足現代的正子發射斷層掃描儀，為提高計數率、減少病人劑量，所以在臨床掃描時，會縮回準直儀，以三維模式收集搞合事件。但是三維空間影像重建法較為耗時，不利於臨床上使用，另一種選擇是應用重分裝演算法，將三維空間的資料事先分裝進二維空間的資料堆疊，再利用二維空間的影像重建法處理二維空間的資料堆壘，以節省時間，使之能夠應用於臨床上。常用的重分裝法包括單一切面重分裝法、多切面重分裝法，以及f專利葉重分裝法。本文提出一新型正交投影多切面重分裝法，其目的在於改善多切面重分裝法中z軸模糊的兒象。新方法乃藉由單一切面重分裝法的資料，以正交投影法評估射源落在z軸的位置，再將三維資料以加權方式重分裝到二維的資料堆疊，以避免產生模糊的現象。利用華盛頓大學所研發之SimSET軟體，模擬三維空間中不同位置的點射源與線射源，比較比四種重分裝方法。由結果可證明正交投影多切面重分裝法，對於z軸的解析度具有良好的分辨能力。

In order to increase counting rate and reduce patients' dose， modern PET scanners retract septa during scanning and collect coincidences in 3D mode. However， the traditional direct 3D reconstruction algorithm is time-consuming. An alternative is to use rebinning algorithms that re-arrange the original 3D sinograms to a stack of 2D parallel sinograms. Then the rebinned sinograms are reconstructed by 2D image reconstruction methods. Several rebinning algorithms have been previously suggested such as single slice rebinning (SSRB)， multi-slice rebinning (MSRB) and Fourier rebinning (FORE). In this work， we propose a novel multi-slice rebinning using orthogonal projection (OrthoMSRB) to prevent the blurring along the z direction that is commonly observed in MSRB. By constructing a template sinogram using SSRB， we can roughly estimate the source distribution between slices by using orthogonal projection estimation and hence compute non-uniform weighting for each slice during the rebinning. A series of computersimulated studies of 3D point sources and 3D line sources were conducted to evaluate the performance of four different rebinning schemes. As a result， the proposed Ortho-MSRB results in the best performance in terms of the resolution