Improving Online Adaptive Radiotherapy Quality Assurance with Streamlined Clinical Workflow through In-House Development

Abstract

<h3>Purpose/Objective(s)</h3> Online adaptive radiotherapy has been clinically feasible through decades of research and development and is now available in the clinical routine with dedicated platforms using CBCT-LINAC or MR-LINAC. The quality assurance (QA) of online plans has become a bottleneck in the clinical workflow as it often requires manual data export/import through third party software, lengthy wait for the secondary dose calculation, and tedious work on report generation and checking, etc. The QA bottleneck not only impedes clinical workflow but also risks plan and treatment quality due to the time pressure with the patient waiting on the treatment table. We have developed and implemented in-house solutions to address such a bottleneck. <h3>Materials/Methods</h3> We developed dedicated in-house infrastructure to facilitate online adaptive plan QA and clinical workflow: 1) RT-PACK with SQLite database, which receives and organizes DICOM-RT data from any DICOM sender (e.g., clinical TPS); 2) a general purpose dose calculation server powered by the in-house GPU-based Monte Carlo particle transport engine with electromagnetic field modelling; 3) a report generator that automatically generates reports for secondary dose and plan quality check; 4) a web-portal that allows clinicians to review and sign the report anywhere with a network and VPN connection, even from a mobile phone. The reports and related documents will be automatically sent to the Record & Verify system after obtaining both physician and physicist signatures. An auto-commissioner was also developed to facilitate beam modeling based on water-tank measurements, which serves the purpose of fully independent verification. The dose calculation was performed on the whole patient volume with grid resolution of 2 mm x 2 mm x 2 mm. The 3D dose volume, treatment fields, point dose, Gamma Index with global 3%, 2mm criteria, three orthogonal dose profiles, dose volume histograms, and dosimetric constraints are metrics to compare the plan retrieved from TPS to the secondary dose verification and reported for clinicians' review. The whole in-house infrastructure went through extensive unit tests, dosimetric accuracy verifications, and alpha and beta tests before adopted for clinical use. <h3>Results</h3> The secondary dose and plan check infrastructure went live for CBCT-LINAC and MR-LINAC in our clinic and has since performed secondary check for every plan generated (846 plans and 622 plans, respectively, in less than 6 months.) The average processing time, from DICOM receiving to report generation, was 62 seconds for CBCT-LINAC and 130 seconds for MR-LINAC, which was 3-5 minutes shorter compared to using commercial software. More than 95% of CBCT-LINAC (MR-LINAC) plans had Gamma passing rates >99% (>95%). <h3>Conclusion</h3> The in-house developed software, through strict verification and validation, has been integrated into the clinical system. It has greatly improved online adaptive quality assurance with efficient and streamlined clinical workflow.