Implementation of L4 Automation for Patient-Specific Quality Assurance Using the AUTOFRAME Platform and a Robotic Mechanism

Abstract

The AUTOFRAME platform has been previously developed for automating clinical workflows and has achieved L3 automation for patient-specific quality assurance (PSQA). The aim of this study is to further develop the platform to achieve L4 automation, which extends AUTOFRAME to operate Varian's control console (VCC) for the linear accelerator. The goal is to automate all PSQA operations outside of the treatment room. The automated stages, listed in Table 1, are ranging from L0 to L5. The current study focuses on L4 automation of PSQA, which uses a 6-axis robot arm to control buttons on the VCC. The robotic arm is mounted on a custom frame that is attached onto the VCC to maintain its relative position. The arm is controlled through python scripts, which is run on a raspberry pi and communicates with the AUTOFRAME platform via the PyFlow and AutoFlow subsystems. This integration allows the robotic arm to receive commands from AutoFlow and execute them through PyFlow, activating real-time button-pressing actions. The integrated L4 system was tested using real PSQA verification plans, which involved a minimum of 4 actions and a maximum of 9 actions on the VCC, including a couch kick that required overriding the couch position. The results showed that the 6-axis arm was able to correctly execute the action groups and guide the robotic arm to consistently press and hold each button with sufficient force and accuracy. All buttons on the VCC, including the Motion-enable, Preparing, Readying, and Delivering buttons, were correctly pushed and held according to the PSQA steps. The system successfully delivered all beams without interruption, and all beam delivery data was collected by the detector for PSQA analysis. The study has further developed the AUTOFRAME platform for L4 automation of PSQA procedures can be achieved. The robotic arm is capable of pushing all required buttons, eliminating all human interactions on the VCC. When combined with previous L0-L3 automations, this will move towards full automation of PSQA procedures. Future work will focus on improving the system's flexibility, stability, and extending its operations to other QA tasks.