Encrypted login | home

Program Information

Respiratory Motion Guided 4DCBCT: An XCAT Study to Analyse Image Quality in Fast Scans (1-2min)

no image available
R O'Brien

R O'Brien1*, J Sonke2 , P Keall1 , (1) University of Sydney, Camperdown, New South Wales, (2) Netherlands Cancer Institute, Amsterdam, NH


TH-AB-601-10 (Thursday, August 3, 2017) 7:30 AM - 9:30 AM Room: 601

Purpose: We have developed, and implemented a method on an Elekta Synergy linac, called respiratory motion guided 4DCBCT (RMG-4DCBCT). With RMG-4DCBCT, the gantry speed and projection frequency are varied in real-time during 4DCBCT acquisition in response to the patient’s real-time respiratory signal to reduce streaking artefacts and to suppress duplicate projections. The purpose of this study was to determine if reconstructed images (with FDK) and short scan times (1-2min) contain enough information to reliably segment the XCAT lesion.
Methods: The XCAT software phantom was used to simulate 4DCBCT acquisition (120kVp + Poisson noise) with 20 (RMG-4DCBCT_20), 30 (RMG-4DCBCT_30) and 45 (RMG-4DCBCT_90) full fan projections per respiratory phase bin for 44 lung cancer patient breathing traces. Scan times were recorded and a region growing algorithm was used to segment the lesion. The segmentation failure rate, contrast to noise ratio (CNR) and error in tumor diameter in all three directions and phase bins (via the full width at half maximum) were calculated.
Results: Scan times of 240s, 62±20s, 75±24s and 101±34s were recorded for conventional, RMG-4DCBCT_20, RMG-4DCBCT_30, and RMG-4DCBCT_45 respectively. The CNR for conventional 4DCBCT dropped from 12 for patients with 3s breathing periods to 4 for patients with 10s breathing periods compared to 4.5±0.5, 5.4±0.4 and 6.4±0.4 for RMG-4DCBCT_20, RMG-4DCBCT_30 and RMG-4DCBC_45 respectively. The failure rate in segmenting the lesion was 13%, 13%, 9% and 5% with conventional 4DCBCT, RMG-4DCBCT_20, RMG-4DCBCT_30, and RMG-4DCBCT_45 respectively. For successful segmentations, the tumour diameter errors were 1.5±1.0mm, 2.3±0.7mm, 2.0±0.6mm and 1.7±0.5mm for conventional 4DCBCT, RMG-4DCBCT_20, RMG-4DCBCT_30, and RMG-4DCBCT_45 respectively.
Conclusion: RMG-4DCBCT with short scan times produces consistent image quality from which the tumour can be segmented. However, with FDK, the contrast suffers due to the low project count so the next phase of this research will examine the ability of iterative techniques to improve contrast.

Funding Support, Disclosures, and Conflict of Interest: This project is supported by an NHMRC Senior Professorial Research Fellowship, NHMRC project grant 1034060 and Cancer Australia grant number 1084566.

Contact Email: