Phantom study of PET/CT guided delineation of radiation therapy volume

Abstract

Objective To propose a model-based method for calculating the threshold in GTV determination by 18 F-FDG PET in a phantom study.Methods A phantom was constructed of a 9 L cylindrical tank.Glass spheres with volumes ranging from 0.5 to 16 ml (0.5,1,2,4,8 and 16 ml) were suspended within the tank.The six spheres were filled with an identical concentration of FDG ( 203.5 MBq/L) and suspended within 3 different background baths of FDG (6.179,16.021,0 MBq/L) solutions,creating 3 target-to-background ratios of 32.96∶ 1,12.69∶1 and target to zero background.A linear regressive function was constructed which represented the relationship between the threshold and the average activity concentration of the target.A 40％ of maximum intensity threshold and the linear regressive function method were applied to define the spheres filled with 18F-FDG.The volume differences between the two methods and the true volumes of the spheres were compared with t-test.Results The linear regressive function model was derived as:threshold =(mean target concentration + 2.6227)/1.9752.The results indicated that a smaller deviation occurred when the function was utilized to estimate the volumes of the phantoms as compared to the 40％ of maximum intensity threshold method,but there were no significant differences between them ( t =0.306,P ＞ 0.05 ).The effect of the linear regressive function on volume was such that when the phantom sphere volumes were ≥ 1 ml,the average deviation between the defined volumes and the true volumes of phantoms was 1.01％ ; but when the phantom sphere volume was 0.5 ml,the average deviation was 9.53％.When the 40％ of maximum intensity threshold method was applied to define the phantom spheres of volume≥2 ml,the average deviation between the defined volumes and the true volumes of phantoms was -4.62％ ; but,the average deviation of that was 19.9％ when the volumes of spheres were 0.5 and 1 ml.When the linear regressive function was applied to phantom spheres with volume ≥ 1 ml,the effect of targetto-background ratios on volume estimation was such that the average deviation between the defined volumes and the true volumes of phantoms were 2.66％,1.11％, - 0.74％ under 3 target-to-background ratios,which showed no statistically significant differences among them (t values were 1.373,2.798,1.328,all P ＞0.05 ).When the 40％ of maximum intensity threshold method was applied to define the phantom spheres of volumes ≥2 ml,the average deviation between the defined volumes and the true volumes of phantoms decreased in parallel to the decreasing target-to-background ratios.Conclusions The threshold constructed by the linear regression function was strongly dependent on the mean target concentration,and the function can be adequately applied to estimate phantoms with volume≥ 1 ml.Target-to-background ratios had no obvious effect on the linear regressive function method when the volumes of the spheres were above 1 ml. Key words: Neoplasms;  Radiotherapy;  Tomography, emission-computed ;  Tomography, X-ray computed ;  Target delineation;  Models, theoretical