1269054 玖、發明說明： 【發明所屬之技術領域】 本發明係關輻射性斷層掃描儀（radiative tomographic scanners)。特別是關於一種應用於轄射性 斷層掃描儀之散射修正裝置（scatter correction device) 和方法。此裝置和方法能被應用在發射式掃描儀 (emission type scanners)，例如質子發射斷層（positron emission tomography，PET)掃描儀和單光子發射斷層 (single photon emission tomography，SPECT)掃描儀， 以及穿透式掃描儀（transmission type scanners)，如電腦 斷層掃描儀（computed tomography，CT)。 【先前技術】 由於發射斷層掃描儀具有高解析度和高敏感度等特 性，其遂成為研究人體功能上可供選用的儀器設備。斷 層掃描儀的優點在於其具有在人體内輻射性核種 (radionuclides)之活度分佈（activity distribution)的镇 測能力。量化的精確度依原始資料（raw data)的適當修 正而定。底下，以一 PET系統為例來作說明。 核子醫學領域認為PET的衰減（attenuation)和散射 修正是產生精確量化資料的重要元件。衰減的問題能在 PET中完全被修正，其主要限制是所取得資料的統計誤 1269054 差。在PET中對散射輻射的修正是最具挑戰性的工作 之一0 PET系統利用一對的加瑪（gamma)射線的巧合债測 (coincidence detection)，該加瑪射線是由正子與電子碰 撞造成互毁作用（annihilation interaction)而產生。為了 增加敏感度，三維PET掃描常將準直儀（septa)移除。此 準直儀是用來校準（collimate)以防散射的輻射線。散射 事件（scatter events)顯著地增加。據估計，移除準直儀 裝置後，散射部分（scatter fraction)由14%增至36%。散 射事件引起射源（source)位置的誤判且降低系統的空間 解析度。在真實的分佈中加入背景，導致在重構的輻射 活度濃度（reconstructed radioactivity concentrations)對 比降低和量化被高估的現象。由於使用相當寬廣的能量 視窗來維持精確的計數統計量，利用能量差別（energy discrimination)來摒棄（rejection)散射的光子只能達到 有限的效果。 散射修正不僅對量化，對病灶損害債測（lesion detection)和影像分割（image segmentation)也是重要 的。影像散射的影響一般係依偵測器的能量解析度、能 量視窗的設定、物體的大小、形狀和化學成分，以及射 源分佈等因素而定。這些參數大部分都是非固定的 (non-stationary)，這也就是在研發適當的散射修正技術 1269054 時所存在的潛在的困難。 精確的散射修正是面對量化三維PET的主要問題之 一，且仍是一個未獲解決的問題。很多研究開發都集中 在量化二維PET所需的散射補償上。這些修正方法之 間的差異是散射元件被估算的方式，主要區分為四大 類·以能量視窗（energy window)為基礎的方法、以曲線 擬合（curve fitting)為基礎的方法、以旋積運算 (convolution)為基礎的方法，和以重構（rec〇nstructi〇n) 為基礎的散射補償方法。 多重能量視窗法在SPECT上已使用多年。最近，在 PET取樣模式和偵測器能量解析度的發展使得以能量 譜（energy spectra)為基礎的散射修正可以被實現。此後 方法假設高能量（尖峰）視窗裡含有主要的和散射的事 件，而低能量視窗裡含有大部分的散射事件。總散射事 件和低能量視窗事件之間有一個固定的散射比。此兩組 資料集的線性組合可算出散射事件。然而，使用固定的 散射比值對大體積的物件未必恰當。散射比值也會隨著 材料的不同而變化。因此，此種這方法不適合處理不均 勻的物件。另一個缺點是，有些商業系統並未裝設多重 能量視窗。 1269054 曲線擬合法係基於假設散射空間分佈能以高斯 (Gaussian)函數或是二階多項式來描述。在中心的散射 是由射源物件（source object)的外部區域所债測的資料 作内插。對大體積且不均勻（inhomogeneous)區域之物件 而言，這樣的假設可能會有問題。以類似的技術為基 礎’此方法更在三維（高散射）和二維（低散射）掃描之間 的差異作比較’以估算散射污染（scatter contamination)。此方法需要一個額外的二維掃描。 旋積運算法從標準尖峰值利用常數散射核心 (constant scatter kernels)來估算散射的分佈。此散射核 心可以是單指數函數（1110110-6叮01]^11丨丨&1)或高斯函數。散 射分佈以核心與尖峰投射作遞迴地旋積運算。其缺點為 它沒有考慮來自視野（field of view，FOV)之外的散射， 並且’核心是由假體所量測（measured with phantoms) 可能不適用於人類的解剖（human anatomy)。 以重構為基礎的方法是利用一嚴密的蒙地卡羅 (Monte Carlo)模擬來估算散射元件。散射元件可用蒙地 卡羅模擬直接由穿插式掃描得到之影像來估算。 Ollinger利用Klein-Nishina公式直接計算單一散射分 佈，且與影像作旋積運算這個散射分佈，以估算多重散 射。由於在很多參數上作重複的迴圈，這些方法需要大 量的運算和處理時間。 1269054 在美國專利6,590,213的揭露中，Scott D ·1269054 发明, invention description: [Technical field to which the invention pertains] The present invention relates to radiation tomographic scanners. More particularly, it relates to a scatter correction device and method for use in an accommodative tomography scanner. The apparatus and method can be applied to emission type scanners, such as positron emission tomography (PET) scanners and single photon emission tomography (SPECT) scanners, and through Transmission type scanners, such as computed tomography (CT). [Prior Art] Since the emission tomography scanner has characteristics such as high resolution and high sensitivity, it has become an instrument for selecting human body functions. An advantage of a tomography scanner is its ability to maintain the activity distribution of radioactive nucleus in the human body. The accuracy of the quantification depends on the appropriate correction of the raw data. Below, a PET system is taken as an example for illustration. The nuclear medicine field believes that PET attenuation and scatter correction are important components for producing accurate quantitative data. The problem of attenuation can be completely corrected in PET. The main limitation is the statistical error 1269054 of the data obtained. Correction of scattered radiation in PET is one of the most challenging tasks. The PET system utilizes a pair of coincidence detection of gamma rays caused by collisions between electrons and electrons. Produced by annihilation interaction. To increase sensitivity, 3D PET scans often remove the collimator (septa). This collimator is a radiation that is used to collide to prevent scattering. Scatter events are significantly increased. It has been estimated that the scatter fraction increased from 14% to 36% after removal of the collimator device. A sporadic event causes a false positive of the source location and reduces the spatial resolution of the system. Adding background to the true distribution results in a phenomenon in which the reconstructed radioactivity concentrations are reduced and quantified. Due to the use of a fairly wide energy window to maintain accurate count statistics, the use of energy discrimination to reject scattered photons has only a limited effect. Scattering correction is also important not only for quantification, but also for lesion detection and image segmentation. The effects of image scattering are generally determined by factors such as the energy resolution of the detector, the setting of the energy window, the size, shape and chemical composition of the object, and the distribution of the source. Most of these parameters are non-stationary, which is a potential difficulty in developing the appropriate scatter correction technique 1269054. Accurate scatter correction is one of the main problems facing the quantification of 3D PET and remains an unsolved problem. Much of the research and development has focused on quantifying the scattering compensation required for two-dimensional PET. The difference between these correction methods is the way in which the scattering elements are estimated, mainly divided into four categories: an energy window-based method, a curve fitting-based method, and a convolution operation. (convolution)-based methods, and scattering compensation methods based on reconstruction (rec〇nstructi〇n). The multiple energy window method has been used for many years on SPECT. Recently, the development of PET sampling mode and detector energy resolution has enabled energy spectrum based scatter correction to be achieved. The method then assumes that the high energy (spike) window contains major and scattered events, while the low energy window contains most of the scattering events. There is a fixed scattering ratio between the total scattering event and the low energy window event. A linear combination of the two sets of data sets calculates the scattering event. However, the use of a fixed scattering ratio may not be appropriate for large volumes of objects. The scattering ratio will also vary from material to material. Therefore, this method is not suitable for handling uneven objects. Another disadvantage is that some commercial systems do not have multiple energy windows installed. 1269054 The curve fitting method is based on the assumption that the scattering spatial distribution can be described by a Gaussian function or a second-order polynomial. The scattering at the center is interpolated from data that is measured by the outer region of the source object. Such assumptions can be problematic for objects that are large and inhomogeneous. Based on a similar technique, this method compares the difference between three-dimensional (high scattering) and two-dimensional (low scattering) scans to estimate scatter contamination. This method requires an additional 2D scan. The convolution algorithm uses constant scatter kernels from standard peak peaks to estimate the distribution of the scattering. This scattering kernel can be a single exponential function (1110110-6叮01]^11丨丨&1) or a Gaussian function. The scatter distribution is a recursive operation of the core and spike projections. The disadvantage is that it does not take into account scattering from outside the field of view (FOV), and that the 'core is measured by the phantoms may not be suitable for human anatomy. The reconstruction-based approach uses a rigorous Monte Carlo simulation to estimate the scattering elements. Scattering elements can be estimated directly from the images obtained by interspersed scanning using Monte Carlo simulation. Ollinger uses the Klein-Nishina formula to directly calculate a single scattering distribution and convolutes the scattering distribution with the image to estimate multiple scattering. These methods require a large amount of computation and processing time due to repeated loops on many parameters. 1269054 In the disclosure of U.S. Patent 6,590,213, Scott D.
Wollenweber藉由結合某個範圍内的軸數據（axial daty 成為複合跨軸平面（composite transaxial planes)或超級 切片（super-slices)，改良了以模型為基礎的散射運算法 則的執行速度，因此有效地瓦解沿著軸向的數據。經由 如此的組合轴向數據，而可藉由平面上（in_plane)參數χ 和Υ的迴圈運算（looping)，在每一超級切片内執行模型 為基礎的散射運算法則，而不是在平面上參數和方位角 維度（azimuthal angle dimension)兩者作迴圈運算。由於 消去結合方位角維度迴圈運算的計算，因此降低執行模 型為基礎的散射運算法則的計算時間。 在美國專利5,903，008的揭露中，Jianying Li發明了 一種發射斷層掃描系統，以對有興趣的物件成像。此系 統包含一旋轉臂（gantry)和一病患台（patient table)。一 包含準直器（collimator)的谓測器固定在此旋轉臂上，且 一台電腦耦合（coupled)至旋轉臂和偵測器，用以偵測和 控制偵測器相對於病患台的位置。此系統先量測一個穿 透值（transmission measurement)，並利用此穿透值產生 一散射百分比（scatter fraction)。然後利用此散射百分 比，由雙能量視窗所擷取資料決定出在主要能量視窗裡 的非散射光子。 大部分上述的散射修正方法皆間接、地執行蒙地卡羅 1269054 模擬來計算輻射的散射量，並且假設知道例如身體的結 構和射源活度的分佈等資訊。然而，結果並不真的完全 如此，因此不可能達成精確的散射修正。 【發明内容】 本發明克服上述傳統之斷層掃描儀散射修正的缺 點。其主要目的是提供一種斷層掃描儀散射修正褒置和 方法，可應用在所有使用輻射線成像的斷層掃描儀上。 本發明之散射修正裝置和方法的優點在於，實際地量測 散射量而改善實驗結果的精確性。本發明產生較佳的對 比和降低均方根誤差（mean SqUared err〇r)。 根據本發明，此斷層掃描儀散射修正裝置衰減一部 分的主要射線，而幾乎不會對散射造成影響，主要射線 阻擋掉的百分比（fracti〇n)是由一空氣掃描（ah saM的 修正而得到。在處理來自兩次掃描，含與不含射束擋塊 (beam stopper)的數據（datam，可得到在射束擋塊位置 之散射量。接著，以一内插法對整個正弦圖（sin〇gram) 執行散射分佈的回復。對整個正弦圖，由原始數據中扣 除散射量，遂能得到主要射線的正弦圖（primary sin〇gram)。在影像重構之後，可得到修正過而無散射 污染的影像。 在較佳實施例中，本發明之散射修正裝置被應用在 1269054 CT或SPECT掃描儀上。使用本發明之相同方法，能夠 對每一投射角計算出主要投射數據。沿著物件四周所有 角度掃描完之後，就能得到僅含有主要射線的正弦圖可 用來做影像重構。實驗結果顯示使用主要元件的影像比 傳統方法得到較佳的對比（contrast)。而根據本發明，結 果也顯示散射修正後，重構影像的均方根誤差由 0·38χ106 降至 〇·21χ1〇6。 茲配合下列圖式、實施例之詳細說明及專利申請範 圍’將上述及本創作之其他目的與優點詳述於後。 【實施方式】 圖1說明本發明之散射修正裝置。参考圖1，本發 明之散射修正裝置主要由一個或數個射束擋塊1021、 1022、1023、1024、1025、…、102η 和一個或更多個固 定支撑元件（steady support component)101a 和 101b 所 組成。每一個射束檔塊的厚度在2mm至5mm的範圍 内’且其兩端分別固定在固定支撐元件上。兩端上的固 定支撐元件使得這些射束檔塊能穩定地支撐在固定的 位置上。 射束擋塊用來有效地遮擋住輻射線。由於高元子序 (Z值）的材料具有高阻擋能力，故本發明之射束擋塊可 12 1269054 以是如圖1所示之高z值的細棒條。高z質材料可包括 雜和鶴。根據實驗的結果’ 3mm厚的射束擋塊具有最佳 的阻擋輻射線的能力而又不會妨礙散射線的測量。由於 低能量之光子的因素’將本發明之散射修正袭置用到 SPECT上時，射束擋塊的最理想厚度範圍為2mm至 4mm ° 固疋支擇元件的功月b係用來支樓細的射束擔塊且將 此射束擔塊支撑在固定的位置上。然而，此固定支撑元 件不應干擾到輻射線的行進。就這一點而言，固定支撐-元件通常是由當輻射線通過時較不會引起任何影響的 塑膠所製成。 本發明之散射修正裝置可以有不同的外觀設計，1 外觀設計依待掃描之物件的大小和形狀而變。如果不同 的待掃描之物件或器官具有相似的形狀和大小，則可使 用相同的散射修正裝置。在掃描物件的期間，本發明之 散射修正裝置通常置於物體與偵測器之間。為了降低散 射的輻射線的影響，本發明之散射修正裝置應盡量靠近 待掃描的物件，且覆蓋住整個物件。而當本發明之散射 修正裝置應用在PET掃描儀時則是一個例外，此時本 發明之散射修正裝置並不須覆蓋欲掃描之物件的底部。 13 1269054 圖2說明本發明之散射修正裝置應用在動物pet的 第一較佳實施例。該散射修正裝置盡量靠近物件201( — 隻老鼠）且完全覆蓋該物件201。該散射修正裝置包含 一對塑膠框（plastic frame) 202，其上並插入數隻鉛棒 (lead bar) 203。本發明之散射修正裝置能夠很容易地從 檢驗台（couch) 204移走或置於檢驗台204頂上。 根據本發明，圖3a說明應用射束擋塊到一 CT掃描 儀。圖3a中，射束擋塊301置於物件3 02之前，以阻 擋來自點射源的主要射束。輻射事件為偵側器Dj上的 讀取值。如圖3b所示，可加上更多的射束擋塊以得到 更多的請取值。 圖4說明本發明在SPECT系統上可能的具體實現。 其包含兩部份：上半部401為包含數個射束擋塊4011、 4012、…、401m和一固定支律元件403的半圓柱型鳥 籠（cylindrical birdcage);下半部402為包含數個射束擋 塊4021、4022、…、402K和一固定支撐元件404的一 薄板（sheet)。此薄板能夠從置於物件4〇5底下的盒子中 取出或插入。 圖5說明實施在圖1之裝置上的散射修正方法的流程 圖，說明本發明之散射修正的詳細步驟。此散射修正方 1269054 法包含五個主要步驟：第一步驟為分別以含與不含射束 擂塊來掃描物件，如步驟5011和5012所示，並且分別使 用含與不含射束檔塊的兩個空氣掃描，如步驟5〇13和 5014所示；第二步驟5 〇2為處理從步驟5〇1卜5012、5013 和5014所得到的資料（^、〇11、（^()和(：11();第三步驟503 係對散射正弦圖回復其散射量。正弦圖是儲存掃描儀之 投射數據的一個檔案。第四個步驟504係從原始數據扣 除散射量而得到僅有主要射線的正弦圖。最後的步驟 505是重新建構已修正過散射的原始物件的影像。 為了更詳細的描述步驟5011和5012，圖6和圖7 为別說明知描儀與回應線（line of response，LOR)的幾 何結構，以及為了阻擋真實的（主要的）事件免於被記錄 的射束檔塊的系統幾何結構。 参考圖6，回應線表示連接一對偵測器的直線。 LOR(i，Θ)上的任意點（x，j；)皆滿足方程式+ =，，其 中ί為LOR與中心點的距離，而0為投射角。掃描儀的 投射數據儲存在正弦圖。正弦圖的元素c((，的包含沿著 LOR(i, Θ)分佈來源點所發出的主要射線，以及從物件全 身所發出的散射線。本方法的目的係用來分離主要射線 (primary radiation)和散射線（scattered radiati〇n)。 15 1269054 參考圖7射束標塊插入在横測器的前方，以阻擋真 實的（主要的）事件免於被紀錄。射束檔塊是小的，且被 安置在盡量靠近物件的位置，使得被阻擋的散射事件為 最小。依射束檔塊的材質和厚度而定，主要射線中只百 分比Γ· (Γ<1)能穿越射束檔塊且被紀錄在偵測器上。注 意’ r==exp(-私）其中Α和Ζ分別為線性衰減係數和射束構 塊的厚度。一個射束檔塊能阻擋數個主要射線。但對於 每一投射角0，射束檔塊只能（部分）阻擋由位於，内 之射源所發出的主要射線，其中 尤,cos0 + 巧 sin0 =， (Uy)為射束檔塊的位置。 回到步驟5013和5014，執行含與不含數個射束檔 塊之空氣掃描以計算出主要射線之穿透百分比 (transmission fracti〇n)r。百分比Γ依射束檔塊的材質 和厚度而定。在空氣掃描中，將一點射源在視野中每一 位置停駐一固定時間。假設（:⑽和Crq為含與不含射束 檔塊之空氣掃描的計數值（count)。 在步驟502中’计异數據r、CB和（^以得到更有用的 資訊。因此，LOR(i，的的穿透百分比Γ很簡單地就是此兩 次掃描的商： TV，0) = CBQ(t，0)fCRQ(t，0)。 1269054 注意，r等於1表示此LOR沒有被射束檔塊擋住。 空氣掃描只須操作一次即可。將Γ的資訊儲存起來，以 給同一台裝置作接續的掃描用。 假設S和p表示不含射束檔塊之原始訊號的散射元 件和主要元件，而cR表示不含射束檔塊在正弦圖中 LOR(i，的上的計數值，可表示為下列公式： =户⑽以⑽。 而CB表示含射束檔塊在正弦圖中L〇R(i，^上的計數值， 係散射事件和主要事件之百分比的加總，亦即 <^Β(Μ) = Γα0)χΡ(ί，0) + 5(Μ)。Wollenweber improves the execution speed of model-based scattering algorithms by combining a range of axis data (axial daty into composite transaxial planes or super-slices), thus effectively Dissipating data along the axial direction. Through such combined axial data, model-based scattering operations can be performed in each super slice by means of in-plane parameters χ and Υ looping. The rule, rather than the on-plane parameter and the azimuthal angle dimension, is used as a loop operation. Since the calculation of the azimuth dimension loop operation is eliminated, the computation time of the execution model-based scattering algorithm is reduced. In the disclosure of U.S. Patent No. 5,903,008, Jianying Li invented an emission tomography system for imaging an object of interest. The system includes a gantry and a patient table. A collimator predator is attached to the swivel arm and a computer is coupled to the swivel arm A detector is used to detect and control the position of the detector relative to the patient station. The system first measures a transmission measurement and uses the penetration value to generate a scatter fraction. Then using this scattering percentage, the data extracted from the dual energy window determines the non-scattered photons in the main energy window. Most of the above-mentioned scattering correction methods indirectly perform Monte Carlo 1269054 simulation to calculate the amount of radiation scattering. And it is assumed that information such as the structure of the body and the distribution of the activity of the source is known. However, the result is not completely true, and thus it is impossible to achieve accurate scattering correction. [Invention] The present invention overcomes the above-described conventional tomography scanner Disadvantages of scatter correction. Its main purpose is to provide a tomographic scatter correction device and method that can be applied to all tomographs using radiation imaging. The advantage of the scatter correction device and method of the present invention is that the actual amount Measuring the amount of scattering to improve the accuracy of the experimental results. The present invention produces better contrast and lowers both Square root error (mean SqUared err〇r). According to the present invention, the tomographic scanner scatter correction device attenuates a part of the main ray, and hardly affects the scatter, and the percentage of the main ray blockage is fracti〇n An air scan (obtained by ah saM correction. In processing data from two scans with and without beam stoppers (datam, the amount of scattering at the beam stop position is obtained). Next, the response of the scattering distribution is performed on the entire sinogram by an interpolation method. For the entire sinogram, the amount of scattering is deducted from the original data, and the primary sin〇gram of the main ray is obtained. After the image is reconstructed, images that have been corrected without scattering contamination can be obtained. In a preferred embodiment, the scatter correction device of the present invention is applied to a 1269054 CT or SPECT scanner. Using the same method of the present invention, the main projection data can be calculated for each projection angle. After scanning all angles around the object, you can get a sinogram with only the main rays for image reconstruction. The experimental results show that the image using the main components is better contrasted than the conventional method. According to the present invention, the result also shows that the root mean square error of the reconstructed image is reduced from 0·38χ106 to 〇·21χ1〇6 after the scatter correction. The above and other objects and advantages of the present invention will be described in conjunction with the following drawings, the detailed description of the embodiments and the scope of the patent application. Embodiments Fig. 1 illustrates a scatter correction device of the present invention. Referring to Figure 1, the scatter correction apparatus of the present invention is primarily comprised of one or more beam stops 1021, 1022, 1023, 1024, 1025, ..., 102n and one or more steady support components 101a and 101b. Composed of. Each of the beam stops has a thickness in the range of 2 mm to 5 mm and both ends are fixed to the fixed support members, respectively. The fixed support members on both ends allow these beam stops to be stably supported in a fixed position. The beam stop is used to effectively block the radiation. Since the material of the high order sub-order (Z value) has a high blocking ability, the beam stop of the present invention can be 12 1269054 to be a thin bar of high z value as shown in FIG. High-grain materials can include miscellaneous and cranes. According to the results of the experiment, the 3 mm thick beam stop has the best ability to block radiation without obstructing the measurement of the scattered line. Due to the low-energy photon factor's use of the scatter correction of the present invention on SPECT, the optimum thickness range of the beam stop is 2mm to 4mm °. A thin beam is supported and the beam is supported in a fixed position. However, this fixed support element should not interfere with the travel of the radiation. In this regard, the fixed support member is usually made of plastic that does not cause any influence when the radiation passes. The scatter correction device of the present invention can have different designs, and the design varies depending on the size and shape of the object to be scanned. The same scatter correction device can be used if different objects or organs to be scanned have similar shapes and sizes. The scatter correction device of the present invention is typically placed between the object and the detector during scanning of the object. In order to reduce the influence of the scattered radiation, the scatter correction device of the present invention should be as close as possible to the object to be scanned and cover the entire object. An exception is when the scatter correction device of the present invention is applied to a PET scanner, in which case the scatter correction device of the present invention does not have to cover the bottom of the object to be scanned. 13 1269054 Figure 2 illustrates a first preferred embodiment of the use of the scatter correction device of the present invention in an animal pet. The scatter correction device is as close as possible to the object 201 (the mouse) and completely covers the object 201. The scatter correction device includes a pair of plastic frames 202 on which are inserted a plurality of lead bars 203. The scatter correction device of the present invention can be easily removed from the couch 204 or placed on top of the inspection station 204. In accordance with the present invention, Figure 3a illustrates the application of a beam stop to a CT scanner. In Figure 3a, the beam stop 301 is placed before the object 302 to block the main beam from the spot source. The radiation event is the read value on the side detector Dj. As shown in Figure 3b, more beam stops can be added to get more values. Figure 4 illustrates a possible implementation of the invention on a SPECT system. It comprises two parts: the upper half 401 is a semi-cylindrical birdcage containing a plurality of beam stops 4011, 4012, ..., 401m and a fixed branch element 403; the lower half 402 is a number A beam stop 4021, 4022, ..., 402K and a sheet of fixed support member 404. This sheet can be removed or inserted from a box placed under the object 4〇5. Figure 5 is a flow chart showing the scatter correction method implemented on the apparatus of Figure 1 illustrating the detailed steps of the scatter correction of the present invention. The scatter corrector 1269504 method comprises five main steps: the first step is to scan the object with and without the beam slabs, respectively, as shown in steps 5011 and 5012, and using the containing and non-beam stops, respectively. Two air scans are shown in steps 5〇13 and 5014; the second step 5〇2 is to process the data obtained from steps 5〇1, 5012, 5013 and 5014 (^, 〇11, (^() and ( :11(); The third step 503 is to restore the amount of scattering of the scattered sinogram. The sinogram is a file for storing the projection data of the scanner. The fourth step 504 is to subtract the amount of scattering from the original data to obtain only the main ray. The final step 505 is to reconstruct the image of the original object that has been corrected for scattering. To describe steps 5011 and 5012 in more detail, Figures 6 and 7 illustrate the line and response. The geometry of the LOR), as well as the system geometry of the beam stop to be blocked in order to block the real (primary) event. Referring to Figure 6, the response line represents the line connecting a pair of detectors. LOR(i, Any point (x, j;) on Θ) Satisfy the equation + =, where ί is the distance between the LOR and the center point, and 0 is the projection angle. The projection data of the scanner is stored in the sinogram. The element c of the sine diagram (containing the LOR(i, Θ) The main rays emitted by the distribution source point, and the scattered rays emitted from the whole body of the object. The purpose of this method is to separate the primary radiation and the scattered radiation. 15 1269054 Refer to Figure 7 The target block is inserted in front of the traverse to block the actual (primary) event from being recorded. The beam stop is small and placed as close as possible to the object, minimizing the blocked scatter event Depending on the material and thickness of the beam block, only the percentage Γ· (Γ<1) in the main ray can pass through the beam block and be recorded on the detector. Note 'r==exp(-private) Where Α and Ζ are the linear attenuation coefficient and the thickness of the beam block, respectively. One beam block can block several main rays, but for each projection angle 0, the beam block can only be (partially) blocked by, The main rays emitted by the source inside, Zhongyou, cos0 + 巧 sin0 =, (Uy) is the position of the beam block. Returning to steps 5013 and 5014, performing an air scan with and without several beam stops to calculate the percentage of penetration of the main ray (transmission fracti〇n) r. The percentage depends on the material and thickness of the beam block. In air scanning, a point source is parked at each position in the field of view for a fixed time. Assume (: (10) and Crq are Count value (count) for air scans with and without beam stops. In step 502, 'different data r, CB, and (^ are used to get more useful information. Therefore, the percentage of penetration of LOR (i, is simply the quotient of this two scans: TV, 0) = CBQ (t,0)fCRQ(t,0). 1269054 Note that r equal to 1 means that the LOR is not blocked by the beam stop. The air scan only needs to be operated once. The information of the Γ is stored to give the same device. For subsequent scans, suppose S and p represent the scattering elements and main components of the original signal without the beam stop, and cR represents the count value of the LOR (i, without the beam stop in the sinogram, It can be expressed as the following formula: = household (10) with (10), and CB represents the sum of the count value of the beam block in the sinogram L 〇 R (i, ^, the scattering event and the main event, that is, <^Β(Μ) = Γα0)χΡ(ί,0) + 5(Μ).
Cr扣除CB且計算後，產生被射束檔塊阻擋之lor處的 主要元件，亦即 1，，0) 而在射束檔塊位置上的散射元件可由下列公式得到· 使用數個射束檔塊可得到多重的散射事件的讀取 值。在步驟503中，本發明以内插法回復整個正弦圖之 17 1269054 散射分佈。根據散射的空間分佈是緩慢變化的事實，本 發明以内插方式來回復整個分佈。若斷層掃描儀含有雙 能量視窗，散射的回復將更精確。雙能量視窗法的優點 為實際地量測輻射量。不像上述的討論，其假設可能不 完全正確。若結合雙能量視窗法，下列為其運作的過程。 假設心(⑽和仏(⑽分別表示LOR(i，Θ)在雙能量視窗 技術中高能量視窗和低能量視窗的計數值，則 WL(t^) = g(t90)xS(t90) 其中，介，取0表示部分的散射掉進尖峰能量視窗且大 部分為單一散射事件，而你邓，0為在低能量視窗中的 多重散射事件。定義Fl/h(⑽為％(⑽和％(⑽兩者間的比 值，則 g(t90)xS(t90) P(t^) + /(t^)x5(t^) g(t90)xSm P(t90) + S(t90) + f(t90) x S(t,0) - S(ty0) g(t，0)xS(t，6〇 _ g(t,0)xSF(U0) P(t^) + S(t,0) + 5(1^)(1 - f{\M 1 + (1 - /(t，)) x W(t，0 傳統的雙能量視窗法假設/U，0/g(i，W為一定值且主 18 1269054 要事件是由％(/，外扣除散射事件《⑺+〜…❼仏⑺⑼而 得。利用先前提到的方法，在射束檔塊位置作，的和 的值由％⑽·大略估計而得。現在，假設 = /(U)x叩，印(Ρ…〇+/(，，岣*叩，〇)。對某一特定的投射角 0’我們可利用⑺，來當作校正規，由的w⑽内插 W 0，<9)。在射束播塊的投射執跡上^厂，⑺A的值能利用 先前提到的方法求得。對於f/<i< G⑺.和〇•為相鄰兩射 束檔塊的投射距離），1(/，的和，…㊇之間的比值可以假 設為一線性函數。那麼，就能由下列公式得到其内插值： SF，_ = Wuli(;，0)x JF^e\ +^L.( SF^J^ SF\th0) > trtt \WLin{tj^WLln{thff)j 在此技術之下，在射束檔塊上0的内插值恒等 於使用/5和（P+/S)者。一旦坪，的正弦圖算出後就能得 到主要射線的正弦圖。本發明的方法可視為質化地使用 雙能量視窗資訊和量化地使用量測數據來執行散射内 插。此内插的優點為考量到散射事件的區域變化。其他 優點為重構的影像將含較小的雜訊，因為在計算主要事 件時，使用除法，而不是減法。從射束檔塊量測到的數 據來回歸SF，為_的一個二階多項式，並且從^對每 一投射角來估算SF，都是可能的。 得到散射正弦圖之後，就能得到主要正弦圖，而影 像就可以重構。 1269054 圖8說明沿著物體以等距的角度，半徑為K5mm， 12支鉛製圓柱所形成的射束檔塊裝置。利用一種叫做 Simset的蒙地卡羅軟體來模擬本實驗。利用蒙地卡羅模 擬’可以分開主要射線和散射線，因此而證實本發明。 在本模擬中，圖8說明一橢圓形物件（長轴=17〇3cm， 短軸=12.66cm)包含4個圓形（半徑分別為〇 625、〇 935、 1.25和5cm)具有相同的活度（c〇ncentrati〇n 〇f activity)。橢圓形物件的背景活度為圓形的15%。最大 的圓内部的中央區活度為〇,以模擬冷光點（c〇ld sp〇t) 空氣掃描和物件掃描的總計數分別為4·6χ1〇6和 1·2χ 10。圖9說明此兩個影像橫跨區域的水平透視圖 (profile)。使用主要元件的影像產生較佳的對比（〇 47比 0.82)。結果也顯示出散射修正後重構影像的均方根誤差 由 〇·38χ1〇6 降至 〇 21χ1〇6。 木准以上所述者，僅為本創作之較佳實施例而已， 二不此以此限定本創作實施之範圍。即大凡依本創作申 月專利犯圍所作之均等變化與修飾，皆應仍屬本創作專 利涵蓋之範圍内。 20 1269054 【圖式簡單說明】 圖1說明本發明之散射修正裝置； 圖2為根據本發明之散射修正裝置的第一較佳實施例； 圖3a說明應用本發明之射束擋塊到一 掃插儀· 圖3b說明本發明應用更多的射束擋塊到一 [τ掃描儀; 圖4說明本發明在SpECT系統上可能的具體實現； 圖5說明實施在圖1之裝置上的散射修正方法的流程 圖，說明本發明之散射修正的詳細步驟； 圖6說明掃描儀與回應線的幾何結構； 圖7a說明含一射束擋塊之系統的幾何結構； 圖7b說明射束擋塊阻擋數個主要射線； 圖8說明沿著物件以等距的角度，半徑為1 ，12 支鉛製圓柱所形成的射束擋塊裝置； 圖9說明影像橫跨區域的水平透視圖。 圖號說明： 1021〜102η射束擋塊 101a、101b固定支撐元件 2〇1物件（一隻老鼠） 202 —對塑膠框 203數隻鉛棒 204檢驗台 21 1269054 301射束擋塊 302物件 401上半部 4011〜401m射束擋塊 4021〜402K射束擋塊 405物件 402 下半部 403固定支撐元件 404固定支撐元件 5011含射束擋塊來掃描物件 5012不含射束擋塊來掃描物件 5013含射束擋塊的空氣掃描 5014不含射束擋塊的空氣掃描 502處理資料CB、CR、CB〇和CR〇 503對散射正弦圖回復其散射量 504從原始數據扣除散射量 5 05重新建構已修正過散射的原始物件的影像 22After the C is deducted by CB and calculated, the main components at the lor blocked by the beam stop are generated, that is, 1, 0, and the scattering elements at the position of the beam stop can be obtained by the following formula. • Several beam files are used. The block can get read values for multiple scattering events. In step 503, the present invention recovers the 17 1269054 scattering profile of the entire sinogram by interpolation. According to the fact that the spatial distribution of the scattering is slowly changing, the present invention reverts the entire distribution by interpolation. If the tomograph contains a dual energy window, the scatter back will be more accurate. The advantage of the dual energy window method is to actually measure the amount of radiation. Unlike the discussion above, the assumptions may not be completely correct. If combined with the dual energy window method, the following is the process for its operation. Suppose the heart ((10) and 仏((10) respectively represent the count value of LOR(i,Θ) in the high energy window and low energy window in the dual energy window technology, then WL(t^) = g(t90)xS(t90) , taking 0 means that part of the scattering falls into the peak energy window and most of them are single scattering events, and you Deng, 0 is the multiple scattering event in the low energy window. Define Fl / h ((10) is % ((10) and % ((10) The ratio between the two is g(t90)xS(t90) P(t^) + /(t^)x5(t^) g(t90)xSm P(t90) + S(t90) + f(t90) x S(t,0) - S(ty0) g(t,0)xS(t,6〇_ g(t,0)xSF(U0) P(t^) + S(t,0) + 5( 1^)(1 - f{\M 1 + (1 - /(t,)) x W(t,0 The traditional dual-energy window method assumes /U,0/g(i,W is a certain value and the main 18 1269054 The event is obtained by %(/, externally subtracting the scattering event "(7)+~...❼仏(7)(9). Using the previously mentioned method, the value of the sum at the beam stop position is roughly estimated by %(10)· Now, suppose = /(U)x叩, India (Ρ...〇+/(,,岣*叩,〇). For a particular projection angle 0' we can use (7) as a correction gauge. By w(10) interpolating W 0, <9). The projection of the projection on the factory, (7) A value can be obtained by the method mentioned above. For f / < i < G (7). And 〇 • is the projection distance of the adjacent two beam stops), 1 (/, The ratio between the sum of ... and octave can be assumed to be a linear function. Then, the interpolated value can be obtained by the following formula: SF, _ = Wuli(;, 0)x JF^e\ +^L.( SF^ J^ SF\th0) > trtt \WLin{tj^WLln{thff)j Under this technique, the interpolated value of 0 on the beam block is always equal to the use of /5 and (P+/S). The sinogram of the sinogram can be calculated to obtain the sinogram of the main ray. The method of the present invention can be regarded as qualitatively using the dual energy window information and quantitatively using the measurement data to perform the scatter interpolation. The advantage of this interpolation is to consider The change in the area of the scattering event. The other advantage is that the reconstructed image will contain less noise because the division is used instead of subtraction when calculating the main event. The data measured from the beam block is returned to SF. A second-order polynomial of _, and estimating SF from each projection angle is possible. After obtaining the scattered sinogram, you can get To sinogram, the image can be reconstructed. 1269054 Figure 8 illustrates the beam stop device formed by 12 lead cylinders at an equidistant angle of the object, with a radius of K5mm. Using a Monte Carlo called Simset Luo software to simulate this experiment. The present invention can be confirmed by using Monte Carlo simulation to separate the main rays and the scattered rays. In this simulation, Figure 8 illustrates an elliptical object (long axis = 17 〇 3 cm, short axis = 12.66 cm) containing four circles (radius 〇 625, 〇 935, 1.25, and 5 cm, respectively) having the same activity. (c〇ncentrati〇n 〇f activity). The background activity of an elliptical object is 15% of a circle. The central area of the largest circle has an activity of 〇, and the total count of air scanning and object scanning is 4·6χ1〇6 and 1·2χ 10, respectively, to simulate the cold spot (c〇ld sp〇t). Figure 9 illustrates a horizontal profile of the two images across the area. Using the images of the main components produces a better contrast (〇 47 vs. 0.82). The results also show that the root mean square error of the reconstructed image after scatter correction is reduced from 〇·38χ1〇6 to 〇21χ1〇6. The above is only the preferred embodiment of the present invention, and the scope of the present invention is not limited thereto. That is to say, the equal changes and modifications made by the patents in accordance with the creation of this patent should still fall within the scope of this patent. 20 1269054 BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 illustrates a scatter correction apparatus of the present invention; FIG. 2 is a first preferred embodiment of a scatter correction apparatus according to the present invention; FIG. 3a illustrates a beam stop to a sweep insert to which the present invention is applied. Figure 3b illustrates the application of more beam stops to a [τ scanner; Figure 4 illustrates a possible implementation of the invention on a SpECT system; Figure 5 illustrates a scattering correction method implemented on the device of Figure 1. Flowchart illustrating the detailed steps of the scatter correction of the present invention; Figure 6 illustrates the geometry of the scanner and response line; Figure 7a illustrates the geometry of the system with a beam stop; Figure 7b illustrates the number of beam stop blocks The main ray; Figure 8 illustrates the beam stop device formed by a radius of 1, 12 lead cylinders at equidistant angles; Figure 9 illustrates a horizontal perspective view of the image spanning area. Figure number description: 1021~102η beam stop 101a, 101b fixed support element 2〇1 object (a mouse) 202 - plastic frame 203 number of lead bar 204 inspection table 21 1269054 301 beam stop 302 object 401 Half 4011~401m beam stop 4021~402K beam stop 405 object 402 lower half 403 fixed support element 404 fixed support element 5011 with beam stop to scan object 5012 without beam stop to scan object 5013 Air Scan 5014 with Beam Stops Air Scan 502 without Beam Stops Processing Data CB, CR, CB〇, and CR〇 503 Respond to the Scattering Sinusoidal Rebound 504 Reconstructed from Raw Data Deducted Scattering Volume 5 05 Image of the original object that has been corrected for scattering 22