A Fast and Accurate Contour Tracing (FACT) Method for Custom Electron Cutout Using a Beam’s Eye View (BEV) Camera
J Sohn1,2*, J Park3, S Kim1, (1) Virginia Commonwealth University, Richmond, VA, (2) Northwestern University Feinberg School of Medicine, Chicago, IL (3) Memorial Sloan Kettering Cancer Center, Basking Ridge, NJ
Presentations
PO-GePV-T-76 (Sunday, 7/10/2022) [Eastern Time (GMT-4)]
ePoster Forums
Purpose: To develop a novel methodology and software generating electron cutout shape from the clinical setup images in optically capturing the projected contour with a CCD camera mounted at the electron applicator.
Methods: In clinical electron beam treatment without CT imaging, the treatment target is marked directly on the patient’s skin by a physician. First, two patient setup images were taken at the treatment gantry angle using a CCD camera that has a single lens at the center. A 3D printed tray for the camera was mounted at either lower and upper scrapers of the applicator in order to capture two images at different depths. The region-of-interests (ROIs) were detected along the clinical target contour at each setup image by binarizing the images and then extracting the edges with skeletonization. And, each distance from the center of the ROIs at every 5 degrees has been calculated for both images and analyzed to reconstruct the target contour on the skin. The final cutout shape was then calculated with a magnification factor at 95 cm source-to-point distance. An in-house software for the camera calibration and image processing was developed with MATLAB for the study. The result was compared with the clinical cutout drawn in the Eclipse TPS and the correlation coefficient was calculated using the mutual information applied to the masked binary images.
Results: The resultant cutout shape showed good agreement with the clinical cutout in terms of similarity and correlation coefficient. The concave/convex body shape of the patient could be accounted for using the developed algorithm compared to the single setup image.
Conclusion: Electron cutout can be generated from the clinical setup using the proposed methods. It provides a more accurate cutout shape and automatic image process of cutout preparation.
Keywords
Treatment Techniques, Pattern Recognition, Image Processing
Taxonomy
TH- External Beam- Electrons: Development (new technology and techniques)
Contact Email