Purpose: To integrate an Electro-Magnetic Tracking enabled (EMT-enabled) brachytherapy afterloader with CT scan image for accurate implant localization and reconstruction.
Methods: Nine 6F-catheters (20cm long) and two gynecological applicators were placed in a custom-made phantom to test EMT-enabled technology. The applicators were reconstructed by forward stepping an EMT sensor mounted on the afterloader check cable at a speed of 1cm/s through each dwell position every 2mm and dwell time of 3s. A CT scan of the phantom inside a water cube was obtained (0.6mm slice thickness, 120kV & 100mAs). Iterative Closest Point (ICP) algorithm was used to compare EMT reconstructed implants in the following cases: A) before CT scan was performed (insertion direction of V-shape-arranged-catheters parallel to y-axis of Field Generator (FG)), B) after CT (parallel to y-axis), C) phantom flipped (parallel to z-axis), D) phantom flipped and FG on the side (parallel to x-axis). EMT to CT registration was done using ICP initialized with a rigid transformation matrix obtained from four calibration templates (each with a 5mm diameter ceramic marbles). The marbles’ centers in EMT were obtained by aligning a 3D scan cross sections of these templates to the EMT reconstructed catheters inside them. The marbles’ centers in CT were obtained by fitting a sphere to its edge pixels obtained by the Canny edge detector.
Results: Mean distance difference for EMT implant reconstruction between case A&B is: 0.35±0.16mm, B&C 1.35±0.74mm, B&D 1.26±0.61mm. Mean distance difference between EMT and CT registration for catheters and calibration points (except applicators) is: A) 1.35±0.75mm, B) 1.61±0.68mm, C) 0.77±0.49mm, D) 0.79±0.48mm.
Conclusion: All differences for the multiple cases are within the EMT system expected accuracy. The calibration template constitutes an efficient approach for accurate EMT and CT reference frame registration. Registration could be improved by using more calibration points and using robust ICP variants.
Funding Support, Disclosures, and Conflict of Interest: This work was supported by the National Sciences and Engineering Research Council of Canada (NSERC) via the NSERC-DG and NSERC Alliance grant program. This study was partially funded by Elekta.