| 摘要: | 本研究採用單光子放射顯微鏡系統（Single Photon Emission Microscope, SPEM）作為投影影像擷取裝置。單光子放射顯微鏡系統為單光子放射電腦斷層掃描系統（Single Photon Emission Computed Tomography, SPECT）的一種高空間解析度應用分支，通用於小動物的影像研究。其影像擷取設備包含電子增益電荷耦合元件（Electron Multiplying Charge-Coupled Device, EMCCD）、高質量光影像縮倍管、摻鉈碘化銫晶體[CsI(Tl)]及多針孔式準直儀。
在電腦斷層掃描系統中，為達到高品質的重建影像，必須建立具有高解析度與精確度的影像系統矩陣。該矩陣將三維物空間中每一體素點透過成像系統計算出在偵測器上之投影點。矩陣內含有每一格點各針孔不同角度之參數，雖然持續增加針孔數能夠提高系統靈敏度，但也不是不斷地增加針孔就能夠優化系統，過多的針孔容易引起多工效應的問題，進而導致解析度的降低，本研究將探討在不同針孔數的影像系統矩陣在影像重建中多工效應的影響。
研究以各種參數分析針孔數對影像品質之影響，首先採用取樣完整性(Sampling Completeness)係數，評估螺旋軌道取樣對於重建物體大小的限制，以及在已建立的不同針孔數影像系統矩陣的假體取樣情形，比較圓形軌道取樣和雙螺旋軌道取樣的差異，藉此提升取樣完整性，進而改善重建影像的品質。此外，由於取樣軌跡不同會影響頻域響應分佈，可能導致空間解析度變化與產生混疊效應(Aliasing)，因此本研究進一步使用Fourier Crosstalk Matrix分析影像系統的頻域行為。透過計算點響應函數(Point Response Function, PRF)與調制傳遞函數(Modulation Transfer Function, MTF)，以取得影像系統矩陣的混疊和空間解析度。
最後，透過假體影像重建實驗，分析各針孔數與軌跡之重建差異。重建過程中皆採用序列子集期望值最大化演算法（Ordered Subset Expectation Maximization, OSEM）作為重建的迭代演算法，使用解析度假體與Defrise假體作為模擬與評估之物體。最終，將單、四和七針孔系統在不同軌跡設計下的重建結果，分析在不同多工比例系統的重建品質，作為SPEM系統優化之理論依據。
;In this study, a Single Photon Emission Microscope (SPEM) system
was employed as the projection imaging acquisition device. SPEM is a
high spatial resolution version of Single Photon Emission Computed
Tomography (SPECT), commonly used in small animal imaging studies.
The imaging system consists of an Electron Multiplying Charge-Coupled
Device (EMCCD), a high-quality optical demagnifier, a thallium-doped
cesium iodide [CsI(Tl)] scintillator, and a multi-pinhole collimator.
In computed tomography systems, achieving high-quality
reconstructed images requires the construction of an image system matrix
with high resolution and accuracy. This matrix maps each voxel in the 3D
object space to its corresponding projection image on the detector through
the imaging system. It incorporates parameters from various pinhole angles
at each grid point. While increasing the number of pinholes can enhance
system sensitivity, doing so indiscriminately does not necessarily optimize
performance. An excessive number of pinholes may introduce
multiplexing effects, which can degrade spatial resolution. This study
investigates how the number of pinholes in the system matrix influences
multiplexing effects during image reconstruction.
To evaluate the impact of different pinhole numbers on image quality,
the Sampling Completeness Coefficient (SCC) is first used to assess the
limitations of helical sampling trajectories on object size. The sampling
conditions of phantoms using system matrices with different pinhole
configurations are analyzed, comparing circular and double-helical
trajectories to improve sampling completeness and ultimately enhance
image quality. Furthermore, since sampling trajectories affect the spatial
frequency response—potentially altering spatial resolution and inducing
aliasing—the study employs the Fourier Crosstalk Matrix (FCM) to
analyze the frequency-domain behavior of the imaging system. By
calculating the Point Response Function (PRF) and the Modulation
Transfer Function (MTF), both aliasing and spatial resolution
characteristics of the image system matrix can be determined.
Finally, phantom reconstruction experiments are conducted to analyze
the reconstruction performance under different pinhole configurations and
sampling trajectories. The Ordered Subset Expectation Maximization
(OSEM) algorithm is used throughout the reconstruction process.
Resolution and Defrise phantoms are used for simulation and evaluation.
Finally, the reconstructed results of single-, four-, and seven-pinhole
systems under different trajectory designs are analyzed to evaluate image
quality across varying degrees of multiplexing, serving as a theoretical
basis for optimizing the SPEM system. |
