ESTRO 38 Abstract book

S1128 ESTRO 38

[2] R.T. Shinohara, E.M. Sweeney, J. Goldsmith, N. Shiee, F. J. Mateen, P. A. Calabresi, S. Jarso, D. L. Pham, D. S. Reich, and C. M. Crainiceanu, “Statistical normalization techniques for magnetic resonance imaging,” NeuroImage Clin. , vol.6, pp.9–19, 2014. EP-2052 Commissioning monoenergetic CT images for optimal proton dose calculations using TwinBeam scans P. Randers 1 , M. Fuglsang 1 , V. Trier Taasti 1 1 Aarhus University Hospital, Danish Centre of Particle Therapy, Aarhus, Denmark Purpose or Objective Monoenergetic CT images derived from two CT images acquired at different kVps offer the potential of reducing the beam hardening (BH) effect compared to single energy CT (SECT) images. Commissioning monoenergetic CT images for proton dose calculations could therefore lead to a reduced range uncertainty. The first step is to determine the optimal energy leading to the lowest possible BH effect, and then to establish stoichiometric calibration curves from CT numbers to proton stopping power ratios (SPRs) and mass densities. Material and Methods CT scans of a tissue equivalent electron density phantom (Gammex, Middleton, WI) were acquired with a SOMATIOM Definition Edge dual energy CT (DECT) scanner (Siemens Healthineers, Forchheim, Germany). This is a single source scanner with a TwinBeam (TB) mode (Fig.1), which enables simultaneous dual energy acquisition, which reduces the risk of motion blurring. CT scans were acquired in both TB and SECT mode. Seven different clinically relevant configurations were used and fifteen different tissue equivalent inserts were scanned individually within each configuration (Fig 1). All remaining positions were filled with solid water inserts of the same composition as the bulk phantom, to avoid interference between the different inserts. Monoenergetic images were generated using the Siemens syngo.via software at different energies (monoEs). The size of the BH effect was quantified as the standard deviation of the monoenergetic CT numbers over the various configurations for a specific insert at a given monoE. The BH effect was analysed for all inserts together by summing over the inserts, as well as for each insert separately. The CT number differences for the SECT scans of the seven configurations were also extracted and used to investigate if the TB monoenergetic CT images decreased the BH compared to SECT. After obtaining the optimal monoE, stoichiometric calibration curves were fitted for the small and the large phantom, respectively.

method. For T2w-images, JD was 0.182[0.496], 0.036[0.038] and 0.095[0.152] for M1, M2 and M3, with respect to 0.090[0.142] in raw images. M2 was found to outperform all the other methods according to JD, whereas the use of M1 was supported by npCV (Tab. 1). For both T1w- and T2w-images it was noticed that WM variability measured by npCV was lower than the GTV one, apart from M2 for which median values were close to zero. Differences in variability varied for different normalization methods, but all of them qualitatively kept the WM-GTV relation observed in raw images (Fig.1). Conclusion Overall, M1 showed better performance for T1w-images and both M1 and M2 for T2w-images. M3 is the simplest method but it does not seem to guarantee a comparison between MRI acquisitions. The evaluation of the performance of different MR normalization methods is not trivial but required for MRI analysis in longitudinal studies, as well as for applications in feature extraction for patient classification or stratification. Harmonized evaluation of such (and other) pre-processing method is therefore needed.

Bibliography : [1] L.G. Nyul, J.K. Udupa, and Xuan Zhang, “New variants of a method of MRI scale standardization,” IEEE Trans. Med. Imaging , vol.19, no.2, pp.143–150, 2000.