ESTRO 37 Abstract book

S1148

ESTRO 37

EP-2087 Can respiratory-induced diaphragm motion in children be reliably estimated from a single 4D-CT? S. Huijskens 1 , I.W.E.M. Van Dijk 1 , J. Visser 1 , B.V. Balgobind 1 , C.R.N. Rasch 1 , A. Bel 1 1 Academic Medical Center, Radiation Oncology, Amsterdam, The Netherlands Purpose or Objective In adults, a single pre-treatment four-dimensional CT (4D-CT) acquisition is often used to account for respiratory-induced target motion during radiotherapy. However, studies have indicated that a 4D-CT is not always representative for respiratory motion. Paediatric data on respiratory-induced target motion during radiotherapy is limited. Our aim was to investigate if respiratory-induced diaphragm motion during radiotherapy in children can be reliably estimated from a single pre-treatment 4D-CT. Material and Methods To investigate the predictive value of a 4D-CT for respiratory-induced diaphragm motion (as a surrogate for target motion), nine patients (mean age 14yrs; range 8.5– 17.9yrs) were retrospectively included when the diaphragm was visible on upper abdominal or thoracic free breathing imaging data. From each patient, a 4D-CT for planning purposes, and daily/weekly CBCTs (total 102; range 4–32 per patient) acquired prior to dose delivery were available. For each CBCT, a two- dimensional Amsterdam Shroud image was created, allowing for selection of the cranial-caudal position of the end-inspiration and end-expiration positions of the top of the right diaphragm dome. Pixel coordinates were corrected for the scanner geometry and translated into millimetres relative to the patients’ isocenter. The mean peak-to-peak amplitude was defined as the average displacement between end-inspiration and end-expiration diaphragm positions. Additionally, the peak-to-peak diaphragm motion, corresponding to the difference in position of the diaphragm in end-inspiration and end- expiration phases, was extracted from the 4D-CTs, and compared to the distributions of diaphragm motion in CBCTs (one-sample t-test). Results On average, peak-to-peak diaphragm motion over all CBCTs was 9.6 mm, and peak-to-peak diaphragm motion in 4D-CT was 9.3mm. For 7 out of 9 patients, peak-to- peak diaphragm motion on 4D-CT differed significantly (p<0.05) from diaphragm motion on CBCT (Figure 1). Differences >3mm were found in 60 of the 102 fractions (59%), mostly with 4D-CT underestimating daily/weekly respiratory-induced diaphragm motion.

processing software. CT and MR image data of six pediatric patients (mean age of 2.4±0.4 years, images covered the torso from neck to thighs) was adopted for the construction of the atlas and the conversion algorithms. The method was tested for the same patients with a leave-one-out strategy. The sCT quality was evaluated by comparing sHUs to actual CT-based HUs. Dose calculation accuracy in the sCT images was quantified by three “virtual” target volumes in each patient. The target volumes were positioned into treatment sites lacking prior research; the lungs, the vertebra (thVI-X), and the liver. Five-field IMRT plans were constructed for the lung and liver cases and a three-field posterior plan for the thoracic vertebrae using 6 MV photons. Results The software transformed a standard T2-weighted MR image to a sCT image within ~5 minutes (3-5 minutes for atlas segmentation and 10-30 seconds for the MR intensity to sHU conversion). Figure 1 shows an example of a generated sCT, the original MR, and the reference CT image. The mean differences (CT-sCT) between sHUs and actual HUs were, 16±51, -20±29 and, -18±44 HUs for lung, soft tissues and, bone respectively. The dose-volume- histogram (DVH) parameter D98%, D50% and, D2% differences (CT-sCT) between target volumes in sCTs and real CTs in liver PTV were -0.4±0.4, -0.5±0.3 and, - 0.7±0.2 – in lung PTV 0.5±1.6, 0.1±1.4 and, -0.2±1.3 – in vertebra PTV 0.2±0.5, 0.1±0.2 and, 0.2±0.1, respectively. Figure 2 presents the DVH curves for sCT and CT images of the six cases for the lung, vertebrae, and liver plans, respectively.

Figure 1: Example of the MR, constructed sCT and, reference CT images.

Figure 2: DVH comparison of the three target sites.

Conclusion The study shows feasibility of generating high quality sCT images from a standard T2-weighted MR image. The method can be applied either separately in different body parts or for larger body volumes. The work continues by verifying feasibility of the method for targets located anywhere in the body and treated either with photons or protons.

Conclusion For most patients, respiratory-induced diaphragm motion estimated from a single pre-treatment 4D-CT was not representative for respiratory-induced diaphragm motion during treatment. Respiratory-induced diaphragm motion is very variable and not well predicted by a single measurement. Hence, when respiratory-induced target