Abstract Book

ESTRO 37

S540

ge of motion could be recovered with a median error of - 3.3% (range=[6.12%; 7.2%]) (Figure 1c). The time bin that showed the smallest SUVmax variation from the static sit uation was not consistently the same but varied across m otions and institutions (Figure 2). Similar results were obt ained for the variations in SUVpeak.

formula has relevant potentials in guiding strategies for ART and treatment individualization.

PO-0977 Assessment of motion and SUV recovery in 3D and 4D PET/CT: a multicentre phantom study K.J. Ortega Marin 1 , M. Lambrecht 1 , A.J. Arends 2 , S. Adebahr 3 , U. Nestle 4 , W. Vogel 5 , M. Verheij 6 , J.J. Sonke 6 , C.W. Hurkmans 1 1 Catharina Hospital, Radiation Oncology, Eindhoven, The Netherlands 2 Catharina Hospital, Medical Physics, Eindhoven, The Netherlands 3 University Medical Center Freiburg, Radiation Oncology, Freiburg, Germany 4 University Medical Center Freiburg / Kliniken Maria Hilf, Radiation Oncology, Freiburg / Mönchengladbach, Germany 5 The Netherlands Cancer Institute, Nuclear Medicine, Amsterdam, The Netherlands 6 The Netherlands Cancer Institute, Radiation Oncology, Amsterdam, The Netherlands Purpose or Objective In PET/CT studies, tumour motion results in the apparent decrease of SUV and apparent increase of its volume in t he PET image with respect to static conditions. The goal of this multicentre dynamic phantom study is to quantify the variation in SUV and motion recovery in 3D and 4D PE T/CT images. Material and Methods A customized CIRS- 008A phantom was scanned at 13 institutions participatin g in the EORTC LungTech trial on SBRT for centrally locat ed lung tumours. The phantom body contains a moving cy linder of 63.5 mm diameter with a sphere of either 15 m m or 25 mm diameter placed inside it. The body, cylinder and sphere were filled with homogeneous 18F- FDG solutions representative of activity concentrations in mediastinum, lung and tumour, respectively. 3D and 4D PET/CT scans were acquired for static and dynamic tumo urs. Motion was simulated with cos 6 respiratory patterns o f various amplitudes and periods: 15mm/6s, 15mm/3s an d 25mm/4s. Institutional scan protocols were used. SUVm ax and SUVpeak (average SUV of the volume encompassed by a 70% SUVmax threshold) for all scans acquired were a ssessed centrally using MATLAB. To recover the range (am plitude) of motion (ROM), the centre of mass of the tumo ur was estimated from a volume segmented for SUVs larg er than 40% of the SUVmax in each bin (respiratory phase) . SUV metrics were normalized to their corresponding sta tic values. Wilcoxon rank tests were used for assessment of SUV variations. Results Variations in SUVmax due to tumour motion with 3D PET/ CT were the largest for the 25mm/4s pattern (median=- 36.4%, range=[44.7%; 268.4%]) and significantly different from the 15mm/6s pattern (-27.6% [-36.3%; -9.3%], p=0.037) and from the 15mm/3s pattern (-23.6% [- 35.3%; 10.7%], p=0.009) (Figure 1a). These variations dec reased significantly with 4D PET/CT (p<0.001): -10.4% [- 35.2%; 241.1%] for the 25mm/4s pattern, -5.1% [ 42.2%; 81.3%] for the 15mm/6s pattern and -5.1% [- 32.3%; 41.4%] for the 15mm/3s pattern (Figure 1b). There were no significant differences in the variations between patterns with 4D PET/CT (p>0.15). Nevertheless, a larger spread in variations was observed in 4D PET/CT for the 1 5mm/6s and 15mm/3s patterns. Moreover, an overestima tion of SUVmax could be sometimes observed with 4D PET /CT (Figure 1b). For the acquisitions with 10 bins, the ran

Conclusion 4D PET/CT scans provide the potential to recover SUV an d range of motion, although a large variation of SUV amo ng bins and institutes is seen. The selection of not one, b ut all bins in 4D PET/CT may be a better approach to acc urately recover SUV. Further investigation is necessary to propose guidelines for optimal acquisition and reconstruc tion of 4D PET/CT.