Research On Image Registration Between Stand-Alone Micro-PET And Micro-CT

In recent years, with the development of computer technology and the needs for medical research, more and more medical imaging modalities appear, such as X-ray, CT (computer tomography), MRI (magnetic resonance imaging), SPECT (single photon emission computed tomography), PET (positron emission tomography), ultrasound, fluorescence molecular imaging and so on. Medical imaging modalities tend to differ from each other in their focus. If the images from different modalities can be merged together, more information will be presented. Due to the high cost of multi-modality imaging equipment and the existing equipment cannot meet the needs of most multimodal image fusion, the general acquisition pattern of registered images becomes a research hotspot.In this paper, taking the registration method between micro-PET and micro-CT as an example, we try to study the general registration method for stand-alone imaging modalities. To achieve this objective, we improve the existing registration scheme and design a small animal chamber with an optimized registration device fixed outside, so that the chamber with small animal can be transferred between different imaging modalities. Since the animal is anesthetized during the scanning and transferring, the animal together with the animal chamber can be assumed as a rigid body. The registration device mentioned above appears as feature points in the transaxial images. The space transformation relationship between PET and CT images will be built based on those matched feature points, thus the PET images can be mapped into CT space. Then the image registration and fusion will be done.In order to guarantee a certain accuracy, the scheme above needs to increase the number of feature points, which leads to higher complexity of the experiments and calculation. A further research based on the scheme above was performed, which proposes a new registration method of establishing a reference coordinate system in two image sets. In this scheme, two cross-straight catheters filled with hybrid contrast agent were fixed outside the animal chamber, which appeared as feature points in the transaxial images of PET and CT images. A straight line fitting method was adopted to find those two catheters, which can help reduce the errors introduced in the estimation of the feature points’ location. To obtain the transformation matrices between micro-CT and micro-PET, those two catheters were used to build the reference coordinate system. Thus the image registration and fusion can be realized subsequently.In this paper, two kinds of schemes of the registration results are assessed from the simulation data, phantom data and in-vivo experiments. Through visual inspection and feature points matching error calculation, the results demonstrate that the two schemes have satisfactory performance on image registration between micro PET and micro CT. Compared with the existing methods, the new scheme has the advantage of higher registration accuracy and computing efficiency.