Premium
Accuracy of treatment planning based on stereolithography in computer assisted surgery a)
Author(s) -
Schicho Kurt,
Figl Michael,
Seemann Rudolf,
Ewers Rolf,
Lambrecht J. Thomas,
Wagner Arne,
Watzinger Franz,
Baumann Arnulf,
Kainberger Franz,
Fruehwald Julia,
Klug Clemens
Publication year - 2006
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.2242014
Subject(s) - image registration , fiducial marker , computer science , artificial intelligence , stereolithography , point set registration , radiation treatment planning , computer assisted surgery , computer vision , point (geometry) , mathematics , radiology , medicine , image (mathematics) , materials science , geometry , composite material , radiation therapy
Three‐dimensional stereolithographic models (SL models), made of solid acrylic resin derived from computed‐tomography (CT) data, are an established tool for preoperative treatment planning in numerous fields of medicine. An innovative approach, combining stereolithography with computer‐assisted point‐to‐point navigation, can support the precise surgical realization of a plan that has been defined on an SL model preoperatively. The essential prerequisites for the application of such an approach are: (1) The accuracy of the SL models (including accuracy of the CT scan and correspondence of the model with the patient's anatomy) and (2) the registration method used for the transfer of the plan from the SL model to the patient (i.e., whether the applied registration markers can be added to the SL model corresponding to the markers at the patient with an accuracy that keeps the “cumulative error” at the end of the chain of errors, in the order of the accuracy of contemporary navigation systems). In this study, we focus on these two topics: By applying image‐matching techniques, we fuse the original CT data of the patient with the corresponding CT data of the scanned SL model, and measure the deviations of defined parameter (e.g., distances between anatomical points). To evaluate the registration method used for the planning transfer, we apply a point‐merge algorithm, using four marker points that should be located at exactly corresponding positions at the patient and at connective bars that are added to the surface of the SL model. Again, deviations at defined anatomical structures are measured and analyzed statistically. Our results prove sufficient correspondence of the two data sets and accuracy of the registration method for routine clinical application. The evaluation of the SL model accuracy revealed an arithmetic mean of the relative deviations from 0.8% to 5.4%, with an overall mean deviation of 2.2%. Mean deviations of the investigated anatomical structures ranged from 0.8 mm to 3.2 mm . An overall mean (comprising all structures) of 2.5 mm was found. The fiducial registration error of the point‐merge algorithm ranged from 1.0 mm to 1.4 mm . The evaluated chain of errors showed a mean deviation of 2.5 mm . This study verifies that preoperative planning on SL models and intraoperative transfer of this plan with computer assisted navigation is a suitable and sufficiently reliable method for clinical applications.