摘要： An in-house developed EPID based QA strategy was deployed clinically to test daily linac performance before treatment. Beam dosimetric parameters, including output, flatness, symmetry, uniformity, TPR20/10, and MLC repeatability, as well as image quality metrics, such as high and low contrast resolutions, were recorded based on MV and kV EPID imaging analysis. Over a period of three years, data was collected and analyzed. Trend analysis with data visualization, comparisons to ionization chamber (IC) results and QA failure studies were performed to evaluate the reliability and robustness of the EPID based daily QA system. The EPID output tests over the three-year period cross-correlated with the monthly QA IC system within 1.3% and TG51 results within 1%. A consistent positive output drift was detected for all energies during that time. A faster drifting speed was observed after the installation of a new beam monitoring chamber on the machine but slowed down over time. Symmetry, flatness, and uniformity were stable and the changes were directly related to beam steering or QA baseline resetting. TPR20/10 was not sensitive to dosimetry fluctuations and a good delivery repeatability of MLC fields was observed. The image quality metrics were stable and a passing rate greater than 96% was observed. For the most recent eight month of data, a one-time QA passing rate was calculated to be 87.6%. Beam interruptions, human errors and data transfer issues were the major reasons for QA failures. The data gathered over 3 years indicates that the EPID-based process is an efficient and reliable QA solution for daily linac performance checks. The long-term trend analysis and data visualization can help physicists understand the behavior of a linac, optimize the daily QA procedure, and has the potential to predict machine failures prior to incidents.