In order to achieve the above purpose, the present invention provides a computerized tomography imaging reconstruction method comprising the steps of:
Step 1: Gather information - collecting all the different directions corresponding to a projection of the data by using the object as a center when an image system goes around an object every predetermined angle, and the data are recorded as geometric parameters of the imaging projection system; then the directions of the geometric parameters of the system are created as a set of three-dimensional spatial imaging reconstruction information.
Step 2: Projection - building the imaging virtual data model and simplify its three-dimensional characteristics into a simplified data model, and virtualize a set of data on the area of interest as a forecast data, and finally input the data of interest into the simplified data model to simulate the projection.
Step 3: Comparison - comparing and matching the simulation projection with the collected information at each angle of the projected image, and calculating a difference value.
Step 4: Back-projection - processing iterative calculation of back-projection by the different value.
Step 5: Correction - correcting the original three-dimensional imaging spatial reconstruction information by the difference between the value of the back-projection.
Step 6: Cycle - verifying the difference whether is less than a predetermined difference value to decide to go around the steps or not. If not, then go back to the projection step; and if so, then go for the next set of three-dimensional spatial imaging reconstruction starting with the steps of projection, comparison, backprojection and correction in sequence, until each direction of the three-dimensional spatial reconstruction image completes the cycle, and the data of three-dimensional imaging reconstruction space is smaller than that of the predetermined value in the previous iteration.
Step 7: Image output - outputting the reconstruction image from all directions of the resulting simulated projection data forms a three-dimensional imaging.

透過本法所建立之運算架構，可有效減少傳統三維或多維影像之重建時所建立龐大系統矩陣的方式，使用時僅需二維截平面之陣列資料來建立部分的系統矩陣取代傳統建置三維或多維影像之系統矩陣，透過其球座標之對稱特性減少運算時記憶體用量與儲存之空間量。

目前CT運用疊代式演算法進行影像重建蔚為趨勢，舉凡醫療儀器有名之大廠：奇異公司、飛利浦、西門子皆有開始開發各自的疊代式演算法。多數演算法考量其經濟效益，採用傳統解析式演算法；代數式疊代搭配最佳化技術在近期慢慢開始出現；僅有極少數採用統計疊代模型來進行影像重建。