A Scanner-Specific Simulation Platform for Virtual Imaging Trials in Deep Si-Based Photon-Counting CT
S Sharma1*, D Pal2, E Abadi3, P Segars4, J Hsieh5, E Samei6, (1) Duke University, Durham, NC, (2) GE Healthcare, Menlo Park, CA, (3) Duke University, Durham, NC, (4) Duke University, Durham, NC, (5) GE Healthcare, Milwaukee, WI, (6) Duke University, Durham, NC
Presentations
WE-D-TRACK 3-6 (Wednesday, 7/28/2021) 2:00 PM - 3:00 PM [Eastern Time (GMT-4)]
Purpose: Deep Si-based photon-counting CT (Si-PCCT), a candidate emerging CT detector technology, is expected to provide multiple benefits by virtue of fast charge collection times, reduced K-fluorescence crosstalk, and ease of manufacturing. To translate Si-PCCT to clinical use, its performance needs to be comprehensively evaluated and optimized in a task-specific manner. Often performed through large-scale imaging studies on human subjects, such evaluations are impractical for Si-PCCT due to the lack of access to clinical prototype systems. This limitation can be overcome by using computational anthropomorphic phantoms and imaging simulators, in a process known as virtual imaging trials (VITs). To enable such VITs, we present a scanner-specific simulation platform to simulate realistic clinical images for Si-PCCT systems.
Methods: A hybrid simulation platform for VITs (DukeSim) was extended to model the scanner-specific geometry and components of a PCCT prototype, including models incorporating energy-dependent response and crosstalk characteristics for a deep Si-based detector. To simulate images, a potential pilot system was emulated to image a high-resolution (0.1 mm) virtual patient with a lung lesion. The obtained projections with energy-binned photon counts were processed to introduce Poisson noise, summed across energy bins, and corrected for beam hardening before reconstruction. To demonstrate the utility of this platform, the reconstructed images were utilized to qualitatively and quantitatively compare the spatial resolution performance of PCCT to conventional energy-integrating CT (ECT), including their relative performance for estimation of morphological radiomics features.
Results: The platform successfully simulated realistic images for prototype Si-PCCT designs. The evaluation of spatial resolution revealed superior performance of Si-PCCT over ECT both qualitatively and quantitatively, with a dramatic improvement of accuracy (30.7% mean) in the estimation of morphological radiomics features.
Conclusion: The developed simulation platform simulated realistic images for a pilot Si-PCCT system, enabling the qualitative and quantitative evaluations integral to performing clinical VITs.
Funding Support, Disclosures, and Conflict of Interest: This research was partly supported by GE Healthcare.
Handouts
Keywords
CT, Simulation, Image Analysis
Taxonomy
Contact Email