Evaluation of setup accuracy for NSCLC patients; studying the impact of different types of cone-beam CT matches based on whole thorax, columna vertebralis, and GTV

Figures & data

Table I. Characteristics of the UnTagged and the CBCT image sets.

	UnTagged:	CBCT:
Distance between slices/[mm]	2.0	2.0
Pixel size/[mm]	0.78 – 0.98*	0.88
Image dimensions/[px]	512 × 512	512 × 512
Number of slices	122 – 231	80

*The pixel size was identical in the LR and AP direction (square), ranging between 0.78 × 0.78 mm² to 0.98 × 0.98 mm² for the UnTagged image sets. The pixel size in the CC direction was determined by the slice thickness of 2 mm.

Figure 1. A user-defined 3D rectangular ROI (white box) when performing a WT (A), CV (B) and ST (C) CBCT match. (A) The cranio-caudal (CC) borders of the ROI were limited by the length of the CBCT scan. The left-right (LR), and the anterior-posterior (AP) borders were positioned just outside the thorax volume to assure whole thorax volume inclusion in the ROI. (B) The CC borders of the ROI were limited by the extent of the PTV (blue line). The LR, and the AP borders were positioned just outside the columna vertebralis when using bone window level in Offline Review. The volume within the ROI utilized for the CV match is colored red (i.e. the pixels with an intensity range of 50–3000 HU). (C) The CC borders of the ROI were limited by the extent of the PTV. The LR, and the AP borders were positioned next to the PTV. The volume within the ROI utilized for the ST match is colored red (i.e. the pixels with an intensity range of -150-150 HU) (Color version of figure is available online.)

Figure 2. Translational shifts (VRT, LNG and LAT) between the 3 DOF CV match and the 2D kV planar imaging match on CV.

Figure 3. Rotations (Pitch, Roll and Yaw) for the 6 DOF CV, WT and ST matches. On each box, the central mark is the median, the edges of the box are the 25th and 75th percentiles, the whiskers corresponds to approximately ± 2.7 SD, outliers are plotted as red crosses individually, the notches corresponds to the confidence interval of the median. If the notches do not overlap for the various group data it can be concluded, with 95% confidence, that the true medians differ.

Figure 4. (A) Translational shifts (VRT, LNG and LAT) between the non-IGRT setup and the 6 DOF CV and ST matches. (B) Translational shifts for the 3/6 DOF CV/ST, as well as the WT/ST matches.

Table II. Summary of the calculated CTV to PTV margins.

	CBCT IGRT				2D kV planar IGRT	Non-IGRT
	CV/ST		WT/ST		CV/ST	Tattoo/ST	Tattoo/CV
DOF	3/6	6/6	3/6	6/6	3/6	3/6	3/6
VRT/[cm]	0.4	0.4	0.4	0.3	0.4	0.7	0.9
LNG/[cm]	0.4	0.2	0.6	0.5	0.3	1.0	1.0
LAT/[cm]	0.3	0.2	0.2	0.2	0.3	0.6	0.6

Left column set (CBCT IGRT): The listed margins are based on translational shifts between CV/ST and WT/ST matches for various 3 and 6 DOF combinations. Middle column set (2D kV planar IGRT): The listed margins are based on the translational shifts between the 2D kV planar imaging match on CV in 3 DOF and the 6 DOF CBCT ST match. Right column set (Non-IGRT): The listed margins are based on translational shifts between patient positioning in 3 DOF, using only skin markers (tattoos) and the 6 DOF CBCT ST, as well as the CV matches.