Quantitative and qualitative assessment of thoracic CBCT image quality for multiple imaging systems

Purpose or Objective: A Varian TrueBeam with OBI was commissioned in 2014. During early clinical use concerns were raised regarding thoracic CBCT image quality in comparison with that observed in Elekta XVI images. Streaking artefacts caused by respiratory motion were the primary reason for the perceived poor quality. This study compared the image quality of the TrueBeam OBI with the other CBCT systems at the centre, a Varian Trilogy OBI and Elekta XVI, using quantitative and qualitative methods. Material and Methods: A static Catphan phantom (The Phantom Laboratory) was used to assess image quality quantitatively, and to create a HU calibration curve for the XVI. A Quasar phantom (Modus Medical Systems) with moving inserts was used to investigate the effect of motion on CBCT image quality. A systematic pattern of motion artefacts was revealed. Artefacts (created by a high density object moving along the axis perpendicular to the slice) were assessed in an axial slice. Circular profiles were used to quantify the artefacts on the three systems. Clinical image quality was assessed through a qualitative study where two experienced observers independently scored 24 randomly selected clinical thoracic CBCT scans (8 per system). CBCT images were viewed alongside the planning CT scan, with the PTV outline being the only visible delineated structure. Scoring was based on a five point scale and reflected the image quality for matching purposes, the clinical task. Eight anatomical regions, sharpness, contrast, impact of artefacts, and the overall image quality were scored. Comments were also recorded. Results: Quantitative assessment using the Catphan revealed no differences between systems that was deemed significant. The variation in magnitude of the streaking artefacts in the Quasar phantom was found to depend on scan time, but not on the system, as shown in Table 1. The Mann-Whitney test was applied to each observer's scores for each metric of the clinical image analysis. No significant (p<0.05) differences between any systems for any metric for either user were detected. Conclusion: Investigations to date indicate no significant difference between the systems assessed. Image quality must allow matching of the CBCT to CT with confidence. Staff were thus reassured that all systems were assessed as “acceptable” (mean score of 3) for most metrics. It was felt that patient size was often the cause of particularly good or poor scores; therefore improvement of patient size dependent protocols is identified as a key area of future work. (Figure Presented).