PET探测器数据获取与晶体识别-《中国医学物理学杂志》

文章信息/Info

作者:: 岳松1; 佟林格2; 尹永智2; 裴曦1; 1.中国科学技术大学物理学院，安徽合肥 230027； 2.兰州大学核科学与技术学院，甘肃兰州 730000

Author(s):: YUE Song1; TONG Lin’ge2; YIN Yongzhi2; PEI Xi1; 1. School of Physics, University of Science and Technology of China, Hefei 230027, China; 2. School of Nuclear Science and Technology, Lanzhou University, Lanzhou 730000, China

Keywords:: Keywords: positron emission tomography; crystal identification; watershed algorithm; image segmentation

摘要:: 目的：旨在设计一个小像素PET成像系统，并获得用于断层扫描的晶体查找表。方法：对PCI-ADC系统获取的LYSO闪烁体（pitch_2.0_mm_10×10、pitch_1.6 mm_10×1）探头晶体位置分辨图像，采用分水岭算法完成图像分割任务，实现晶体识别。结果：实验中，pitch_2.0_mm_10×10、pitch_1.6 mm_10×10两种像素探头晶体能够全部识别；同时对华盛顿大学pitch_1.6 mm_20×20探头数据进行处理，没有观察到晶体粘附现象。结论：PET数据采集系统性能良好，分水岭算法能够识别不同的像素晶体，为后期断层扫描和三维重建提供了基础。

Abstract:: Abstract: Objective To design a small-pixel PET imaging system and obtain the crystal lookup table used for computed tomography. Methods After the resolution images of the probe positions of LYSO scintillators (pitch_2.0_mm_10×10 and pitch_1.6 mm_10×1) were obtained by PCI-ADC system, watershed algorithm was employed to complete image segmentation and realize crystal identification. Results For the pixel probes of pitch_2.0_mm_10×10 and pitch_1.6 mm_10×10, all the crystals were identified. Meanwhile, the data of pitch_1.6 mm_20×20 from University of Washington were processed, and no crystal adhesion was observed. Conclusion The PET data acquisition system performs satisfactorily, and watershed algorithm is capable of identifying different pixel crystals, which provide a basis for the further computed tomography and three-dimensional construction.