Implanted fiducial markers are often used in radiotherapy to facilitate accurate visualization and localization of tumors. Typically, such markers are used to aid daily patient positioning and to verify the target's position during treatment. These markers can also provide a wealth of information regarding tumor motion, yet determining their accurate position in thousands of images is often prohibitive. This work introduces a novel, automated method for identifying fiducial markers in planar x-ray imaging.Methods:
In brief, the method was performed as follows. First, using processed CBCT projection images, an automated routine of reconstruction, forward-projection, tracking, and stabilization generated static templates of the marker cluster at arbitrary viewing angles. Breathing data were then incorporated into the same routine, resulting in dynamic templates dependent on both viewing angle and breathing motion. Finally, marker clusters were tracked using normalized cross correlations between templates (either static or dynamic) and CBCT projection images. To quantify the accuracy of the technique, a phantom study was performed and markers were manually tracked by two users to compare the automated technique against human measurements. Then, 75 pretreatment CBCT scans of 15 pancreatic cancer patients were analyzed to test the automated technique under real-life conditions, including several challenging scenarios for tracking fiducial markers (e.g., extraneous metallic objects, field-of-view limitations, and marker migration).Results:
In phantom and patient studies, for both static and dynamic templates, the method automatically tracked visible marker clusters in 100% of projection images. For scans in which a phantom exhibited 0D, 1D, and 3D motion, the automated technique showed median errors of 39 μm, 53 μm, and 93 μm, respectively. Human precision was worse in comparison; median interobserver differences for single markers and for the averaged coordinates of four markers were 183 μm and 120 μm, respectively. In patient scans, the method was robust against a number of confounding factors. Automated tracking was performed accurately despite the presence of radio-opaque, nonmarker objects (e.g., metallic stents, surgical clips) in five patients. This success was attributed to the distinct appearance of clusters as a whole compared to individual markers. Dynamic templates produced higher cross-correlation scores than static templates in patients whose fiducial marker clusters exhibited considerable deformation or rotation during the breathing cycle. For other patients, no significant difference was seen between dynamic and static templates. Additionally, transient differences in the cross-correlation score identified instances where markers disappeared from view.Conclusions:
A novel, automated method for producing dynamic templates of fiducial marker clusters has been developed. Production of these templates automatically provides measurements of tumor motion that occurred during the CBCT scan that was used to produce them. Additionally, using these templates with intrafractional images could potentially allow for more robust real-time target tracking in radiotherapy.