J Encaoua1, R Abgral2, C Leleu3, O El Kabbaj1, P Caradec3, D Bourhis2, O Pradier1, U Schick4. 1. Service de radiothérapie, institut de cancérologie et d'hématologie, CHRU Morvan, 2, avenue Foch, 29200 Brest, France. 2. Service de médecine nucléaire, CHRU Morvan, 2, avenue Foch, 29200 Brest, France. 3. Service de radiothérapie, centre hospitalier de Cornouaille, 14bis, avenue Yves-Thépot, 29107 Quimper cedex, France. 4. Service de radiothérapie, institut de cancérologie et d'hématologie, CHRU Morvan, 2, avenue Foch, 29200 Brest, France. Electronic address: ulrike.schick@chu-brest.fr.
Abstract
PURPOSE: To study the impact on radiotherapy planning of an automatically segmented target volume delineation based on (18F)-fluorodeoxy-D-glucose (FDG)-hybrid positron emission tomography-computed tomography (PET-CT) compared to a manually delineation based on computed tomography (CT) in oesophageal carcinoma patients. METHODS AND MATERIALS: Fifty-eight patients diagnosed with oesophageal cancer between September 2009 and November 2014 were included. The majority had squamous cell carcinoma (84.5 %), and advanced stage (37.9 % were stade IIIA) and 44.8 % had middle oesophageal lesion. Gross tumour volumes were retrospectively defined based either manually on CT or automatically on coregistered PET/CT images using three different threshold methods: standard-uptake value (SUV) of 2.5, 40 % of maximum intensity and signal-to-background ratio. Target volumes were compared in length, volume and using the index of conformality. Radiotherapy plans to the dose of 50Gy and 66Gy using intensity-modulated radiotherapy were generated and compared for both data sets. Planification target volume coverage and doses delivered to organs at risk (heart, lung and spinal cord) were compared. RESULTS: The gross tumour volume based manually on CT was significantly longer than that automatically based on signal-to-background ratio (6.4cm versus 5.3cm; P<0.008). Doses to the lungs (V20, Dmean), heart (V40), and spinal cord (Dmax) were significantly lower on plans using the PTVSBR. The PTVSBR coverage was statistically better than the PTVCT coverage on both plans. (50Gy: P<0.0004 and 66Gy: P<0.0006). CONCLUSION: The automatic PET segmentation algorithm based on the signal-to-background ratio method for the delineation of oesophageal tumours is interesting, and results in better target volume coverage and decreased dose to organs at risk. This may allow dose escalation up to 66Gy to the gross tumour volume.
PURPOSE: To study the impact on radiotherapy planning of an automatically segmented target volume delineation based on (18F)-fluorodeoxy-D-glucose (FDG)-hybrid positron emission tomography-computed tomography (PET-CT) compared to a manually delineation based on computed tomography (CT) in oesophageal carcinomapatients. METHODS AND MATERIALS: Fifty-eight patients diagnosed with oesophageal cancer between September 2009 and November 2014 were included. The majority had squamous cell carcinoma (84.5 %), and advanced stage (37.9 % were stade IIIA) and 44.8 % had middle oesophageal lesion. Gross tumour volumes were retrospectively defined based either manually on CT or automatically on coregistered PET/CT images using three different threshold methods: standard-uptake value (SUV) of 2.5, 40 % of maximum intensity and signal-to-background ratio. Target volumes were compared in length, volume and using the index of conformality. Radiotherapy plans to the dose of 50Gy and 66Gy using intensity-modulated radiotherapy were generated and compared for both data sets. Planification target volume coverage and doses delivered to organs at risk (heart, lung and spinal cord) were compared. RESULTS: The gross tumour volume based manually on CT was significantly longer than that automatically based on signal-to-background ratio (6.4cm versus 5.3cm; P<0.008). Doses to the lungs (V20, Dmean), heart (V40), and spinal cord (Dmax) were significantly lower on plans using the PTVSBR. The PTVSBR coverage was statistically better than the PTVCT coverage on both plans. (50Gy: P<0.0004 and 66Gy: P<0.0006). CONCLUSION: The automatic PET segmentation algorithm based on the signal-to-background ratio method for the delineation of oesophageal tumours is interesting, and results in better target volume coverage and decreased dose to organs at risk. This may allow dose escalation up to 66Gy to the gross tumour volume.