10. Assessment and optimization of PET contours recovery for respiratory-gated radiotherapy

Introduction The development of medical imaging, and more specifically functional imaging, open new perspectives for radiotherapy treatments. The joint use of the dosimetric CT and the PET improve the determination of target volume by combining functional and morphological data. The purpose of this study is to evaluate the recovery process of the PET contours for respiratory-gated radiotherapy according to two methods: the Replanning algorithm (RP) and the module Integrated Registration (IR) from General Electric (GE) Medical System. Methods The 4D acquisitions are achieved on a dosimetric CT Brillance Big Bore® (Philips Medical System) associated to an abdominal belt Pneumo chest Bellows (Lafayette Instrument) and on a PET/CT Discovery 710 (GE Medical System) coupled to the respiratory gating system RPM (Real-Time Position Management™, Varian Medical System). The Quasar™ Programmable Respiratory Motion Phantom (Modus Medical Device Inc) is used with a lung density insert. The lesions are simulated by solid spheres of 1 and 3 cm diameter and another empty one of 2 cm. The acquisitions have been performed with breathing amplitude of 2 and 4 cm and breathing frequency of 10, 15 and 20 bpm. The influence of the acquisition parameters, reconstruction parameters and variability of the patient parameters (amplitude, frequency, lesion size) are being assessed. On the CT images, the spheres are segmented with a threshold allowing to obtain theoretical volume of the sphere on a 3D acquisition. On the PET images, the spheres are segmented with a 42% threshold of the SUVmax. The PET/CT contours obtained with the RP algorithm and the IR module have been compared to the contours delineated on the dosimetric CT with Eclipse v13.5, using the Dice’s coefficient. Results The influence of the acquisition and reconstruction parameters studied with the RP algorithm show Dice’s coefficients ranging from 0.93 to 0.96. Dice’s coefficients variating between 0.85 and 0.94 are obtained for the study of variability of patients parameters. With regard to the two retrieval methods of PET contours, they show similar Dice’s coefficients. For an amplitude of 2 cm and a respiratory frequency of 15 bpm, the Dice’s coefficient is 0.84 with RP algorithm and 0.85 with the IR method. Conclusions For pulmonary lesions, the two retrieval methods of RP or IR contours appear similar. In the field of radiotherapy, a new way to use the replanning algorithm seems possible and allows for the simplification of the recovery of the PET contours. A clinical study is in the process of being evaluated. The objective is to validate this method in lung lesions cases.