Ozgur Ates, PhD, Ergun E. Ahunbay, PhD, Guang-Pei Chen, PhD, JingQiao Zhang, PhD, J. Frank Wilson, MD, X. Allen Li, PhD; Medical College of Wisconsin
BACKGROUND: MR-Linac, an integration of a magnetic resonance imaging (MRI) scanner with a Linac, has many advantages, such as excellent soft tissue and tumor contrast, allowing us to deliver MRI-based online-replanned radiation treatment (RT) with real-time tumor motion tracking. In this work, we introduce an online adaptive replanning tool to address interfractional changes in tumor and organs at risk (OARs), based on the synthetic CT converted from the pretreatment MRI.
METHODS AND DISCUSSION: The online tool is developed to perform three different plan adaptation strategies—simple, moderate, and complex—based on the degree of the interfractional variations. The simple strategy is designed for situations in which translational shift is the dominant interfractional change, where the adaptive plan is generated by shifting all beam apertures based on the rigid registration of the planning and daily images and by optimizing all beam weights based on the dose map from the original plan. The simple strategy requires no segmentation and can be performed quickly. The moderate strategy will use the previously developed segment aperture morphing (SAM) algorithm to account for the changes, including target deformation, which requires delineation of the target. The complex strategy is a general solution to fully address interfractional variations; this approach requires segmentation of target and OARs. In this strategy, the adaptive plan is generated by the SAM and warm-start optimization (WSO) algorithms.
A software tool is developed to perform a comprehensive pretreatment and posttreatment delivery quality assurance, including segmentation verification, plan check, and delivery validation. The entire procedure takes 4 minutes for the simple strategy, 6 minutes for the moderate strategy, and 10 minutes for the complex strategy. The adaptive plans developed from the three techniques have shown to be comparable with those generated from the full-blown reoptimization plans based on daily images.
Proceedings of the 98th Annual Meeting of the American Radium Society — americanradiumsociety.org