Background: The mid-term performance of clinical linear accelerator (LINAC) during volumetric modulated arc therapy (VMAT) treatment period is not performed in clinical practice and usually replaced with one-time plan quality assurance (QA). In this research we aim to monitor daily reproducibility of VMAT delivery from tracking individual leaf movement error and dosimetric error to evaluate the mid-term quality of the machine used. Materials and Methods: First, multileaf collimator (MLC) information was imported into MATLAB program to determine which of the MLC leaves in the leaf bank had the maximum RMS position error (maxRMS). We estimated where the maximum positional errors (maxPE) of the chosen leaf occur along its path length and tracked its daily variations over the entire treatment period. Secondly, picture information of dosimetric error from portal dosimetry was imported into MATLAB where representative high gamma index region (HGR) was determined as HGR with length of > 1 cm and their centers were daily tracked. Results and Discussion: The maxPEs in the brain and tongue cases were distributed broader than in other cases, but all data were found located within ± 0.5 mm. From first day to last day all of five cases show the similar visual pattern of HGRs and Centers of the longest HGRs remained within ± 1 mm of that in first day. These findings prove excellent mid-term performance of the LINAC used in VMAT treatments over a full course of treatment. Conclusion: Tracking the daily location changes of leaf movement and dosimetric error can be a good indicator of predicting the daily quality like stability and reproducibility of beam delivering in VMAT treatment.
Domain generalization approaches aim to learn a domain invariant prediction model for unknown target domains from multiple training source domains with different distributions. Significant efforts have recently been committed to broad domain generalization, which is a challenging and topical problem in machine learning and computer vision communities. Most previous domain generalization approaches assume that the conditional distribution across the domains remain the same across the source domains and learn a domain invariant model by minimizing the marginal distributions. However, the assumption of a stable conditional distribution of the training source domains does not really hold in practice. The hyperplane learned from the source domains will easily misclassify samples scattered at the boundary of clusters or far from their corresponding class centres. To address the above two drawbacks, we propose a discriminative domain-invariant adversarial network (DDIAN) for domain generalization. The discriminativeness of the features are guaranteed through a discriminative feature module and domain-invariant features are guaranteed through the global domain and local sub-domain alignment modules. Extensive experiments on several benchmarks show that DDIAN achieves better prediction on unseen target data during training compared to state-of-the-art domain generalization approaches.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.