Purpose: Prior to thoracic radiotherapy treatments, patients are commonly aligned using conventional4DCBCT scans (1320 projections over 4 minutes). The ADAPT clinical trial performs adaptive 4DCBCTscans by modulating gantry rotation speed and projection acquisition in response to real-time changesin the patients breathing rate with as low as 200 projections. This study determines reconstructionalgorithms that best complement adaptive 4DCBCT acquisitions for reducing imaging dose and scantime whilst maintaining or improving image quality compared to conventional 4DCBCT acquisitions from20 ADAPT clinical trial patients (NCT04070586).
Methods: Adaptive acquisitions from 20 patients treated for lung cancer as part of the ADAPT clinicaltrial were used in this study. Each patient dataset consisted of a planning 4DCT and 3 4DCBCT scansacquired on two treatment fractions (conventional 1320, adaptive 600 and adaptive 200 projectionscans were acquired). All scans were reconstructed using Feldkamp-Davis-Kress (FDK) as a comparator, and the adaptive scans are reconstructed using Mckinnon-Bates (MKB), Motion Compensated MKB (MCMKB) andmotion compensated FDK (MCFDK). Reconstructed image quality was quantified using StructuralSimilarity Index (SSIM), Signal-to-Noise-Ratio (SNR), Contrast-to-Noise-Ratio (CNR), Tissue InterfaceSharpness at the diaphragm (TIS-D) and tumor (TIS-T).
Results: 200 and 600 projection adaptive 4DCBCT acquisition delivered 85% and 55% less imagingdose and 62% and 4% shorter scan times, respectively. Motion compensated reconstructionsmaintained or improved image quality metrics comparable to conventional reconstructions. The 200projection adaptive scan with MCFDK (SNR=21.0±9.9, CNR=13.0±4.9, TIS-D=0.8±0.9, TIS-T=0.8±0.6)and MCMKB (SNR=22.1±10.7, CNR=14.1±5.3, TIS-D=1.3±1.1, TIS-T=0.9±0.3) reconstructionsimproved image quality metrics compared to conventional scans (SNR=11.9±4.2, CNR=9.5±3.1, TISD=0.58±0.3, TIS-T=0.4±0.1).
Conclusion: Combining adaptive acquisition with motion compensation allows significant reductionsin 4DCBCT scan times to be realized in the clinic on a standard linac.
Funding Support, Disclosures, and Conflict of Interest: This study is funded by NHMRC grant 1138899 and partly by a Cancer Australia grant number 1123068. RO is supported by a Cancer Institute NSW career development fellowship and PK is supported by an Australian NHMRC Senior Principal Research Fellowship.