2242 - Improving Liver SBRT Target Localization with Hybrid MRI/CBCT Technology

M. S. Chen¹, Z. Zhang¹, K. Lu¹, H. Zhong², Z. Jiang³, F. F. Yin¹, and L. Ren⁴; ¹Department of Radiation Oncology, Duke University Medical Center, Durham, NC, ²Department of Radiation Oncology, Medical College of Wisconsin, Milwaukee, WI, ³Duke University, Durham, NC, ⁴Department of Radiation Oncology, University of Maryland School of Medicine, Baltimore, MD

Purpose/Objective(s): Liver SBRT is a powerful modality for the treatment of hepatocellular carcinoma and other malignancies of the liver. Yet, the rate of local recurrence can be as high as 20-30%. Improved accuracy of tumor localization may improve the rate of local control of liver tumors. Cone beam CT (CBCT) is the most commonly used imaging modality for liver SBRT, but is limited in pre-treatment target localization due to its poor soft tissue contrast of liver tumor. We hypothesize that hybrid virtual-MRI/CBCT (hMRI/CBCT) imaging will improve pre-treatment target localization. Here, we evaluate a method to improve liver tumor localization accuracy using hMRI/CBCT. Ultimately, we aim to demonstrate the utility of virtual multi-modality imaging to improve outcomes for patients with liver tumors. Materials/

Methods: To improve the soft-tissue contrast of CBCT liver, we developed a method to generate onboard virtual MRI using CBCT, prior MRI, and finite element method (FEM) for deformable mapping. Patients data that contain pretreatment MRI and onboard CBCT were obtained from the clinic planning system. Next, livers were segmented from pre-treatment MRI and on-board CBCT images. MRI and CBCT livers were converted to 3D meshes and imported to the hypermesh software, which aligned the MRI liver surface to CBCT liver surface. MR surface displacements were calculated and served as the input of the Ansys FEM software. The internal deformation vector fields (DVFs) of each MRI liver were calculated based on FEM with the liver elasticity parameter. With the above DVFs, the prior liver MR images were deformed to map to the CBCT geometry to generate onboard virtual liver MRI, which was then combined with CBCT to generate hMRI/CBCT. To evaluate hMRI/CBCT images, we performed two rigid registrations: 1) hMRI/CBCT to planning CT, 2) onboard CBCT to planning CT. We identified landmarks in hMRI/CBCT, onboard CBCT, and planning CT image sets. The errors were calculated based on the landmark differences of two CBCT images and ground truth planning CT. A total of 48 cases (18 simulated and 30 real patients) were evaluated in this study.
Results: Hybrid MRI/CBCT demonstrated excellent soft-tissue contrast with clear tumor delineation. Wherever possible, tumor was used as landmarks. Onboard CBCT images showed a mean difference of 2.64 + 0.96 mm between landmarks in simulated cases. hMRI/CBCT reduced the onboard CBCT localization error from 4.01 ­+ 3.2 mm to 2.49 + 1.63 mm for tumors away from the liver boundary. In real patient cases, the registered shifts between hMRI/CBCT and onboard CBCT had a mean difference of 5.16 + 2.61 mm.
Conclusion: Hybrid-virtual MRI/CBCT resulted in a reduction of target localization error compared to CBCT alone for tumors away from the liver boundary in CBCT-guided liver SBRT. The results demonstrate the feasibility and potential utility of on-board hMRI/CBCT for improved target localization for liver SBRT potentially leading to margin reduction and reduced treatment toxicity.