Comparison of Target Volumes Delineated on 3DCT Referencing 18f-FDG PET/CT and the Deformed Image Derived from 3DCT with 18f-FDG PET/CT of Primary Thoracic Esophageal Cancer

Abstract

To compare positional and volumetric differences between the gross target volumes (GTV) contoured on three-dimensional CT (3DCT) referencing 18F-FDG PET/CT and the GTV on the deformed image derived from 3DCT and 18F-FDG PET/CT for primary thoracic esophageal cancer (EC). Eighty-one patients with EC sequentially underwent 18F-FDG PET/CT scans used for diagnosis and 3DCT scans for simulation. Firstly, the GTV3D was contoured based on 3DCT without referencing 18F-FDG PET/CT. Secondly, the GTVPET-ref was contured on 3DCT referencing 18F-FDG PET/CT. Thirdly, the GTVPET-reg wes contured on the deformed image derived from 3DCT and 18F-FDG PET/CT by deformable registration software. The standardised uptake value (SUV) 2.0 and 20% of maximal SUV (SUVmax) were selected to delineate the GTV on PET-CT. Lastly, the GTVPET-reg-auto was obtained by the deformed image using the auto-contouring function of the software program. All the manual contours were completed by a experienced radiotherapist. The differences in position, volume, length, conformity index (CI), and degree of inclusion (DI) of target volumes were compered, respectively. The mean volume ratio of GTVPET-ref to GTV3D, GTVPET-reg, GTVPET-reg-auto were 1.073, 1.184, 3.974, and that of GTV3D to GTVPET-reg, GTVPET-reg-auto were 1.123, 3.431. The minimum target volume was GTVPET-reg, the middle was GTV3D, and the maximum was GTVPET-reg-auto. The mean length of GTV3D, GTVPET-ref, GTVPET-reg, GTVPET-reg-auto were 85.241, 94.031, 86.6941,86.6941, respectively. The length of GTVPET-ref was longer than GTV3D(p=0.008). The maximum diameter of GTV3D, GTVPET-ref , GTVPET-reg, GTVPET-reg-auto were 3.9447, 3.7399, 3.7960, 3.7960. The width of GTV3D was wider than GTVPET-ref(p=0.000). The mean DIs of GTVPET-ref in GTV3D, GTVPET-reg, GTVPET-reg-auto were 0.815, 0.813, 0.565, and that of GTV3D in GTVPET-ref, GTVPET-reg and GTVPET-reg-auto were 0.827, 0.774, 0.585. The best DI was GTV3D in GTVPET-re. The mean CIs of GTVPET-ref in GTV3D, GTVPET-reg, GTVPET-reg-auto were 0.696, 0.679, 0.450,and that of GTV3D in GTVPET-reg and GTVPET-reg-auto were 0.641,0.450. The best CI was GTVPET-ref in GTV3D. The minimum target volume was GTVPET-reg, the middle was GTV3D, and the maximum was GTVPET-reg-auto. Therefore, The deformed image derived from 3DCT and 18F-FDG PET/CT has an potential value to delineate the volume and length of the gross target volume. At the same time, GTVPET-reg-auto should be chosen carefully due to the possible error of it.