Image Quality Optimization Of Micro-computed Tomography

In recent years, with the rapid development of the in vivo small animal imaging techniques, the longitudinal study for a single small animal using small animal imaging systems has become a common approach in biomedical researches. Among kinds of in vivo small animal imaging modalities, Micro-CT is the most ancient, most widely used and most effective imaging means. However, the signal-to-noise ratio of the Micro-CT is relatively low because of the limited number of X-ray photons arriving at each pixel. As a result, the Micro-CT image is rather sensitive to the artifacts and the instability of the X-ray source, the detector and the acquisition geometry. Thus, how to reduce the image artifacts and improve the image quality becomes a very important research topic for Micro-CT.This thesis develops a gantry-rotated Micro-CT system consisting of a micro-focal spot X-ray source and an amorphous silicon based flat-panel detector. In this Micro-CT system, the X-ray source and the detector are fixed at the gantry and the gantry rotates in the vertical plane around the object. The small animal is placed on the animal bed and kept still during the scan. It has been shown that the spatial resolution is about200um and the associated contrast-to-noise ratio illustrate this system can only differentiate the bony tissue, fat tissue and lung tissue clearly. No obvious beam hardening artifact could be observed in the slice of a water phantom after correction.For the first time, this paper proposes and proves the symmetry property of sum of projections (SOP) in cone beam Micro-CT. Then, several numerical phantoms are simulated to study the factors which may affect the symmetry property including the X-ray source, the flat-panel detector, the system geometry and the imaging sample. The simulation results show that factors including random movement of the focal spot, the abnormal pixels (APs) of the detector, the system geometric parameters and the respiratory movement of small animals can seriously affect the symmetry property, while others factors have a negligible effect.Using the relationship between the geometric parameters and the SOP symmetry, an objective function is formulated to illustrate the dependence of SOP symmetry on the main geometric parameters of the skew (in-plane rotation angle of the detector), the roll (rotation angle around the projection of the rotation axis on the detector), the rotation axis, the mid-plane and source-to-detector distance. The objective function will converges to minimum as the geometric parameters approach real. Thus, by minimizing the objective function, we can obtain the geometric parameters for image reconstruction. We next validated this method by simulating numerical phantom scans and small animal scans. The results show that the proposed on-line geometric calibration method can greatly improve the reconstruction image quality and achieved comparable image quality in the reconstructions as some off-line methods.Based on the simulation result that APs from the detector will introduce high singularity in the symmetry image of the SOP, this thesis proposes a new method to detect these APs. This method combines dyadic wavelet transform-based singularity detection to extract the APs. Next, the Laplacian solution (LS) method is employed to restore the APs in each projection image. Compared with the traditional methods, the proposed method can deal with the APs with arbitary intensity and will not filter the useful information of original images. Thus, our method effectively suppresses the ring artefacts and the radiant line artefacts and improves the imaging quality.At last, this paper utilizes the developed Micro-CT system and the above two new methods to image the soft tissue of the small animals. To enhance the contrast resolution, the iodinated lipid emulsion contrast agent is introduced. Firstly, the CT contrast enhancements are plotted as a function of the time after injection for each organ. Based on the plots, the compromised imaging time is chosen and we can obtain the images with good soft tissue contrast. Then, the organs in thorax and abdomen were segmented with different approaches depending on the characteristics of the organs with contrast agent. The results show that bone structures, lung, heart, liver, spleen and muscles were separately segmented well.