A COMPARISON OF SOFT TISSUE AND BONE TO VERIFY TREATMENT POSITION FOR LUNG CANCER PATIENTS RECEIVING RADICAL RADIOTHERAPY.
Introduction Two-dimensional (2D) kilovoltage (kV) imaging is currently used in the department for positioning lung cancer patients prior to non-stereotactic ablative radiotherapy (SABR). With enhanced visualisation of soft tissue using three-dimensional cone-beam computed tomography (CBCT), its use in non-SABR lung radiotherapy is highly recommended. Prior to its introduction in the department, matching to bone, carina and tumour was assessed retrospectively in CBCTs collected as part of a service evaluation and compared to the current 2D kV protocol of matching to bones. This determined which anatomical landmark provided the optimum treatment accuracy. The feasibility of introducing CBCT imaging as routine protocol was also assessed. Method In this quantitative study, five radiographers conducted automatic and manual matches to bone, carina and tumour in 88 CBCTs of 20 patients. For each of the 2600 match positions, couch shifts were recorded in the anterior/posterior, left/right and superior/inferior directions. Tumour coverage was graded at each match position according to target volume margins. The level of agreement between automatic and manual matches and the percentage of couch errors out with clinical tolerance (5mm) were calculated for each matching method. The feasibility of CBCT imaging was assessed by examining inter-observer reliability, reporting difficulties in the matching process and comparing the time taken to match to CBCTs compared with 2D kV imaging. Results Paired t-tests demonstrated a highly significant difference in the couch position when tumour was matched to instead of bone (P<0.001). Tumour matching detected the highest proportion of set-up errors out width tolerance (26.1%), followed by carina (19.05%) and bone (18.41%). Bland-Altman analysis demonstrated tumour matching had the lowest levels of automatic and manual matching agreement. Using the intraclass correlation coefficient (ICC), all matching methods demonstrated either good or excellent inter-observer reliability (ICC 0.871-0.957). Problematic matches were reported in 20% of matches and timings were comparable (four minutes) with 2D kV imaging. Conclusion This study supports the introduction of CBCT imaging and matching directly to tumour as routine protocol for all non-SABR lung cancer patients. Visual inspection of the tumour is highly recommended after automatic matching to ensure target coverage, with the use of the carina as a guide when problematic matching occurs.