ESTRO 2020 Abstract book

S1006 ESTRO 2020

approximately 30% of the estimated motion, potentially leading to more exposure of normal tissue. PO-1724 Intelligent 4D CT sequence scanning (i4DCT): First prototype measurements R. Werner 1 , T. Sentker 1 , F. Madesta 1 , T. Gauer 2 , C. Hofmann 3 1 University Medical Center Hamburg-Eppendorf, Dept. of Computational Neuroscience / Dept. of Radiotherapy and Radio-Oncology, Hamburg, Germany ; 2 University Medical Center Hamburg-Eppendorf, Dept. of Radiotherapy and Radio-Oncology, Hamburg, Germany ; 3 Siemens Healthcare, Advanced Therapies, Forchheim, Germany Purpose or Objective 4D CT imaging is an integral part of 4D radiotherapy treatment planning. However, due to breathing irregularity during data acquisition, clinical 4D CT image data often contain motion artifacts that affect treatment planning quality. To reduce such artifacts, approaches to guide 4D CT data acquisition by online respiratory signal analysis have been described, but existing work was always theoretical and presented as pure in-silico studies. Here, we present the first prototype implementation on a CT scanner and respective measurements for a recently introduced respiratory signal-guided (RSG) 4D CT concept, the so-called intelligent 4D CT sequence scanning (i4DCT; Med Phys 46(8):3462-74). Material and Methods Core blocks of i4DCT are an initial learning period to establish a patient-specific reference breathing cycle representation for data-driven i4DCT parameter selection and online respiratory signal-guided sequence mode scanning to fulfill 4D CT projection data coverage requirements. i4DCT was implemented as a fully automated workflow on a Siemens SOMATOM go. platform. Respiratory signals were acquired using the Varian RPM system. Measurements were performed using the CIRS Dynamic Thorax Phantom with a customized wooden motion insert with oblique titanium plates (see figure). Phantom motion curves were programmed to systematically vary from regular to very irregular, covering typical irregularity patterns like breathing period variability, breathing pauses and signal amplitude variations. i4DCT-acquired and -reconstructed 4D CT images were finally compared to corresponding images acquired by routine spiral 4D CT imaging (retrospective 10- phase reconstruction; similar mAs setting and noise characteristics for both protocols). Results i4DCT was successfully implemented on a clinical CT scanner platform. Applying i4DCT, 4D CT motion artifacts were strongly reduced for all investigated breathing irregularity patterns when compared to routine spiral 4D CT scanning. A case with pronounced irregularity of breathing period and amplitude can be found in the attached figure (from top to bottom: breathing curve and, in green, beam-on period during standard spiral 4D CT scanning; corresponding axial and sagittal views and highlighted artifacts; same breathing curve and i4DCT beam-on periods; axial and sagittal views for i4CDT).

Results For the PlAcq Reference, the MaxIE strategy led to a mean t out of 3.4%. For the Min95 strategy this was 15%; significantly higher than the expected 5% of outliers (Fig 2a). For the FAcqD Reference t out became lower: 0.8% for MaxIE and 6.0% for Min95. These values for Min95 showed that the intrasession variation was smaller than the intersession variation. The intraday variation appeared largely due to outliers and in this case the outlier rejection strategy had the intended effect. The comparison of the thresholds of MaxIE and Min95 showed an average difference of 1 cm (30% of the reconstructed motion (Fig. 2b)).

Conclusion Using a daily reference leads to a better estimation of the motion during treatment than only using a planning MRI. When comparing MaxIE to Min95 it can be seen that including the diaphragm for 5-10% more of the time within the thresholds, these thresholds had to be increased by