ICTR-PHE 2016 S21 A least absolute shrinkage and selection operator (LASSO) method was used for feature selection. Model performance was evaluated using Harrell’s concordance-index (c-index). Fitted model included sum entropy (GLCM), high intensity large area emphasis (GLSZM), volume with a minimum relative intensity of 60% of the maximum SUV – AVRI60% (IVH), grey level non-uniformity and long run emphasis (RLGL) and volume (shape) – Table 1. Internal performance of the model was 0.64 (p<0.01), while externally it achieved a performance of 0.61 (p = 0.05) and 0.58 (0.20), with no further calibration done. Maximum and mean SUV had a univariable performance in the training data of 0.51 and 0.55, respectively. The reduced accuracy of the model validation can be associated with dissimilarities among data, particularly the different timing and delivered dose of the second scan. Nevertheless, we do see benefit on a timely assessment of response to radiotherapy using the described imaging analysis, particularly when compared with the limited capacity of humans to infer accurate predictions and risk groups identification ( 5). From the Radiomics analysis one can optimally benefit from early response metrics based on changes in metabolism measured with FDG-PET, even before anatomic changes become noticeable, while treatment can still be adapted. We developed and validated a predictive model on the percentage variation of Radiomics features, the so-called “ Delta Radiomics” concept, from repeated FDG-PET scans of NSCLC pat ient s. Keywords: Early response assessment, non-small cell lung cancer, 18 F-FDG-PET imaging, metabolic metrics References: [1] Lambin P, Rios-Velazquez E, Leijenaar R, et al. Radiomics: extracting more information from medical images using advanced feature analysis. Eur J Cancer. 2012; 48: 441- 446. [2] van Elmpt W, Ollers M, Dingemans AM, Lambin P, De Ruysscher D. Response assessment using 18F-FDG PET early in the course of radiotherapy correlates with survival in advanced-stage non-small cell lung cancer. J Nucl Med. 2012; 53: 1514-1520. [3] Yossi S, Krhili S, Muratet JP, Septans AL, Campion L, Denis F. Early assessment of metabolic response by 18F-FDG PET during concomitant radiochemotherapy of non-small cell lung carcinoma is associated with survival: a retrospective single- center study. Clin Nucl Med. 2015;40:e215-221. [4] Leijenaar RT, Carvalho S, Velazquez ER, et al. Stability of FDG-PET Radiomics features: an integrated analysis of test- retest and inter-observer variability. Act a Oncol. 2013; 52: 1391-1397. [5] Oberije C, Nalbantov G, Dekker A, et al. A prospective study comparing the predictions of doctors versus models for treatment outcome of lung cancer patients: a step toward individualized care and shared decision making. Radiot her Oncol. 2014; 112: 37-43. 43 Proton scattering radiography using an emulsion detector: a feasibility study A. Ariga 1 , T. Ariga 1 , M. Auger 1 , S. Braccini 1 , T. Carzaniga 1 , A. Ereditato 1 , K. P. Nesteruk 1 , C. Pistillo 1 , P. Scampoli 1,2 1 Albert Einstein Center for Fundamental Physics (AEC) - Laboratory for High Energy Physics (LHEP), University of Bern 2 Department of Physics University of Naples Federico II Purpose: Proton radiography is an imaging technique in proton therapy giving direct information on the density of the tissues, a useful tool to enhance the precision of proton therapy. It is usually performed as a proton range radiography by measuring the position and the residual range of the protons after the target. The properties of the traversed materials are directly related to multiple scattering so that it is possible to obtain an image through the assessment of the proton angular distribution. This work aims at studying the possibility of performing proton scattering radiography using only one nuclear emulsion film. Materials and methods: Nuclear emulsion films allow for high- precision tracking of charged particles and, in particular, for reconstructing their angular distribution with a resolution of the order of 1 mrad. Specific detectors for medical applications can be built by interposing double-sided emulsion films with tissue equivalent materials, as it was done for proton range radiography [1] and to study the halo of a proton pencil beam [2]. In the present study, a detector composed by only one emulsion film was exposed to a 138 MeV proton pencil beam at the Gantry 1 at PSI. Two phantoms were placed in front of the detector: the “ step” phantom consisted of two different thicknesses of PMMA (3 and 4 cm, respectively); the “rod” phantom had a total thickness of 4.5 cm and contained five aluminum rods (5 × 5 mm 2 section) positioned at different depths in a PMMA structure. Following the chemical development and the automatic microscopic scanning of the emulsion film, proton tracks were identified and their angular distribution reconstructed. Results: The RMS of the scattering angle was measured for different segmentations of the emulsion film. Areas were chosen as strips parallel to the direction of the step or of the rods, for the first and the second phantom, respectively. To evaluate the resolution, strips of different sizes were considered. As shown in figure 1 (left), the step is clearly identified as a sharp drop of the RMS of the scattering angle. The signal due to the rods is visible as an enhancement of the RMS corresponding to their positions. The rod located nearest to the detector shows a sharper peak while the farthest one appears broader due to the larger distance travelled by the protons. While the contrast for the step phantom is found to be basically the same for range and scattering proton radiography, the signal due to the rods is more evident with respect to what was obtained with proton range radiography [1]. These preliminary results suggest that atomic number plays a fundamental role to increase the contrast of the image. Conclusions: A feasibility study of proton scattering radiography with a new method based on a single emulsion detector has been performed. The first preliminary results are promising and further studies are under way encompassing in particular Monte Carlo simulations. Keywords: Proton radiography, proton therapy, nuclear emulsion detectors References: [1] S. Braccini et al., First results on proton radiography with nuclear emulsion detectors, Journal of Instrumentation 2010 JINST 54 P09001. [2] A. Ariga et al., Characterization of the dose distribution in the halo region of a clinical proton pencil beam using emulsion film detectors, Journal of Instrumentation 2015 JINST 10 P01007. source: https://doi.org/10.7892/boris.83509 | downloaded: 23.5.2020