中文摘要：

目的建立利用半导体探测器阵列检测多叶准直器（MLC）叶片定位误差的方法，用于MLC的日常质量保证。方法在MapCheck半导体探测器阵列上，相邻行探测器是错开半个探测器间距分布的。针对这种分布特点，设计了台阶式叶片位置模式，代替胶片测量常用的篱笆式（picketfence）叶片位置模式，作为叶片位置参考模式。在参考模式的基础上，人为引入不同大小的叶片位置误差，得到一组新的叶片位置模式——叶片位置校准模式。依次照射参考模式和校准模式，得到每一个探测器测量剂量与叶片位置误差的关系曲线。该曲线作为测量叶片位置误差的校准曲线。在进行MLC日常质量保证时，只需定期照射叶片位置参考模式，就可得到定量的叶片位置误差。结果新方法在一个配置40对MLC的加速器上进行了测试（ElektaPrecise）。篱笆式叶片位置模式一次检测160个叶片位置，而台阶式叶片位置模式一次检测240个位置，检测效率提高50％；对于同样的1mm叶片误差，前后两种模式的剂量变化分别是17％和25％，因此测量灵敏度提高8％。结论建立了一种新的定量的MLC日常QA方法。通过重新设计叶片位置模式，新方法提高了实施效率和测量灵敏度。

英文摘要：

Objective To design leaf patterns for Muhileaf Collimator（MLC） routine quality assurance （QA） with a 2D diode array. Methods According to the detector distribution characteristic of the 2D diode array and basing on the ＂picket fence＂ pattern, design the＂stepwise＂pattern. For each diode involving MLC QA, a calibration curve of relative output versus leaf positioning error was measured through delivering a set of patterns with different intentionally introduced positioning errors. When this proposed technique was delivered, the referenced patterns were exposed, and the calibration curves were used as a mean to quantitative determination of the leaf possible positioning errors through the detector readings. Results Compared with the ＂picket fence＂ pattern, the ＂stepwise＂ pattern not only had a high detecting efficiency, but also increased the dosimetric sensitivity to leaf positioning error. A 1 mm leaf positioning error corresponds to a dose variation of 25 % for the ＂stcpwise＂ pattern, while for the ＂picket fence＂ pattern the same positioning error just causes a 17% dose variation. Conclusions The new ＂stepwise＂ pattern is more efficient to be carried out, and more sensitive to sub-millimeter changes of leaf positioning.