure, local recurrence may still occur in up to 10% of patients and is often due to incomplete resection of the tumor area. To better assess tumor borders, image-guided navigation (IGN) can be used.
A common form of IGN is where the surgeon uses tracked surgical tools that are visualized on a navigation screen in real-time, relative to a 3D anatomical model created from preoperative imaging. Within the abdominal space, the non-rigid rectum can easily move and deform which can strongly affect the navigation accuracy when the 3D model is not updated accordingly. Because high navigation accuracy is especially important around the tumor area, it was investigated if real-time tumor tracking, using electromagnetic (EM) trackers placed close or inside the tumor, can facilitate accurate real-time updates of the tumor position and orientation within the 3D model.
Using an IGN system with wired EM tumor tracking, surgeons assessed proximal rectal tumor borders with a low median error of 3mm. The tracker placement and linking the tracking to the 3D model, however, took 30 minutes of intraoperative time. When using wireless trackers, that can be implanted before surgery, these steps can be done preoperatively after which the IGN system is ready for plug-and-play use in the operating room. Currently there is one commercially available wireless EM tracking system (WL-EMTS) used for tumor tracking but is designed for radiotherapy applications. It was investigated if this WL-EMTS can be used for IGN surgery as well. An IGN system was developed using the WL-EMTS for tumor tracking and an optical tracking system for tool tracking. With this IGN system it was possible to reduce the rate of incomplete resections of non-palpable breast phantom tumors by a factor of 3, compared to a clinically used wireless localization technique. Because the optical tracking system required line-of-sight between its camera and the surgical tools, the IGN system was upgraded to where the WL-EMTS tracked tumor ánd tools. Using this upgraded wireless IGN system, accurate navigated resections were possible on virtual lesions in ex vivo liver specimens, where surgeons cut with 0.1±2.3mm accuracy along a planned resection margin.
Ideally, the 3D model is projected onto the actual anatomy, such that the navigation screen can be omitted, improving usability. For laparoscopic surgeries this requires laparoscope camera calibration. In rectal surgery oblique-viewing laparoscopes are often used that allow for shifting the field of view by rotating their telescope without moving the laparoscope itself, which is especially useful in narrow operating areas. This telescope rotation, however, also changes camera parameters and would require recalibration of the laparoscope. To avoid recalibration upon telescope rotation, a camera model was developed modeling these parameter changes. The model was tested on two oblique-viewing laparoscope types and the results showed accurate reprojections of about 4 pixels (0.4mm) accuracy, independent on telescope rotation.