2347 - Comprehensive Evaluation of Two Ring Gantry LINAC Cone-Beam Computed Tomography Systems for Online Adaptive Radiation Therapy

S. C. Lin^1,2, H. C. Kuo³, K. Episcopia¹, G. Tang¹, L. I. Cervino¹, and S. B. Lim¹; ¹Department of Medical Physics, Memorial Sloan Kettering Cancer Center, New York, NY, ²University of Massachusetts Lowell, Lowell, MA, ³Memorial Sloan Kettering Cancer Center, New York, NY

Purpose/Objective(s): This study evaluated the performance of cone-beam computed tomography (CBCT) system on two ring-gantry LINACs for online adaptive radiation therapy. Materials/

Methods: The image quality of fan-beam CT (FBCT), CBCT from a C-arm linac (CBCT_C), and CBCT from two ring-gantry LINACs equipped with a large 86x43cm² imaging panel (CBCT_R) were compared. The comparison was made with the pelvis and thorax imaging protocols. For each protocol, the same field-of-view and a similar volume CT dose index were used to scan an imaging QA phantom. The FBCT and CBCT_R were reconstructed using iterative algorithms, with Boltzmann-based scatter correction considered for CBCT_R, while CBCT_C images were generated with both iterative (CBCT_C,i) and FDK (CBCT_C,FDK) reconstructions. The spatial resolution (MTF_50%), contrast-to-noise ratio (CNR) for polystyrene-LDPE, and noise and non-uniformity in a uniform section were assessed. To compare with the HU calibration curve from FBCT (HU_FBCT), an electron density phantom was scanned on the ring-gantry LINACs to generate HU curves with and without metal artifact reduction, HU_R,I and HU_{R,i_m}, respectively.
Results: The CBCT image quality between the two ring-gantry LINACs were very similar. The MTF_50% of all CBCT images were comparable with FBCT, the difference was within 0.11 cycle/cm for pelvis protocol and 0.75 cycle/cm for the thorax protocol. For CNR, the difference was within 0.4 for CBCT_R,i, while a decrease of CNR by 1.3 and 2.9 was seen in CBCT_C,I and CBCT_C,FDK, respectively, when compared to FBCT. CBCT_R,i, exhibited slightly inferior non-uniformity compared to FBCT by 0.3-0.8%, while CBCT_C,i, and CBCT_C,FDK degraded by 1.2-1.8%. This was consistent with the higher noise observed in CBCT_c,i and CBCT_c,FDK, with 1.33 and 1.87% increases compared to FBCT, respectively, and CBCT_R,i were within 0.4%. In the range of air to liver density of the HU curves, HU_R,i and HU_{R,i_m} showed insignificant differences from HU_FBCT, with a maximum difference of ~90 HU in low density lung tissues. However, more remarkable differences were noted in the higher electron density range. For Inner Bone (~137 HU), differences up to ~107 HU were observed, and up to ~300 HU difference in Cortical Bone (~1200 HU). Beyond Cortical Bone, HU_{R,i_m} was within 576 HU for titanium (6424 HU) and exhibited saturation at 7000 HU with stainless steel.
Conclusion: The CBCT image quality of two ring-gantry LINACs was shown to be similar to FBCT, demonstrating feasibility for treatment planning. However, an independent HU calibration is recommended to minimize systematic dosimetry errors, while care should be taken with high-z material due to potential signal saturation.