ABSTRACT
AIMS: High local control rates are achieved in stage I lung cancer using stereotactic ablative radiotherapy. Target delineation is commonly based on four-dimensional computed tomography (CT) scans. Target volumes defined by positron emission tomography/computed tomography (PET/CT) are compared with those defined by four-dimensional CT and conventional ('three-dimensional') (18)F-fluorodeoxyglucose ((18)F-FDG) PET/CT. MATERIALS AND METHODS: For 16 stage I non-small cell lung cancer tumours, six approaches for deriving PET target volumes were evaluated: manual contouring, standardised uptake value (SUV) absolute threshold of 2.5, 35% of maximum SUV (35%SUV(MAX)), 41% of SUV(MAX) (41%SUV(MAX)) and two different source to background ratio techniques (SBR-1 and SBR-2). PET-derived target volumes were compared with the internal target volume (ITV) from the modified maximum intensity projection (MIP(MOD) ITV). Volumetric and positional correlation was assessed using the Dice similarity coefficient (DSC). RESULTS: PET-based target volumes did not correspond to four-dimensional CT-based target volumes. The mean DSC relative to MIP(MOD) ITV were: PET manual = 0.64, SUV2.5 = 0.64, 35%SUV(MAX) = 0.63, 41%SUV(MAX) = 0.57. SBR-1 = 0.52, SBR-2 = 0.49. PET-based target volumes were smaller than corresponding MIP ITVs. CONCLUSIONS: Conventional three-dimensional (18)F-FDG PET-derived target volumes for lung stereotactic ablative radiotherapy did not correspond well with those derived from four-dimensional CT, including those in routine clinical use (MIP(MOD) ITV). Caution is required in using three-dimensional PET for motion encompassing target volume delineation.
