52. Image registration for focal liver reaction evaluation

Abstract

Purpose Patient who underwent stereotactic body radiation therapy (SBRT) for liver cancer show a signal intensity alteration in follow-up imaging known as focal liver reaction (FLR). Accurate image registration between planning and follow-up imaging is crucial for estimating relationship between planned dose and radiation-induced FLR. This study aims to compare three different strategies of imaging for follow-up studies. Methods 22 patients (13 metastasis, 9 hepatocellular carcinomas [hcc]) underwent SBRT with an abdominal compressor. After therapy patients had follow-up studies at 3, 6, 9, 12, 18, 24 and 30 months. Imaging was performed using either CT (12 cases), MRI with a water-only Dixon sequence (34 cases) or MRI with a T1-weighted, fat-suppressed sequence (19 cases). For each follow-up the liver was contoured and the planning images were registered to the follow up using deformable registration. The registration quality was evaluated by computing the conformal index between the liver in the two image sets. Results the figure shows the computed conformal indices as a function if the follow-up imaging type. The overall average value is 90% (range 79–94%), and CT images have less variability compared to MRI’s. An ANOVA analysis does not show any statistically significant difference between the three imaging modalities (p = 0.06). Conclusions The average registration quality is not optimal, with any image type used. The main reason is the amount of distortion caused by the abdominal compressor used for immobilization in treatment planning and delivery. Beside this issue, the three different methods of follow-up imaging can be registered to the planning CT with similar results. Therefore, the imaging modality that allows better delineation of FLR can be safely chosen for follow-ups. Moreover, MRI can be chosen over CT to avoid unjustified use of ionizing radiation.