A Robust Method to Evaluate Daily Motion Correlation between Diaphragm Domes and Targets
Abstract
Purpose
Monitoring real-time pancreatic target motion during radiotherapy is challenging. The diaphragm can be tracked by ultrasound; however, published findings on its correlation with abdominal motion are inconsistent. We aimed to develop a robust algorithm to assess daily motion correlation between the diaphragm and pancreas.
Methods
An iterative algorithm was developed to detect both right and left diaphragmatic domes and pancreatic fiducial markers in each cone-beam computed tomography (CBCT) projection. In the first round, planning CT and contour information were used to initialize marker and dome positions in each projection. Gradient-based analyses then identified markers and domes, and curve fitting estimated expected locations for refinement in a second detection round. Eleven pancreatic cancer patients receiving breath-hold radiotherapy were enrolled; each had one to three fiducial markers implanted in the pancreas. CBCT was used for positioning and verification; each scan was acquired during two separate breath holds. In total, 123 CBCT scans were analyzed. Superior-inferior distances between the right or left dome and the markers were computed.
Results
Across 24,437 projections, fiducial markers, right dome, and left dome were detected in 23,891, 23,217, and 21,850 projections, respectively. The per-scan mean standard deviations of positions were 1.4, 1.7, and 2.7 mm, and the mean inter-breath-hold gaps were 2.2, 2.9, and 3.5 mm for markers, right dome, and left dome, respectively. Dome-to-marker distances were analyzed in 22,076 (right) and 21,533 (left) projections, with mean standard deviations of 1.6 and 1.7 mm and mean inter-breath-hold gaps of 2.5 and 2.9 mm, respectively.
Conclusion
This algorithm enables routine assessment of residual motion and diaphragm-marker correlation during breath-hold CBCT. In this preliminary cohort, variations in marker-diaphragm distances were smaller than variations in diaphragmatic dome positions, indicating a reliable correlation. The method is scalable to larger patient cohorts with implanted markers.