Comparison Of The Gross Target Volume Based On Diagnostic PETCT For Primary Esophageal Cancer

BackgroundClinically,many esophageal cancer patients who were prepared for radiation therapy already had diagnostic PET-CT image.Because of the lack of validation,it is generally considered that the volume contours of the primary tumor for thoracic esophageal cancer(EC)only referencing PET-CT,which is widely used in the clinic now,may be uncertain to radiation oncologists.However,it is unlikely that these patients would undergo a second PET-CT scan owing to the significant cost and logistical problems involved,as well as the increased radiation exposure of the patient and staff during the scan.As a result,it is relatively difficult to popularize dedicated treatment planning PET-CT as routine management in clinical practice,and diagnostic PET-CT may be the only PET data provided for radiation treatment planning(RTP).ObjectiveTo compare positional and volumetric differences between the gross 18F-FDG PET-CT and the GTV on the deformed image derived from 3D-CT and 18F-FDG PET-CT for primary thoracic esophageal cancer(EC).MethodsSeventy-two patients with EC sequentially underwent 18F-FDG PET/CT scans used for diagnosis and 3D-CT scans for simulation.Then all images were transferred to MIM software.The 3D-CT images were firstly registrated and fused with the CT images of PET-CT sets using deformable registration algorithm.Initially,a rigid registration focusing on the dorsal spine was undertaken automatically using MIM software.Following rigid registration,deformable registration was then performed.Finally,we placed a bounding box over the region of interest(ROI)to remove the influence of other parts of the body because of the automatic image registration algorithm considers the entire image dataset.At the same time,PET voxels were mapped to the planning CT scan during the transformation used in the CT-CT registration process,resulting in a new PET-CT dataset that was deformably registered to the planning CT scan.GTVs were delineated for different phases by the same one experimented radiotherapist.Under the mediastinal window settings(window width=400 HU,window level=40 HU),the GTV3D was delineated on 3D-CT without referencing 18F-FDG PET-CT,the GTVPET-ref was delineated on 3D-CT referencing 18F-FDG PET-CT,and the GTVPET-reg was delineated on the deformed image derived from 3D-CT and 18F-FDG PET-CT by MIM deformable registration software.After the delineation of every target volume,another experimented radiotherapist and an experimented nuclear medicine doctor reviewed the contures.The differences in position,volume,length,conformity index(CI),and degree of inclusion(DI)of target volumes were compered,respectively.ResultsThe median volume of GTV3D,GTVPET-ref,GTVPET-reg were 44.90,40.36 and 41.15 cm3,respectively.There was no statistical difference between the volumes of any two targets.The mean lengths of GTV3D,GTVPET-ref,GTVPET-reg were 8.54,9.29 and 8.38 cm,respectively.The length of GTVPET-ref was longer than that of GTV3D(t=2.134,P<0.05).However,there was no significant difference in length between GTVPET-reg and GTVPET-ref(t=1.357,P>0.05)or between GTVPET-reg and GTV3D(t=-0.087,P>0.05).The paired sample t test results indicated that the maximum transverse diameters of GTVPET-reg and GTVPETref were less than that of GTV3D for all patients,(t=-1.891?2.716,P<0.05)and that there was no significant difference between GTVPET-reg and GTVPET-ref(P>0.05).The median DIs of GTVPET-ref,GTVPET-reg in GTV3D were 0.86,0.82(Z=-2.741,P<0.05),and that of GTV3D in GTVPET-ref,GTVPET-reg were 0.87,0.84(Z=-1.429,P<0.05).The median CIs of GTV3D in GTVPET-ref,GTVPET-reg were 0.72,0.68(Z=2.756,P<0.05).The Cl between GTVPET-reg and GTVPET-ref,GTVPET-reg and GTVPET-ref,and GTVPET-reg and GTV3D showed a significant negative correlation with the 3D vector(R=-0.344,-0.517,-0.527,P<0.05).There was no difference in the centroid distance between GTVPET-reg and GTVPET-ref,GTVPET-reg and GTV3D,and GTVPET-ref and GTV3D(P>0.05)ConclusionsPET-CT could optimize definition of target volumes of esophageal cancer However,there was no significant difference between GTV contoured on three-dimensional CT(3D-CT)referencing 18F-FDG PET-CT and the GTV on the deformed image derived from 3D-CT and 18F-FDG PET-CT either in volume size or in spatial position.Therefore,it is recommended that radiation oncologists can refer to the recent diagnostic PET-CT when delineating the gross target volume for primary thoracic esophageal cancer.