新型PET/CT分子影像动力学建模技术在肺癌诊断中的应用

摘要

目的：采用基于人工免疫网络智能方法的新型动力学建模技术，获得FDG PET/CT成像的病变肺组织动力学参数，通过定量分析来判别肿瘤的良恶性。方法：5例疑似肺癌患者在无锡市第四人民医院核医学诊疗中心行18F-FDG PET/CT检查。使用德国Siemens公司Biograph True Point 64 PET/CT动态成像，动态图像采样协议为8帧×15秒、6帧×30秒、5帧×300秒，总共持续30分钟。通过感兴趣区域勾画技术得到正常肺组织、疑似肺肿瘤组织和腹主动脉血池区域的时间-放射性活度曲线。采用基于人工免疫网络（PKAIN）的新型PET分子影像动力学参数估计方法（PKAIN）分别估计出正常肺组织和疑似肺肿瘤组织的18F-FDG示踪剂动力学模型参数。结果：5例患者术后均经病理证实为肺部恶性肿瘤，肺部肿瘤组织ROI的FDG动力学参数估计值（均数±标准差）分别为k1=0.1746±0.0531、k2=0.4030±0.3324、k3=0.5208±0.2274、k4=0.1046±0.0543、f=0.1468±0.1305、Ki=0.1003±0.0326。结论：实验结果表明，PKAIN的人工免疫网络智能算法更适合于肿瘤组织的PET/CT分子影像动力学建模。与常规的60分钟18F-FDG采样相比，30分钟短采样时间，提高了FDG PET动力学建模方法的临床应用价值，证实了新型PET/CT分子影像动力学建模方法在肺癌诊断中的价值。%Objective: To distinguish the benign and malignant of lung tumors using a new method based on the artificial immune network algorithm for PET/CT molecular imaging kinetic modeling. Methods: All studies were performed at the Nuclear Medicine Central of the Wuxi Fourth People's Hospital, Wuxi, China. There are five lung cancer patients under PET/CT scan. The PET/CT scans were performed with Biograph True Point 64 PET/CT. Images was acquired for 30 min with people under injection of FDG drug about 370 to 555 MBq. The scanning schedule was: 8 15-s scans, 6 30-s scans, 5 5-min scans. Small size of ROIs was manually drawn over the PET images to obtain the time-activity curves for the Normal lung tissue, suspected lung tumor tissue and abdominal aortic blood pool area. Kinetic parameters of the both the normal lung tissue and the suspected lung tumor tissue were estimated by using the new method which was based on the artificial immune network algorithm (PKAIN). Results: The statistics of parameters of the FDG PET kinetic models of five patients with suspected lung tumor tissue were k1=0.1190±0.0023, k2=0.0397± 0.0132, k3=0.7316±0.3421, k4=0.4334±0.3595, f=0.0857±0.0032, Ki=0.1117±0.00299. Conclusion: The PKAIN outperformed the KIS software when fitting the observed TACs especially for the lung cancer regions. Compared with the 60min sample schedule, the 30min sample schedule is more applicable and to be proven well in lung cancer diagnosis.