Dual-energy X-ray imaging for targeting radiotherapy

The DEXTER consortium, consisting of the radART (Paracelsus Medical University, Salzburg, Austria) and CREATIS (Université Lyon 1, Lyon, France) institutes, aims at exploiting the capabilities of PAIR, a newly developed integrated Patient Alignment and x-ray Imaging Ring system for image guided radiotherapy (IGRT). IGRT has been a booming field of research and PAIR pushes forward the capabilities of in-room 2D and 3D imaging. Three main objectives and fields of research were determined in the DEXTER project: 1. Oblique x-ray projection images can provide large field of view (FOV) computed tomography (CT) images. Today, existing cone-beam CT (CBCT) scanners allow in plane displacement of their flat panel. PAIR on the other hand allows independent rotation of the flat panel and x ray source; it can thus reach much larger FOVs. PAIR is, to our knowledge, the only scanner in the field of IGRT enabling this geometry. However, new reconstruction algorithms are required and will be investigated in DEXTER, therefore increasing the maximal size of 3D CBCT FOV. 2. Future x-ray scanners will exploit multiple x-ray energies. This is an important research field in Lyon because a photon counting spectral scanner will soon be installed in the context of France Life Imaging (FLI). One perspective is the better characterization of tissue composition which is essential in hadron therapy for predicting the range of the ion beam. With PAIR, we have the unique opportunity to investigate the capabilities of dual- energy in the treatment room. DEXTER will provide guidelines on the achievable accuracy of ion range prediction which will be validated on real experiments, the current knowledge being essentially restricted to Monte Carlo simulations. Successful improvement will translate in smaller treatment margins for hadron therapy centers equipped with PAIR units. 3. Finally, new opportunities for 2D/3D intrafraction tumor localization will be explored based on PAIR`s capabilities, namely, multi-energy radiographs and maximal collimation of the x-ray beam. There are two objectives, solving target occlusion difficulties while minimizing the imaging dose. Improving intrafractional position verification and online treatment guidance directly impacts the treatment margins accounting for uncertainties in the target position. If one can also minimize the imaging dose, continuous tracking during treatment delivery might be feasible. The end results are twofold. First, the new knowledge will be disseminated via conventional means, i.e., conferences in the medical imaging field and in the radiotherapy domain as well as corresponding journals. Second, the algorithms, developments and representative data sets will be open source via collaborative platforms that have been initiated by the two partners, i.e., the Reconstrution Toolkit (RTK) for CBCT reconstruction and Reg23 in Plastimatch for 2D/3D registration.

DEXTER was a collaborative project between radART (Austria) and CREATIS (France) where novel dual- nergy x-ray imaging techniques were explored and developed that are intended to improve patient alignment and treatment planning accuracy in radiation therapy. Experiments and methods focused on a novel Cone-Beam Computed Tomography (CBCT) scanner, the ImagingRing (medPhoton, Austria), that offered in a research mode new key options for x-ray imaging: (A) independently moveable x-ray source and detector arms to enable variable fields of view, (B) dual-energy acquisitions by alternating the x-ray energy from radiograph to radiograph with additional energy spectra separation by inserting specific filtration materials into the beam path. The DEXTER project has allowed the development of geometrical calibration methods including design of a suited calibration phantom to being able to exploit (A). Furthermore, new methods for (intensity) calibration of the x-ray images supporting the variable imaging geometry (A) and dual-energy characteristics (B) were developed and verified. Based on these developments and adaptations of pre- existing open source software of the project partners, an efficient tomographic reconstruction algorithm for derivation of volumetric information from ImagingRing raw data, and a 2D/3D image registration method to spatially align a reference patient CT with ImagingRing-acquired x-rays were developed and verified. Besides, a realistic x-ray computer model of the ImagingRing was derived and validated with acquired x-ray data of different energies such that it could be used for the dual-energy developments. Performed dual-energy investigations comprised particle range estimation based on CBCT for potentially improving treatment planning (particle therapy), and separating in the projective x-rays soft and bony tissues to potentially enhance the robustness and precision of 2D/3D registration. The tight cooperation of radART and CREATIS during the DEXTER project and intensive research have led up to now in total to 8 articles in main journals of medical physics and 14 related proceeding contributions at international conferences. A part of the software developments is available via open source platforms which is already in fractions used clinically. MedAustron, Austrias first particle therapy center and first ImagingRing installation, has been an important (pre-)clinical collaborator and supported the project with data acquisition, evaluation and experimental data acquisition. The tight cooperation with this clinical institution and the other involved partners such as medPhoton (ImagingRing vendor) as well as the research results of the DEXTER project make progressive translation of the developed methods into the international clinical theater of radiation therapy in the course of subsequent research projects likely.