Abstract
1936
Objectives 123I-FP-CIT is used for the diagnosis of Parkinson's disease (PD) and Dementia with Lewy Bodies (DLB). It can visualize the distribution by binding to the dopamine neurons. In negative cases, the dopamine neurons are accumulated in form of comma, and in positive cases, the dopamine neurons are accumulated in form of dot. SPECT images can provide physiological information of a brain, but it is not possible to read anatomical information such as shape of corpus striatum. Therefore, it is possible to improve of diagnosis by presenting fusion image of SPECT image and MR image to readers as shown on Fig. 1.
Methods Since image information represented by SPECT and MR images are different each other, an accurate image fusion of the two images is not easy. In this study, we propose a method fusion of SPECT image and MR image by intervening CT image taken by SPECT/CT devices (Fig. 2). Once the image fusion between CT and MR images was performed, the locations of MR images were transferred on SPECT images. SPECT/CT devices are general equipments integrated with gamma camera and multi-slice CT. Mostly, the CT images on SPECT/CT device were obtained with very low dose and were used for the absorption correction to generate SPECT images with high image quality. The CT images were utilized for the image registration with MR images. The location of CT and SPECT images were mechanically registered because the images were obtained at the almost same time. For the image registration between CT and MR images, we used a mutual information (MI) between the two images. The MI is the amount that represents a measure of the mutual dependence of two random variables. Before the MI was calculated, spatial resolutions of CT and MR images were unified, and the resolution of CT images were changed from 0.4297 mm to 2.9454 mm. After the resolutions were unified, rough alignments and angles of rotation were determined by calculating maximum mutual information. It is possible to limit search ranges and processing times. The maximum MI was obtained with fine steps of locations and angles to determine the final parameter of the location and the angles. Parallel computing method using CUDA was applied to obtain the parameter.
Results IRB approved 30 clinical cases of SPECT/CT and MR examinations were used in this study. 25 of 30 cases were registered correctly with registration errors up to 2mm. The other 5 cases were registered with registration errors from 7 to 16mm. Figure 3 shows an example for the image registration. The registration errors of 2mm are acceptable mis-registrations on SPECT images because the image resolution of SPECT is almost 3mm. The error can be ignored as mis-registration within a pixel size. The cause of errors in the other 5 cases of clinical results are in the configuration of initial ranges of deformation parameters during the global registration because the patient positioning at the MR examinations were quite different from SPECT/CT imaging. Figure 4 shows the distribution of registration errors.
Conclusions Image registration between brain SPECT and MR images can be performed by calculating MI between low dose CT images taken by SPECT/CT devices and the same patient’s MR images. Based on our registration scheme, the shape information determined on MR images could be superimposed on SPECT images to obtain quantitative counts.
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.