ESTRO 38 Abstract book

S1133 ESTRO 38

positioning errors and (in proton therapy) range uncertainties. However, it is important to maintain sufficient image quality for position verification and visualisation of potential anatomical changes during the course of radiotherapy. In this work, we have evaluated the image quality for soft tissue visualisation of low-dose CBCT scans simulated for phantom and paediatric patient acquisitions. Material and Methods CBCT scans of a Catphan 503 phantom were acquired with Elekta XVI v5.02 using 16mA/10ms and 100 kV, 200 projection images. In addition, CBCT scans acquired as part of routine treatment were collected from 10 paediatric patients (aged 3 to 12 years) with tumours in the abdominal-pelvic region. All scans were acquired using the same exposure (16mA/10ms and 100 kV), small field of view, half rotation. Low-dose images were created by adding non-uniform Gaussian noise to the original projection images prior to reconstruction, accurately simulating exposures of 10mA/10ms, 10mA/8ms, and 5mA/5ms while keeping kV constant. In addition, the imaging dose was estimated as fraction of the CTDI for the acquisition dose (0.8 mGy). Contrast-to-noise ratio (CNR) was determined between subcutaneous fat and abdominal/pelvic muscle for the paediatric scans and between the acrylic and LDPE inserts (most similar to fat and muscle tissue, respectively) of the Catphan. Results Figure 1 shows an example of simulated low-dose (8mA/10ms and 5mA/5ms) scans compared to the original (16mA/10ms) image. Figure 2 shows the relationship between CNR and dose in paediatric scans and phantoms. Lower-dose simulations created images with more noise and artefacts than the original scans; however the muscle and subcutaneous fat could still be distinguished at all simulated exposures. CNR remained higher than 3 until around 0.5 mGy in paediatric simulations and above 0.4 mGy in the phantom images (sufficient for observation of the selected contrasts for small objects according to the Rose criterion) (Rose, 1974). There was considerable inter-patient variation in CNR for the same dose, which does not seem to correlate with patient age, suggesting these are related to image artefacts.

For the healthy subjects, the analyses were performed on 3D T2-weighted images (0.9mm isotropic) as these provide optimal contrast to differentiate the sclera, while for the UM patients a post-contrast T1 (1.0mm) was used to differentiate between UM and retinal detachment (RD). The analysis strategy is described in figure 1. For the UM patient only the tumour-vitreous border could be accurately compared, as the fast retinal wash-out of the contrast agent changed the appearance of the eye-wall between both acquisitions.

Figure 1. The image in flexed position is registered to the image in supine position and the sclera in both positions is segmented. Subsequently the distances between both segmented sclera meshes were calculated. Results In healthy controls the median difference between the supine and flexed scans was 0.1mm (95 th percentile (P): 0.3mm), which is in the order of the reproducibility of the method (95 th P: 0.3mm), figure 2-A. The slightly larger difference in eye-shapes of subjects 5 was caused by eye- motion artefacts. In the UM patient we found a median UM deformation of the tumour of 0.1mm (95 th P: 0.4mm).

Figure 2: A. Violin plots of the measured distances. Controls (C1&C2) show a measurement reproducibility of 0.3mm. The subjects (S1–S7) and UM patient show similar distances as the controls. B. 2D T2 showing the UM and Retinal detachment (RD). C. 3D-T1 in supine position with segmented UM. D. Registered 3D-T1 in flexed position with segmented UM. Conclusion Changes in gravity direction produce no substantial changes in sclera and tumour shape. We are currently expanding the number of UM patients, but the results so far indicate that supinely acquired MR images can be used to accurately plan ocular PBT, which is performed in sitting position. EP-2059 Assessment of image quality in simulated low-dose paediatric cone beam CT J.W.H. Lindsay 1 , A. Bryce-Atkinson 1 , M. Van Herk 1 , M.C. Aznar 1 1 The University of Manchester, Faculty of Biology- Medicine and Health, Manchester, United Kingdom Purpose or Objective Reduction of dose to healthy tissue is imperative to limit the risk of secondary cancers in paediatric patients. Lower imaging dose would naturally reduce this risk and allow for more frequent imaging, which may reduce dose due to