2959 - Prospective Study Assessing C-Arm Linac High-Performance CBCT for Reliable Target and OAR Contouring

12:30pm - 01:45pm ET

Poster Hall - Exhibit Hall A

Screen: 14

POSTER

Presenter(s)

Anita Nwiloh, MD - University of Maryland Radiation Oncology, Baltimore, MD

J. Cammin¹, J. Chen², R. Zarinshenas³, Y. Khamis⁴, A. A. Olabumuyi⁴, P. Armstrong⁴, A. C. Nwiloh², A. Sawant⁵, and J. K. Molitoris⁴; ¹University of Maryland School of Medicine, Baltimore, MD, ²University of Maryland Medical Center, Baltimore, MD, ³Department of Radiation Oncology, University of Maryland Medical Center, Baltimore, MD, ⁴Department of Radiation Oncology, University of Maryland School of Medicine, Baltimore, MD, ⁵Department of Radiation Oncology, University of Maryland, School of Medicine, Baltimore, MD

Purpose/Objective(s):

This study evaluates if high-performance CBCT on a C-arm linac delivers adequate image quality and imaging volume to support direct target and OAR contouring for treatment planning.

Materials/Methods:

40 cancer patients in a prospective non-interventional trial received radiation therapy on a C-arm linac equipped with a high-performance CBCT system (larger imaging volume and improved pixel-value accuracy compared to conventional CBCT). Treated sites included 8 head/neck, 8 breast, 8 thorax, 6 upper-GI, and 10 pelvis patients. For each patient, PTV size was measured on simulation CT (simCT). The fraction of patients whose PTVs were fully covered on CBCT was assessed. Six radiation oncologists (RO) evaluated subsets of patients, with each CBCT rated by one RO for target (GTV/ CTV) and critical organs-at-risk (cOAR) contouring feasibility using a 4-point Likert scale. A Wilcoxon signed-rank test assessed if contouring difficulty differed from that of simCT. Target-structure similarity between simCT and CBCT was quantified for a subset of 16 patients using Dice similarity coefficient (DSC) and Hausdorff distance (HD). Correlation between Likert scores and structure similarity was evaluated using Kendall’s tau.

Results:

All 40 (100%) PTVs were fully encompassed by the CBCT, but ten (25%) PTVs had <2 cm margin to the superior or inferior CBCT boundary. The largest PTV measured 24 cm in the cranio-caudal direction. Contouring was assessed as similar to or slightly more challenging than simCT for head/neck and breast, and more challenging or only partially feasible for thorax, upper-GI, and pelvis (Table 1). Overall, rating was most consistent with the hypothesis that contouring on CBCT was feasible but more challenging than simCT (p-value < 0.001). Similarity metrices demonstrated a moderate correlation of –0.45 (95%CI = [–0.659, –0.215]), p-value 0.002, between Likert score and the resulting contour similarity to simCT (Table 1).

Conclusion:

The evaluated high-performance CBCT shows promise for enabling CBCT-based (re)planning, potentially reducing the need for repeat simulation CTs for selected treatment sites. Contouring on CBCT was generally feasible for all investigated sites but with some limitations due to respiratory and peristaltic motion artifacts and motion-averaging acquisition compared to simCT.

Table1: Disease site average rating results using a four-point Likert scale and average HD and DSC similarity indices for targets.

	Disease Site
Structure	H/N	BREAST		THORAX		UPPER-GI		PELVIS
Targets (*)	1.4±0.5	2.6±1.1		2.4±0.9		2.5±1.0		1.9±0.9
cOARs (*)	1.3±0.5	1.9±1.1		2.0±0.5		3.2±0.8		2.3±0.9

	(*) Likert scale ratings
Similarity metric	1=similar to contouring on simCT		2=feasible but more challenging		3=partially feasible due to poorer image quality		4=not feasible due to strong artifacts
HD (mm)	15.7 ± 11.2		16.2 ± 12.5		15.5 ± 15.9		31.1 ± 21.9
DSC	0.89 ± 0.05		0.78 ± 0.09		0.80 ± 0.07		0.69 ± 0.12