4D CT amplitude binning for the generation of a time-averaged 3D mid-position CT scan

Abstract

The purpose of this study was to develop a method to use amplitude binned 4D-CT (A-4D-CT) data for the construction of mid-position CT data and to compare the results with data created from phase-binned 4D-CT (P-4D-CT) data. For the latter purpose we have developed two measures which describe the regularity of the 4D data and we have tried to correlate these measures with the regularity of the external respiration signal. 4D-CT data was acquired for 27 patients on a combined PET-CT scanner. The 4D data were reconstructed twice, using phase and amplitude binning. The 4D frames of each dataset were registered using a quadrature-based optical flow method. After registration the deformation vector field was repositioned to the mid-position. Since amplitude-binned 4D data does not provide temporal information, we corrected the mid-position for the occupancy of the bins. We quantified the differences between the two mid-position datasets in terms of tumour offset and amplitude differences. Furthermore, we measured the standard deviation of the image intensity over the respiration after registration (σregistration) and the regularity of the deformation vector field (Delta J) to quantify the quality of the 4D-CT data. These measures were correlated to the regularity of the external respiration signal (σsignal).The two irregularity measures, Delta J and σregistration, were dependent on each other (p<0.0001, R2=0.80 for P-4D-CT, R2=0.74 for A-4D-CT). For all datasets amplitude binning resulted in lower Delta J and σregistration and large decreases led to visible quality improvements in the mid-position data. The quantity of artefact decrease was correlated to the irregularity of the external respiratory signal.The average tumour offset between the phase and amplitude binned mid-position without occupancy correction was 0.42 mm in the caudal direction (10.6% of the amplitude). After correction this was reduced to 0.16 mm in caudal direction (4.1% of the amplitude). Similar relative offsets were found at the diaphragm. We have devised a method to use amplitude binned 4D-CT to construct motion model and generate a mid-position planning CT for radiotherapy treatment purposes. We have decimated the systematic offset of this mid-position model with a motion model derived from P-4D-CT. We found that the A-4D-CT led to a decrease of local artefacts and that this decrease was correlated to the irregularity of the external respiration signal.

Evaluation of gantry speed on image quality and imaging dose for 4D cone-beam CT acquisition

“…In particular, localization error associated with the treatment of moving targets in lung SBRT must be minimized [ 3 ]. The ability of four-dimensional CT (4DCT) to map motion and tissue deformation during respiration, while reducing artifacts, allows for accurate targeting of tumors in the thorax [ 4 , 5 ]. Strategies to improve target coverage, such as breath hold treatments, gated delivery, and mid-position treatments can also be utilized as part of an approach for respiratory motion management [ 4 , 6 , 7 ].…”

Quantification of Esophageal Tumor Motion and Investigation of Different Image-Guided Correction Strategies

“…A midposition pCT scan was subsequently reconstructed from the 4DCT scan for delineation and dose calculation. 11,12 The GTV was contoured on the pCT by the radiation oncologist with all diagnostic information of diagnostic CT, positron emission tomography CT, endoscopy, or EUS with fine-needle aspiration, reflecting clinical practice. Before December 2016 the clinical target volume (CTV) was defined as follows: the CTV comprised the GTV, extended with a 35-mm cranial-caudal (CC) margin and a 5-mm margin in the transverse plane and was extended to include all pathologic lymph nodes.…”

Investigation of internal target volumes using device and deviceless four-dimensional respiratory monitoring systems for moving targets in four-dimensional computed tomography acquisition

“…[ 23 ] AveIP images were used in our study because of their importance in creating contours in the clinical practice and to avoid the additional complexity of contouring each phase. [ 14 ] The AveIP images were exported to Monaco (version 5.10) treatment planning system (Elekta, Crawley), and the ITV delineation was done with a consistent window width of 600 and window level of 40.…”

“…[ 13 ] The more recent studies show that more robust 3D planning CT can be created which is less susceptible to respiration artifacts by registering all frames to the time-weighted average position and taking the median or mean over time (mid-position) to generate AveIP. [ 14 ]…”