Introduction: The aim of this study is to determine the interobserver variability (IV) between radiation oncologists (RO) in target volume delineation for postoperative gastric cancer (GC) radiotherapy planning.
Materials and methods: Four physicians were asked to delimitate clinical target volume (CTV) on the same 3D CT images in 9 postoperative radiochemotherapy GC patients. Instructions were given to include tumour bed, remaining stomach, anastomosis, duodenal loop and local lymph nodes. The principal variable was spatial volume discrepancy between the main observer (called "A") and other observers (all called "B"), which were compared using the mathematical formula A⌣B/A⌢B, applied to the 3D CT images using Boolean operators. Analysis of variance with two random effects (observers and patients) was performed.
Results: Mean volumes were 1410 cm(3) for OBA, 1231 cm(3) for OB2, 734.6 cm(3) for OB3 and 1350 cm(3) for OB4. Discrepancies were 519.9±431.6 cm(3) for OB2, 652.1±294.36 cm(3) for OB3 and 225.90±237.07 cm(3) for OB4. Standard deviation ascribed to patients as random effect was 898.6 cm(3) and that ascribed to observers was 198.10 cm(3), considered as a statistically significant difference.
Conclusions: A significant IV in target delineation that can be attributed to many factors depends more on patients' characteristics than RO delineating decisions.