Author: Olubunmi Odunola Aregbe, Clara Ferreira, Margaret Reynolds, David A. Sterling 👨🔬
Affiliation: University of Minnesota, University of Minnesota Physicians, Department of Radiation Oncology, University of Minnesota, Minneapolis 🌍
Purpose: Current best practices recommend daily imaging and planning for Tandem and Ring (T&R) HDR patient plans. Accurate target delineation is a critical, yet time consuming step in this process. This study aims to quantify the inter-fraction volume similarities in the high-risk clinical target volume (HR-CTV) contours and determine if this is a necessary step.
Methods: In this retrospective study, six randomized T&R patients were selected. Each patient had 5 sets of MRI studies—one for each treated fraction—and contours for each dataset. For each patient, each of the datasets were anonymized and registered with the first fraction. The HR-CTV contours from each dataset were transferred to a common structure set and exported to 3D Slicer, an open-source image analysis tool. A Dice similarity coefficient (DSC) statistical tool was used to measure how similar two sets of HR-CTVs were for each fraction compared to the first treated fraction, i.e. a DSC value of zero means no overlap and 1 indicates a perfect overlap of both structures.
Results: The DSCs indicated good agreement, with the minimum, maximum, mean and standard deviation values across all patients being 0.701, 0.914, 0.840, and 0.058 respectively. The mean values for each of the six patients were 0.868, 0.867, 0.815, 0.886, 0.845, and 0.799 with standard deviation values of 0.027, 0.026, 0.015, 0.024, 0.064 and 0.137, respectively.
Conclusion: Although the DSC indicated that HR-CTV treatment volumes stayed relatively consistent throughout a course of treatment, i.e. DSC > 0.7, the observed variability was sufficient to suggest that daily imaging and contouring remain an important step to ensure accurate treatments delivery. Further research with larger cohorts of patients would allow for the refinement of techniques to explore strategies for improving accuracy without compromising patient safety.