A novel semi-automatic method for accurate registration of FDG_CT and FLT_CT image modalities

摘要

The widely used 18F-fluorodeoxyglucose (18F-FDG) radiotracer serves as a good imaging tool in Computed Tomography (CT) and is essential for cancer diagnosis as FDG uptake is higher in cancerous lesions and lower in benign ones, while 18F-fluorothymidine (18F-FLT) uptake is shown to be closely correlated with cellular proliferation. The registration of both of these tracers, FDG and FLT, complement one another to increase both sensitivity and specificity for imaging cancer. Accurate registration of both imaging modalities is thus sought in this study to optimize the results of the diagnosis. This paper describes a novel feature-based registration method which employs affine transformation and linear interpolation for FDG_CT and FLT_CT image modalities. Instead of using one set of affine transformation parameters, three slices have been selected to calculate all of the transformation parameters for CT image registration by linear interpolation. This method has the merits of: a) improving the 3D registration results for CT images; b) avoiding arbitrary selection of that one slice for calculating the transformation parameters for registration; and c) being cost-effective in both its implementation and computational requirements. Experimental results show that effective registration of these two imaging modalities could prove very useful for cancer diagnosis, including improved means for quantization and visualization.