Program Information
Masking for Gold: Automated Segmentation of Arbitrarily-Shaped Fiducial Markers in Cone-Beam CT Projections and Pseudo-4D Topograms
W.G. Campbell*, D.H. Thomas, M. Miften, B.L. Jones, University of Colorado School of Medicine, Aurora, CO
Presentations
TU-H-CAMPUS-JT-3 (Tuesday, August 1, 2017) 4:30 PM - 5:30 PM Room: Joint Imaging-Therapy ePoster Theater
Purpose: Fiducial markers are commonly used in radiotherapy to help visualize the target volume and accentuate tumors that suffer from poor contrast. Imaging of these markers allows for verification of target position and target motion prior to treatment. However, these highly radio-opaque objects can cause considerable artifacts in CT reconstructions, obscuring nearby structures. Here, we present an automated method of segmenting arbitrarily-shaped fiducial markers in projection data prior to reconstruction.
Methods: We developed a computational method to augment the appearance and identify the location of fiducial markers in CT projection data. Briefly, this method subtracts median filtered data respectively from unfiltered data, thereby enhancing outliers (e.g., small, dense markers). Using enhanced-outlier data, masks were then calculated to denote regions for replacement in unfiltered data, and a repeated subtraction of region-replaced data from original data produces images with highly enhanced markers. The method was tested with two data types: (1) patient CBCT scans featuring cylindrical and coil markers, and (2) pseudo-4D topograms (i.e., repeated topograms from helical CT, sorted according to breathing motion) of cylindrical markers in a phantom. For CBCT data, 2D median filtering was used for marker enhancement. For topograms, markers were enhanced using only slice-wise, 1D median filtering.
Results: The automated routine successfully segmented cylindrical and coil markers in CBCT projection images, allowing for moving metal artifact reduction to take place prior to reconstruction. The routine also successfully segmented markers in pseudo-4D topograms, automatically tracking fiducial marker motion with a helical CT scanner.
Conclusion: An automated, pre-reconstruction method of segmenting arbitrarily-shaped fiducial markers in CT projection data was developed. In addition to facilitating the reduction of moving metal artifacts, average marker locations can be superimposed onto subsequent reconstructions. Furthermore, automated tracking of fiducial markers in pseudo-4D topograms provide an assessment of target motion at the time of simulation.
Funding Support, Disclosures, and Conflict of Interest: Research funding for portions of this work were provided by Varian Medical Systems.
Contact Email: