10. Thomas Harris, PhD

Job Title

Physicist

Academic Rank

Instructor

Department

Radiation Oncology

Authors

Thomas Harris*, PhD, Matthew Jacobson, PhD, Dianne Ferguson, PhD, Yue-Houng Hu, PhD, Raphael Bruegger, Rony Fueglistaller, Marios Myronakis, PhD, Roshanak Etemadpour, PhD, Ross Berbeco, PhD.

Categories

Tags

Clinical Translation of a Novel Prototype Dual-Layer Imager for On-Board Imaging

Scientific Abstract

Purpose: Dual-energy imaging can provide several benefits, including artifact reduction and material decomposition. This imaging may be achieved via multiple detecting layers for which beam hardening between layers creates spectral separation. We describe the construction, characterization, and clinical translation of a novel prototype dual-layer flat-panel detector for kV on-board radiotherapy imaging.

Methods: The dual-layer imager (DLI) was designed with the first layer matching existing current detector construction, with a CsI scintillator and a-Si TFT photodiode array. The second layer has a slightly larger CsI scintillator, to provide extra photon detection efficiency. The prototype DLI was built and initial safety testing performed at Varian’s iLab (Baden, Switzerland). For clinical deployment, top layer information is sent to the treatment console, while the data from both layers is sent to a research PC for retrospective analysis.

Modulation transfer function (MTF(f)) was measured with an edge phantom and noise power spectrum (NPS(f)) was measured with flood-field images for the top, bottom, and combined layers. Detective quantum efficiency (DQE(f)) was calculated from the results. Leeds, CatPhan, and anthropomorphic pelvis phantoms were imaged to further assess detector performance. Initial patient data was analyzed to determine material decomposition feasibility.

Results: The DLI was deployed on a clinical TrueBeam linac as described. For 120kVp the MTF50 (cycles/mm) is top layer:1.327; bottom layer:0.843; combined:0.965. DQE(0) top layer:0.51; bottom layer (recalculated with attenuated fluence):0.64; combined:0.60. Leeds contrast-to-noise ratio (CNR) with top layer as baseline, bottom layer:-45.8%; combined:+12.4%. CatPhan analysis saw a similar CNR trend. Initial log-weighted subtraction of layers from a patient’s lung kV shows promise in material suppression.

Conclusion: A novel kV DLI was constructed and translated on a clinical linac. Combining the layers increases DQE(0) with only small resolution decrease. Preliminary results show promise for spectral imaging applications, such as bone removal to enable better lung tumor visualization.