医用直线加速器质控数据分析与管理系统-QAManager的功能设计与临床应用

目的:基于Python Django框架的加速器质控自动化分析网页系统（QAManager）的开发和测试。方法:利用Python Django和MySQL数据库开发系统QAManager，实现加速器质控数据的统计分析和OBI系统影像自动分析功能。使用QAManager对TrueBeam加速器为期半年的质控结果统计分析并进行功能性测试，包括:①]利用QAManager分析各项质控数据分布和变化趋势，对该段时间内TrueBeam的状态进行综合评估；②]选取一年内TrueBeam OBI系统扫描Catphan504所得影像，分别使用人工、QAManager分析，并比较两者结果，分析QAManager图像自动识别算法的准确性和精确性；③]将模体分别向前、后、左、右4个方向移动1 cm，采用QAManager自动检测，分析算法的鲁棒性。结果:QAManager界面简洁紧凑，用户能够使用QAManager自由选择需要展示的质控项目绘制曲线图，清晰显示各项目整体的分布趋势以及出现偏差最大的时间；QAManager能够准确识别到Catphan504 CBCT影像检测中各模块，除层厚（P=0.05）外，人工分析和自动分析结果比较差异没有统计学意义（P>0.05）。结论:QAManager系统准确可靠，算法能够准确地检测出加速器OBI影像系统参数，具有很强的应用性，对科室实现网络化管理具有重要意义。

Development and clinical application of a data analysis and management system (QAManager) for accelerator quality assurance

Abstract: Objective To develop and test Python Django-based automatic analysis web system (QAManager) for accelerator quality assurance (QA). Methods QAManager was developed based on Python Django and MySQL database, with the functions of statistical analysis on accelerator QA data and automatic analysis on images obtained by on-board imaging (OBI) system. Half-a-year QA results of TrueBeam accelerator were statistically analyzed by QAManager. Besides, several functional tests were also conducted. ①] The distribution and variation trend of all QA items were analyzed by QAManager, thereby comprehensively evaluating the working state of TrueBeam accelerator during the period. ②] The results of manual analysis and QAManager analysis on Catphan504 images obtained by TrueBeam OBI system in the past year were analyzed and compared, thereby analyzing the accuracy and precision of QAManager image automatic recognition algorithm. ③ ]Catphan504 phantom which was set up on the treatment couch was deliberately shifted by 1 cm in lateral and longitudinal (no rotational) directions from the reference position, and then QAManager was used for automatic detection, thereby analyzing algorithm robustness. Results The interface of QAManager was user-friendly and compact. According to the custom QA item graphs drawn by QAManager, the rules of QA data distribution and the trends of data varying with time were pointed out and all modules in Catphan504 CBCT image detection were identified precisely by QAManager. No significant difference was found between manual analysis and QAManager analysis on Catphan504 images obtained by TrueBeam OBI system (P>0.05), except for slice thickness (P=0.05). Conclusion QAManager is accuracy and reliable, and the algorithm can be used to accurately detect all OBI parameters. Therefore, QAManager can be widely used in clinic and is vital for network-based department management.