Target Volume Delineation Study of Compressed Sensing Cone-Beam CT for Head-And-Neck and Breast Cancer Patients During Radiation Therapy
H Zhang1*, W Tan2, J Sonke1, (1) Netherlands Cancer Institute, Amsterdam, (2) Hubei Cancer Hospital, Wuhan, HubeiTH-C-WAB-6 Thursday 10:30AM - 12:30PM Room: Wabash Ballroom
Cone-beam (CB) computed tomography (CT) is widely used for image guided radiotherapy. Compressed sensing (CS) based CBCT reconstruction techniques are investigated to improve CBCT image quality to facilitate adaptive radiotherapy applications. The purpose of this study was to assess the impact of CS image quality improvements on delineation uncertainties of targets and organs-at-risk in head-and-neck (H&N) and breast cancer patients.
Eight H&N and eight breast patients were selected. Each H&N patient had one planning CT (CTP) and three repeat CTs (CTR), and each beast patient had one CTP and one CTR. The corresponding CBCTs on the same day of CTR were reconstructed by both the conventional filtered back-projection (CBCTCON) and compressed sensing (CBCTCS) methods. All CTR, CBCTCON, and CBCTCS were rigidly registered to their CTP. Then gross tumor volume (GTV), parotid glands, spinal cord, and sternocleidomastoid muscles were manually delineated for H&N cancer patients, and the breast and tumor bed (TB) for breast cancer patients. The delineated volumes were determined, and the impact of image quality differences were quantitatively assessed by dice similarity coefficient (DSC), and the shortest perpendicular distance (SPD) between contours of two images. The paired T-test was used for statistical comparison.
The average volume ratio of CBCTCS/CBCTCON was 0.96, which indicated delineated volumes of CBCTCS were smaller than CBCTCON. The results of DSC and SPD from H&N and breast patients demonstrated that CBCTCS was closer to CTR than CBCTCON, where CTR was considered as the golden standard and they were statistically significant. In the temporal direction of H&N patient, a trend of decreasing DSC in GTV was found to suggest anatomic changes.
This study demonstrated that CS CBCT improves delineation accuracy in H&N and breast cancer patients over conventional reconstruction. The results suggest that CS CBCT has potential for image guided adaptive interventions.
Funding Support, Disclosures, and Conflict of Interest: This research was sponsored by CTMM, project Breast CARE (grant 03O-104).