论文部分内容阅读
高分辨率是现代正电子发射型断层显像仪(Positron Emission Tomography,PET)设备最重要的技术指标之一,高分辨率PET探测器通常由海量闪烁晶体组成,这使得探测器校准时,晶体查找表建立的工作量大幅增加,从而对相关自动化算法的鲁棒性提出了更高的要求。目前,晶体阵列的矫正主要依赖固定放射源对探测器晶体阵列成像得到的光子二维位置直方图。高分辨率PET探测器的晶体阵列中晶体个数增多,尺寸变小,导致二维位置直方图信噪比下降,且非线性形变更加复杂,使已有的晶体查找表建立算法都无法得到理想的自动化效率。在测试了多种前沿的晶体查找表建立算法后发现,某些算法的结果之间有很强的互补性,可以通过将多种算法结果相融合的方法得到优于单一探测算法的结果。因此提出了一种基于多种探测算法结果分级融合的晶体查找表建立算法,在实现过程中选取了三种互补性较强的晶体探测方法,分级融合其结果,大幅度降低了二维位置直方图上晶体分割的出错概率,获得了鲁棒的结果。
High resolution is one of the most important specifications for modern Positron Emission Tomography (PET) devices. High resolution PET detectors usually consist of a large number of scintillation crystals that allow the detector to be calibrated such that the crystal The workloads created by look-up tables have increased dramatically, placing higher demands on the robustness of the associated automation algorithms. At present, the correction of the crystal array mainly depends on the photon two-dimensional position histogram obtained by imaging the detector crystal array by the fixed radiation source. High-resolution PET detector crystal array increases the number of smaller, smaller size, resulting in two-dimensional position histogram SNR decreased, and the nonlinear deformation is more complex, the existing crystal lookup table algorithm can not be established Automation efficiency. After testing a variety of cutting-edge crystal look-up tables, we found that the results of some algorithms are highly complementary and can be obtained by combining multiple algorithms results better than the single detection algorithm results. Therefore, a crystal lookup table algorithm based on hierarchical fusion of multiple detection algorithm results is proposed. Three kinds of crystal detection methods with high complementarity are selected in the process of realization, and the results are graded and merged to reduce the two-dimensional position histogram The error probability of the crystal segmentation on the graph achieves a robust result.