Kinetic parametric estimation in animal PET molecular imaging based on artificial immune network / 中华核医学杂志

ABSTRACT

Objective To develop an accurate,reliable method without the need of initialization in animal PET modeling for estimation of the tracer kinetic parameters based on the artificial immune network.Methods The hepatic and left ventricular time activity curves (TACs) were obtained by drawing ROIs of liver tissue and left ventricle on dynamic 18F-FDG PET imaging of small mice.Meanwhile,the blood TAC was analyzed by sampling the tail vein blood at different time points after injection.The artificial immune network for parametric optimization of pharmacokinetics (PKAIN) was adapted to estimate the model parameters and the metabolic rate of glucose (Ki) was calculated.Results TACs of liver,left ventricle and tail vein blood were obtained.Based on the artificial immune network,Ki in 3 mice was estimated as 0.0024,0.0417 and 0.0047,respectively.The average weighted residual sum of squares of the output model generated by PKAIN was less than 0.0745 with a maximum standard deviation of 0.0084,which indicated that the proposed PKAIN method can provide accurate and reliable parametric estimation.Conclusion The PKAIN method could provide accurate and reliable tracer kinetic modeling in animal PET imaging without the need of initialization of model parameters.