Registration methods for gross Motion correction during image-guided kidney surgery
Glisson, Courtenay L; Ong, Rowena E; Simpson, Amber L; Burgner, Jessica; Lathrop, Ray A; Webster III, Robert J; Galloway, Robert L
International Journal of Computer Assisted Radiology and Surgery, 6 (S1), pp. 160–161, 2011
Abstract
The National Cancer Institute reported an increased incidence ofkidney cancer over the past 30 years, projecting over 54,000 newcases in 2008 [1]. Surgical intervention is the established method oftreatment, with partial nephrectomies preferred when physiologicallypractical [2, 3]. Along with nephron-sparing approaches, minimallyinvasive methods have gained ground due to reduced blood loss andfaster recovery time compared to open resections [4]. With lessinvasive approaches becoming more common, the incorporation ofquantitative guidance holds promise as surgeons strive toward closermargins and more spared healthy tissue.Image guidance uses a transformation linking preoperative image-space and physical-space, requiring collection of points in bothspaces. Though point-based methods are faster and more robust, non-rigid attachment of abdominal organs makes extrinsic, preoperativefiducials impractical, and intrinsic points are difficult to localize inboth image and physical-space. Thus surface-based registration hasproven useful in abdominal image guidance [5], despite limitationsinvolving symmetry and local minima convergence. Such methodsmay have particular trouble as surgery progresses and the kidney isshifted and deformed by clamping, icing, and cutting.To update an initial surface-based registration, a quick, reliable re-registration between perioperative and preoperative image-spacewould be advantageous. We previously accomplished this byextracting virtual fiducials from laser range scans (LRS) and usingpoint-based registrations to realign image and physical-space [6]. Theextension of this technique to a minimally invasive approach con-stitutes the next logical step. Conoscopic holography has shownpromise as a minimally invasive surface acquisition method. Theconoscope performs similarly to an LRS, acquiring a full point cloudfor the initial surface-based registration, and later obtaining virtualfiducial locations for intraprocedural re-registration. Previous studieshave shown that a conoscope can successfully be used for liverphantom registrations [7]. The current study evaluates the accuracy ofconoscope-based registration in phantoms and ex vivo porcine kid-neys, providing a preliminary validation of this method for minimally