Brigham and Women's Hospital/Dana-Farber Boston, MA, United States
T. C. Harris1, R. Fueglistaller2, R. Bruegger2, D. Ferguson3, M. Jacobson1, M. Myronakis1, Y. H. Hu4, M. Lehmann2, P. Corral Arroyo2, J. OConnell1, V. Birrer2, D. Morf2, and R. I. Berbeco1; 1Department of Radiation Oncology, Brigham and Women’s Hospital, Dana Farber Cancer Institute and Harvard Medical School, Boston, MA, 2Varian Medical Systems, Baden-Dattwil, Switzerland, 3Department of Radiation Oncology, Brigham and Women’s Hospital, Dana-Farber Cancer Institute and Harvard Medical School, Boston, MA, 4Department of Radiation Oncology, Brigham and Women’s Hospital/Dana-Farber Cancer Institute, Boston, MA
Purpose/Objective(s): Dual-energy imaging confers potential advantages including reduced artifacts and material decomposition – e.g. allowing material enhancement for contrast or material suppression to view underlying anatomy better. Spectral separation may be achieved via a dual-layer detector, with beam hardening by the first layer providing the shift in spectra. This approach may offer unique advantages relative to other on-board spectral imaging concepts, in terms of dose savings, motion artifact elimination, and innate view registration. Materials/
Methods: The first layer of the on-board dual-layer imager (DLI) was designed to be identical to existing detector construction, with a CsI scintillator followed by an aSi TFT photodiode array. This enabled seamless clinical implementation, as standard onboard image guidance was maintained, using the top layer only. The bottom layer has a slightly thicker scintillator to aid in photon detection efficiency given the reduced fluence received after the top layer. The DLI prototype was constructed and underwent rigorous safety testing by a commercial partner prior to clinical integration. The DLI was installed on a clinical linac with a novel imaging chain in which the top layer only is used for clinical tasks and both layers are read out to a research computer for retrospective analysis. Modulation transfer function (MTF), noise power spectrum, and detective quantum efficiency (DQE) were measured for the top, bottom, and combined layers. Detector imaging performance was further characterized by Leeds, Catphan, and anthropomorphic phantoms. To date, data collection has been performed for more than 20 patients of diverse disease sites. Results: Clinical installation of the prototype DLI was completed successfully without any interruption to routine workflow. Phantom measurements confirmed that the top layer MTF and DQE were similar to the commercial single-layer imager which had been replaced. Spatial resolution for the combined-layer images were slightly lower than the top layer only but benefited from increased photon detection efficiency. The patients imaged so far include head & neck, pelvis, extremity, thorax, and CNS. All routine treatment imaging and delivery proceeded as usual without interference from the study. Virtual monoenergetic images were generated from the dual-layer CBCT data using a U-net convolutional neural network. In 2D imaging, log-weighted subtraction of two layers successfully removed bone and metal hardware from resulting images, enabling better tumor visualization. Conclusion: A prototype kV DLI was constructed and clinically translated for study under protocol. Combining the layers yields a higher photon detection efficiency with a small loss in resolution. Preliminary clinical results show promise for spectral imaging applications, such as removing ribs to enable better lung tumor imaging. The dual-layer design may be an effective method for adding spectral imaging capabilities to a linac.
