2205 - Investigating the Impact of CBCT Acquisition Speed on Normal Tissue Structure Localization Accuracy in the Presence of Subject Motion Using a Novel CBCT Scanner

D. P. Adam¹, A. Lu¹, M. Salerno², B. K. K. Teo³, T. Li⁴, A. Sisniega⁵, and W. T. Hrinivich⁶; ¹Johns Hopkins University School of Medicine, Baltimore, MD, ²Department of Radiation Oncology, University of Pennsylvania, Philadelphia, PA, ³Department of Radiation Oncology, Division of Physics, University of Pennsylvania, Philadelphia, PA, ⁴Varian Medical Systems, Palo Alto, CA, ⁵Johns Hopkins University, Baltimore, MD, ⁶Department of Radiation Oncology and Molecular Radiation Sciences, Johns Hopkins University School of Medicine, Baltimore, MD

Purpose/Objective(s): Cone beam CT (CBCT) remains the clinical standard for image-guided radiotherapy (IGRT) but can suffer from poor image quality including artifacts arising from patient motion occurring during slow image acquisition. Increasing image acquisition speed through faster gantry rotation and larger detector panels enabling short-scan protocols may decrease sensitivity to patient motion. This work systematically assesses the impact of CBCT acquisition speed on image quality in terms of normal tissue structure localization at varying subject motion amplitudes using a phantom imaged using a novel CBCT system with 6 s acquisition time. Materials/

Methods: Images of an anthropomorphic abdomen phantom were acquired with one diagnostic fan beam CT with 0.5 s/gantry rotation, an advanced closed ring CBCT system with 16.6 s/acquisition and a novel high-speed CBCT system with 6 s/acquisition, denoted FB-0.5, CB-16.6, and CB-6, respectively. Sinusoidal motions with 4 s period and amplitudes ranging from ±0.0 to ±10.0 mm were applied during image acquisition on all systems. 27 bony and 9 soft tissue thoracic and abdominal structures were automatically segmented on each image using a validated open-source deep learning approach (TotalSegmentator, 3DSlicer). All images were rigidly registered to a common frame of reference, and contours were compared using distance-to-agreement (DTA) with the stationary FB-0.5 scenario as the comparison benchmark.

Results: Key results are summarized in Table 1. Both CBCT systems provided accurate detection and localization of most normal tissue structures compared to diagnostic FBCT. The CB-6 and CB-16.6 exhibited similar trends in performance for the motion amplitudes =±2.5 mm, but the CB-6 scanner exhibited improved performance at motion amplitudes =±5.0 mm. At ±5.0 mm motion amplitude, mean±SD DTA were 0.47±0.43 mm and 0.95±0.69 mm for the CB-6 and CB-16.6 scanners, respectively.

Conclusion: Increasing CBCT acquisition speed from 16.6 s/image to 6 s/image reduced bony and soft tissue contouring error for motion amplitudes =±5.0 mm. Fast CBCT holds promise to significantly improve image quality in the presence of patient motion for advanced IGRT workflows. Abstract 2205 -Table 1: Mean±SD normal tissue structure distance-to-agreement (DTA) for varying subject motion amplitudes *±5.0 mm motion amplitude was not available for the FB-0.5 scanner

Motion Amplitude (mm)	±0.0	±1.0	±2.5	±5.0	±10.0
FB-0.5 DTA (mm)	0.00±0.00	0.01±0.00	0.07±0.01	n/a*	0.52±0.28
CB-16.6 DTA (mm)	0.11±0.17	0.11±0.16	0.24±0.24	0.95±0.69	2.31±1.97
CB-6 DTA (mm)	0.07±0.07	0.08±0.07	0.24±0.30	0.47±0.43	1.84±1.96