Difference in gross target volume derived from 18F-FDG PET/CT and from CT images for lung cancer: a systematic analysis

Objective: The aim of this study was to analyze the diversity of gross target volume (GTV)PET/CT and GTVCT in different patients, observers and studies, and to find a statistical method to identify the discrepancy among subsets of GTV. Methods: The articles in PUBMED and EMBASE that defined target volume based on PET/CT images among lung cancer patients were reviewed and analyzed statistically. Results: Four relative papers were found in the searches (till March 2006). The statistical analysis of the 4 studies had shown that significant difference of GTVCT existed among studies (F=9.43, P<0.01), but GTVPET/CT remitted (F=1.83, P=0.149). In the data of Steenbakkers et al, significant differences existed among observers (F=3.93, P<0.01) in both GTVPET/CT and GTVCT (F=29.31, P<0.01). Absolute diversity and concordance were better with 'relative golden standard' ( partial correlations were 0.384 and 0.506, respectively, both P<0.01). Conclusions: There was significant difference of GTVPET/CT, and GTVCT among patients, observer's or studies. PET/CT could significantly change the radiotherapy planning, reduce the variance among observers. In addition, the GTV on hardware fusion images, compared with software fusion, had a narrow range with a significant statistical difference. (authors)