Development of Non-invasive Imaging Systems and a 3D Automatic Registration Method for Molecular Imaging Applications

• Main Authors: Meei-Ling Jan, 詹美齡
• Other Authors: Keh-Shih Chuang
• Format: Others
• Language: en_US
• Published: 2005
• Online Access: http://ndltd.ncl.edu.tw/handle/24754402621627768932

博士 === 國立清華大學 === 原子科學系 === 93 === Molecular imaging is a rapidly emerging biomedical research field, in which the images produced the noninvasive visualization in space and time of normal as well as abnormal cellular processes at a molecular or genetic level of function. Advancements arising from this research can enhance our knowledge of disease, lead to earlier disease detection and accelerate drug discovery. At the imaging system level, many challenges, such as to improve imaging performance and to integrate molecular imaging studies, remain in molecular imaging. In this dissertation, we have investigated both software and hardware techniques to develop non-invasive imaging systems for molecular imaging applications. We have successfully constructed a positron emission imager－PEImager in the Institute of Nuclear Energy Research (INER) for imaging small animals and plants. The PEImager can be used to monitor living objects with planar and tomographic imaging modes. For planar projection imaging acquisition mode, the average spatial resolution is 3.11 mm FWHM (full width of half maximum). There are no significant resolution differences through center to peripheral region of the image. The planar imaging can be applied to dynamic studies of small animals or plants. For collecting data using the tomographic imaging mode, the scanner can be rotated to acquire sufficient views to reconstruct a full volume images. The spatial resolution is 2.73 mm FWHM at the axis of rotation and 3.48 mm FWHM at 20 mm away from the center in tomographic imaging mode. The tomographic imaging can be used in studies, which require more spatial information in images. A planar tomography method was developed also to reconstruct projection data to obtain tomographic images without rotating detectors. The resolutions in the X or Y directions do not differ significantly between planar tomography and conventional tomography. However, The mean FWHM of planar tomography in the Z direction was 2.63 times that of conventional tomography. Locations and sizes of hot spots in a breast phantom can be estimated by using this planar tomography with consideration of efficiency of detector elements, attenuation and geometric factors. Compared with conventional focal-plane reconstruction, the images using planar tomography have better quality of contrast, and have greater tumor-detection capability. In order to integrate multi-modality molecular imaging studies, a software approach to register animal images among microPET, microCT and microSPECT modalities was developed. The 3D registration method is automatic, simple, and object independent. It does not depend on any preparation and post-processing of data sets and requires no user intervention. The 3D registration method is successfully applied to microPET/CT, microPET/CT/SPECT fusion for phantom and animal studies. The maximum registration errors of this method were one voxel for microPET/CT fusion and microPET/SPECT fusion. Besides the software approach for image fusion, a hardware approach is processing by building integrated microPET/CT scanner in the INER. An in-house microCT was constructed to combine with a microPET R4 for this purpose. The availability of microPET/CT scanner, which provide inherent, co-registered CT images as anatomic reference for PET images, is expected to benefit both qualitative and quantitative molecular imaging studies, and to play a key role in drug discovery, development, and delivery at the pre-clinical level.