Limited-field-of-view MR/SPECT for small animal imaging and development of a Keyhole SPECT image reconstruction method using a priori information.

摘要

The main objectives of this dissertation are: (1) to find the optimal geometrical configuration parameters for a high-resolution and high-sensitivity limited-field-of-view (LFOV) magnetic resonance (MR) / single photon emission computer tomography (SPECT) for simultaneous small animal imaging with detector limits, and (2) to develop a new image reconstruction method for LFOV using a priori information from MR images.;To improve the sensitivity without resolution loss, the number of pinholes, orientation of the pinholes, acceptance angles, and magnification factors were simulated. 7-pinhole and 9-pinhole collimators were simulated to verify the sensitivity improvement obtained by using a multi-pinhole SPECT system. The pattern of the pinholes and acceptance angles were designed to avoid multiplexing while maximizing detector utilization. The results from simulation show that the 9-pinhole collimator SPECT provided higher sensitivity without multiplexing. In addition, SPECT with different magnifications were simulated with a resolution phantom of 1.8 mm diameter for hot regions to evaluate resolution improvement from magnification. The simulation results show the resolution improvement with a magnification factor of 1.6 for the given detector size. A prototype of a 9-pinhole collimator was built and images were acquired to verify the pinhole configuration.;To reduce the artifacts caused by radioactivity from outside the LFOV, a new LFOV image reconstruction method was developed and named Keyhole SPECT or K-SPECT. The method was validated through simulation with various types of numerical phantoms. In the simulation, the boundary information of the region of interest (ROI) obtained from high resolution MRI and the functional information within the ROI obtained from the SPECT image reconstructed without a priori information were used to generate an adaptive system matrix. This matrix consisted of probability weightings for the ROI within the LFOV. The maximum likelihood expectation maximization (ML-EM) algorithm, a popular iterative image reconstruction method, was used with the adaptive system matrix for image reconstruction. Numerical phantoms with different target-to-background (T/B) ratios were used to confirm the behavior of the Keyhole SPECT method with different contrasts. Measurements of the root-mean-square-error (RMSE) were used to validate the K-SPECT image reconstruction method. The simulation results show that K-SPECT reduces the RMSE for all cases and works better on high T/B ratios.;The newly designed 9-pinhole LFOV SPECT system will have high sensitivity and spatial resolution that facilitates biomedicine research in drug delivery, early detection of disease, and tracking cell dynamics and pharmaceuticals. It could be upscaled using appropriate collimators, magnifications, and acceptance angles for human imaging. The innovative Keyhole SPECT image reconstruction method directly uses a priori information from simultaneously obtained anatomical MR images. It differs from other reconstruction methods where use a priori information is limited to post-image processing or attenuation correction only after image reconstruction without a priori information.