JPWO2004024002A1 - Spiral CT system - Google Patents

Abstract

It was clarified that the tomographic image reconstructed by the weighted spiral correction reconstruction method is distorted with respect to the plane orthogonal to the orbital axis, which is the cause of the steer step artifact. In the present invention, the distorted tomographic image reconstructed by the weighted spiral correction reconstruction method is segmented to determine the local plane of each part with respect to the orbital axis, and the target tomographic image is obtained by interpolation using this, and is accurately stored in the Euclidean space. By rearranging them, the steer step artifact generated when the tomographic image is converted into volume data is reduced.

Description

  The present invention relates to a helical scanning tomography apparatus (spiral CT apparatus), and more particularly to a volume image obtained by stacking tomographic images generated by using a weighted helical correction reconstruction method from projection data detected by a radiation detector during helical scanning. The present invention relates to a technique for correcting a steer step artifact.

A helical scanning tomography apparatus (hereinafter referred to as a multi-row detector tomography apparatus) having a radiation detector composed of a plurality of rows is obtained when creating a tomographic image from a projection data array obtained by spiral scanning. A weighted spiral correction is applied to each projection data array by applying a weight function consisting of a channel and a view element for each column, and a tomographic image perpendicular to the rotation axis is created by reconstructing the obtained spiral correction data. . However, in such a multi-row detector tomography apparatus, a staircase artifact called a steer step artifact may occur strongly in the body axis direction on the reconstructed tomographic image.
According to the technique disclosed in US Pat. No. 5,802,134 (corresponding to WO 98/44847 and JP 2000-515411), in order to remove cone beam artifacts and spiral artifacts, A plane having an inclination is defined with respect to an orthogonal plane (hereinafter referred to as an axial plane), projection data corresponding to this is extracted, and a plurality of inclined tomographic images are intentionally created by reconstructing the projection data. . Thereafter, a tomographic image (non-tilted image) parallel to the axial plane is generated by interpolation based on the tilted reconstructed tomographic image. However, the throughput cannot be increased by creating an inclined tomographic image in this way. The throughput indicates the table feed amount per scanner rotation. When applied to an apparatus using a 2-4 slice multi-detector, an image apparatus that controls reconstruction must be modified in hardware.

An object of the present invention is to provide a helical CT apparatus capable of correcting a spatial distortion of a tomographic image reconstructed by a weighted helical correction reconstruction method and obtaining a good image in which a steer step artifact is corrected. And
The present inventor has found that the weighted spiral correction reconstruction method is an algorithm ignoring the angle (cone angle) of radiation having a spread in the direction of the rotation axis, so that the non-uniformity of the effective slice thickness obtained by imaging or the image It was clarified that this was the cause of the steer step artifact.
In order to achieve the above object, the present invention provides a radiation source that emits conical or pyramidal radiation having a three-dimensional extension, and is disposed opposite to the radiation source via an object, and transmits the object. A scanner body having a two-dimensional radiation detector for detecting the radiation to be emitted, and the radiation source and the radiation detector of the scanner body are simultaneously rotated about a rotation axis, and the scanner body and the object are mounted on the scanner body. And a two-dimensional projection data detected by the radiation detector during helical scanning of the object. In a helical CT apparatus having a tomographic image reconstruction means for reconstructing a strain image for each slice in the rotation axis direction using a weighted spiral correction reconstruction method, Differentiating, obtaining a local plane for each part, and data corresponding to the plane coordinate position of the pixel from a plurality of local planes sandwiching a pixel on a predetermined target tomographic image in the circumferential axis direction for each pixel It is characterized by having an artifact correcting means configured to interpolate on a tomographic image and obtain a volume image thereof.
Preferably, an input unit is provided for instructing to segment the distortion image obtained for each slice in the circumferential axis direction into an arbitrary region.
Preferably, when determining the local plane for each part, a local plane that substantially matches the plane that includes the phase range of the spiral trajectory used for reconstruction of each part is selected.
Furthermore, the present invention is characterized by being an execution program for the artifact correction means, an image processing apparatus equipped with the artifact correction means, or an image reconstruction apparatus equipped with the artifact correction means.

FIG. 1 is a block diagram showing an outline of a helical scanning tomography apparatus according to an embodiment of the present invention;
FIG. 2 is a flowchart illustrating image creation according to the prior art;
3 (a), 3 (b) and 3 (c) are schematic diagrams showing the principle of the weighted spiral correction method and spatial distortion;
FIG. 4 is a flowchart showing a steer step artifact correction process in the spiral scanning tomography apparatus shown in FIG. 1 according to the embodiment of the present invention;
5 (a) and 5 (b) are perspective views showing the effective local plane generated by the weighted spiral correction method;
6 (a) and 6 (b) are characteristic diagrams showing a spiral trajectory and an inclined local plane trajectory;
FIGS. 7 (a) and 7 (b) are diagrams showing the phase range of projection data necessary for segmenting a distortion image into a plurality of local planes and for reconstruction of each part;
8 (a) to 8 (d) are conceptual diagrams of spatial distortion correction;
FIG. 9 is a side view of the distorted image viewed from two directions;
FIG. 10 is a diagram illustrating a method for obtaining a point where a tilted local plane as a portion intersects the z-axis;
Figure 11 is an diagram showing the extension of the local plan a view of the FIG. 10 from the z-axis direction intersect at a position Z ₀ of the z-axis;
FIG. 12 is a graph showing the relationship between xyz coordinates and tvz coordinates;
FIG. 13 shows pixel values I _m (x _m , y _m ) located at (x, y) in the vicinity on two local planes in order to correct the pixel value I (x, y, z) on the local plane. ) And I _n (x _n , y _n ) are conceptual diagrams for obtaining I (x, y, z) by interpolation.

尚、数式（１）上で、ｔａｂｌｅは、焦点が螺旋軌跡３０に沿って３６０度周回したときのスキャナと患者テーブル３との相対的な移動量［ｍｍ／２π］である。数式（１）におけるｚ方向成分は、ｔａｂｌｅを定数とすると、図６（ａ）に示すように位相θに対して比例した直線で表される。
周回軸（ｚ軸）に対し垂直な面に対してφ［ｒａｄ］傾斜したある局所平面を考え、焦点の周回軸回り方向の位相をθ［ｒａｄ］とする。また、周回軸からの距離をｒ、位相をθ_ｉとすると、ある傾斜した局所平面を伸長した面Ｉ_{ｓｌａｎｔ}（ｘ，ｙ，ｚ）＝Ｉ_{ｓｌａｎｔ}（ｒ，θ_ｉ）は、数式（２）で表される。

数式（２）におけるｚ方向成分は、φ及びｒを定数とすると、図１０（ｂ）に示すように周回軸回りの位相に対して正弦的に変化する曲線で表される。
上記数式（１）及び数式（２）により、焦点軌跡（θ＝θ_ｉ）とある局所平面（ｒ＝ｓｏｄ）の周回軸方向の絶対誤差Ｚ_ｅｒｒは、数式（３）で表される。

図７（ａ）に歪み画像をどのように部分として把えるかを示す。図ではある歪み画像を４行４列のマトリックスに分割して１６の部分に分けている。この分割数分の傾斜した局所平面が作られるので、分割数を多くすると後の補間の精細さも向上し、目的断層像の画質が向上する。これら部分の形状には制限はなく、四辺形に限らず他の多角形や円等でも良い。
ここで歪み画像の生成時そのそれぞれの部分に寄与した実効的な投影データの位相範囲をそれぞれ±α［ｒａｄ］とすると、数式（４）が得られる。

なお、図７（ｂ）を参照して、部分１０１と１０２で、その部分の画像再構成に寄与した実効的投影データの位相範囲は、それぞれ−α_１０１からα_１０１と−α_１０２からα_１０２となる。部分１０１を再構成するには−α_１０１からα_１０１までの位相角（１８０°以上）が必要であるが、部分１０２を再構成するには−α_１０２からα_１０２までの位相角（１８０°未満）で足りる様子がわかる。このように部分毎にαは異なってくる。なお、ここでいう部分は、歪み画像の局所平面に対応する。
局所平面の傾斜角φ［ｒａｄ］は、数式（４）を変形することにより、数式（５）で表される。

なお、実際に適用する場合、α＝±π／２、およびα＝０の３点にて焦点軌跡３０と局所平面が交わるとすると、局所平面の傾斜角φ［ｒａｄ］は、次の数式（６）で表わすこともできる。

ここで、φは、重み関数を考慮せずに求めたが、φは重み関数を考慮することによってより正確に算出してもよい。この場合、処理時のビーム形状（ファンビーム、パラレルビーム）やスライス厚を考慮にいれる方が望ましい。
図８（ａ）から８（ｄ）は、空間的歪みを有する歪み画像（傾斜した局所平面の集合）から空間的歪みのない画像を得るための補正方法を示す概念図である。図８（ａ）中、３０はＸ線管１の螺旋軌跡を、４１、４２、４３はＸ線ビームを示す。図８（ｂ）及び８（ｃ）中、５１と５２は歪み画像を側面から見たものを、２４１、２４２、２４３は歪み画像５１の部分である傾斜した局所平面を、同じく２５１、２５２、２５３は歪み画像５２の部分である傾斜した局所平面を示す。
前述したように重み付け螺旋補正再構成法による歪み画像の局所平面は、アキシャル面に対して傾斜しているが、従来は、アキシャル面と平行な１枚の平面として扱われていたため、再構成画像を積み上げて作成したボリューム画像には図８（ｂ）や８（ｃ）の歪み画像５１や５２のような空間的歪みが生じている。図８（ｄ）は、上記のように空間的歪みを有する各再構成画像２２２、２２１、２２、２２３、２２４を再配置した空間的な歪みのないボリューム画像を示している。
次に、図８（ｂ）や８（ｃ）に示すボリューム画像を、図８（ｄ）に示すボリューム画像２２２、２２１、２２、２２３、２２４になるように再配置する方法について説明する。
図８（ａ）において、周回軸上にて交差するＸ線ビーム４１、４２、４３などの照射によって生成された歪み画像は、Ｘ線ビームのベクトルにほぼ沿った部分化ができる。ここでは簡単のために歪み画像の傾斜した局所平面２４１、２４２、２４３に部分化することを考える。部分化を細かくするほど目的断層像の精細さが向上し、粗くするほど処理速度があがるので、目的に合わせて適宜部分化の程度を決定する。なお、図８は側面図であるため局所平面２４１、２４２、２４３の３つに部分化しているように見えるが、実際は２次元平面上で部分化するので、より多い数に部分化される。
これを説明するため、図９には歪み画像の側面を直交する２方向から見た図が示される。図中、歪み画像５１は図８（ｂ）と同じ側面から見た図であり、歪み画像５６はこれと直交する側面から見た図である。これら２つの側面図からわかるように、歪み画像は３次元的に複雑にうねった形状をしている。また、そのうねりは多種多様である。そこで、このような歪み画像を２次元座標上複数の部分に分割して、それぞれを傾いた局所平面に置き換えることで目的断層像のｚ軸方向の補正が可能になる。
たとえば、図８（ｂ）の側面図で考えた場合、歪み画像５１の部分化に当っては、局所平面２４２が歪み画像５１のもっともくびれた部分、すなわちｚ軸との交点を通過するようにその位置を特定し、局所平面２４１と２４３は、局所平面２４２の端部を始点として歪み画像５１の面積を上下に２分するような線分として決定する。部分化した個々の局所平面は必ずしも同じ形と面積である必要は無いので、図８（ｂ）及び８（ｃ）上で局所平面２４１、２４２、２４３の長さが一致するとは限らない。歪み画像５２の傾斜した局所平面２５１、２５２、２５３も同様に決定する。
図８（ｃ）において、図８（ｂ）で求めた局所平面２４１、２４２、２４３の組と局所平面２５１、２５２、２５３の組を抽出して表示したが、その他の歪み画像を構成する局所平面の組も同様に決定する。たとえば、図８（ｃ）において、目的断層像２２を得たいときは、少なくとも２つの歪み画像５１と５２から得た局所平面２４１と局所平面２５１上の目的断層像２２上のボクセルＩに（ｘ，ｙ）位置上対応するボクセルＩ_ｍとＩ_ｎのデータから補間でＩの目標データを取得する。局所平面２４２と２５２間および局所平面２４３と２５３間でも同様に補正していくことで目的断層像２２の全面に渉りデータを補間して歪みのない画像を得る。同様に目的断層像２２１、２２２、２２３、２２４も画素ごとにそれらを挟み込む少なくとも２つの歪み画像の局所平面から補間していくことで取得していき、最終的に空間歪みのないボリューム画像を得ることができる。
なお、重み付け螺旋再構成法の場合、目的断層像上のある画素を生成するにあたり、周回軸方向にずれた位置の投影データを使用することがある。このため、周回軸方向に伸長した像が得られることがある。たとえば、従来技術では、図８（ｄ）の目的断層像２２３を作成するにあたり、図８（ｂ）の局所平面２５１、２５２、２５３を含む歪み画像５２を目的断層像２２３と擬制しているため実際は図８（ｃ）の２５１上Ｉ_ｎの位置にある情報があたかも２２３の左端部にあるかのように画像化されてしまう。本実施例によれば、周回軸方向においてより実際の位置に近い場所に像が作られるよう投影データを扱うため、このような伸縮を補正できる。
図１０には、局所平面２５３が周回軸と直交する座標軸ｔに対してφの角度を有し、ｚ軸と点Ｚ_０で交わる様子が示してある。
図１１に示すように周回軸（ｚ軸）に垂直な面（アキシャル面）に対して、φ［ｒａｄ］傾斜した局所平面の延長面がｚ軸と交わる位置をＺ_０、このＺ_０の位置に対応するＸ線焦点の位相をθ［ｒａｄ］、焦点が１回転したときのスキャナと患者テーブル３との相対的な移動量をｔａｂｌｅ［ｍｍ／２π］とすると、焦点位相θは、数式（７）に示すようにＺ_０を変数とする関数で表すことができる。

次に、周回軸を通り焦点一周回軸方向に垂直な面（周回軸方向をｚ軸、ｚ軸に垂直な方向で局所平面の勾配方向をｔ軸、とするｔｚ平面（図１０参照））と局所平面が成す傾斜直線の方程式を算出すると、数式（８）となる。

ここで、図１２に示すようにｔ軸及びｚ軸に垂直な焦点一周回軸方向の軸をｖ軸とすると、ｘｙ座標上の位置は、θ回転したｔｖ座標上では、数式（９）で表される。

数式（８）に数式（９）を代入すると、数式（１０）となる。

よって、位置（ｔ，ｖ，ｚ）を通る局所平面のｚ軸上の位置Ｚ_０は、数式（１０）を変形した数式（１１）によって表すことができる。

ただし、数式（１１）上のθ（Ｚ_０）は、数式（７）に示したようにＺ_０・２π／ｔａｂｌｅである。

になるまで複数回（例えば、５回）繰り返し演算して求める。

尚、数式（１２）のＺ_０の肩文字は繰り返し演算における繰り返し回数を示している。
次に、図１３を参照して、周回軸上の位置がＺ_ｍ０ ＜Ｚ_０＜Ｚ_ｎ０であり、それぞれ画素Ｉ_ｍ（ｘ_ｍ，ｙ_ｍ）、Ｉ_ｎ（ｘ_ｎ，ｙ_ｎ）の集合からなる２枚の局所平面２４１、２５１からｚ方向の補間処理により画素Ｉ（ｘ，ｙ）の集合からなる目的断層像２２を求める。ここでは、記述が複雑になるのを防ぐために２枚の局所平面２４１、２５１の画素値を線形的に補間して目的断層像２２の画素値を算出する例を説明するが、３枚以上の画像を用いて多点補間により算出してもよいし、また非線形的に補間してもよい。
位置（ｘ，ｙ）に対応する局所平面２４１と２５１のｚ軸上の位置ｚ_ｍ，ｚ_ｎを、前述した局所平面の位置Ｚ_０を求めた数式（１１）と同様に、数式（１３）により算出する。

ここで、局所平面２４１上画素Ｉ_ｍ及び局所平面２５１上画素Ｉ_ｎ上の座標を（ｘ_ｍ，ｙ_ｍ），（ｘ_ｎ，ｙ_ｎ）とし、（ｘ，ｙ）との対応について考えると、次の数式（１４）のように表される。

目的断層像上の位置（ｘ，ｙ，ｚ）における画素値Ｉ（ｘ，ｙ，ｚ）は、局所平面上画素Ｉ_ｍ，Ｉ_ｎのｚ軸上の位置ｚ_ｍ，ｚ_ｎと、目的断層像のｚ軸上の位置ｚとに基づいて位置ｚ_ｍ，ｚ_ｎを通る局所平面上の位置（ｘ，ｙ）における画素値Ｉ_ｍ（ｘ_ｍ，ｙ_ｍ），Ｉ_ｎ（ｘ_ｎ，ｙ_ｎ）を使用してＩ（ｘ，ｙ，ｚ）を補間することにより、数式（１５）により算出することができる。

上記のようにして目的断層像のすべての位置の画素値Ｉ（ｘ，ｙ，ｚ）を求め、画像を得る。同様にして、目的断層像のｚ軸上の位置毎に、同断層像を構成するすべての画素値を求めることで、空間的な歪みが補正されたボリューム画像を得ることができる。
ここで、目的断層像の作成に際し、同断層像１スライス当たりに必要なスライス数について考えると、歪み画像１スライスの関心領域（直径がＦＯＶ（Ｆｉｅｌｄ ｏｆ Ｖｉｅｗ）の円）における周回軸方向への広がり幅ｗｉｄｔｈ＿Ｚは、ｗｉｄｔｈ＿Ｚ＝ＦＯＶ・ｔａｎφであるから、周回軸上における再構成間隔をｒＰｉｔｃｈ、目的断層像を局所平面から内挿補間により生成するために必要なスライス数をｎＳｌｉｃｅとすると、スライス数ｎＳｌｉｃｅは、数式（１６）で表すことができる。

このようなステアステップアーチファクト補正手段９ａを採用することによって視覚的な検証を行った結果、ステアステップアーチファクトは殆ど見られなかった。
なお、本実施の形態は、２以上のいかなる列数で構成される多列Ｘ線検出器または平面Ｘ線検出器にも適用可能であり、如何なる螺旋ピッチ（３や４などの整数値に限定されるものではなく、１．５や２．５などといった小数値）にも対応することができる。また本実施の形態では、ステアステップアーチファクト補正手段はいかなる画像処理装置や画像再構成装置にも搭載可能であり、ハードウェアの追加なしにソフトウェアとして提供可能である。
さらに本実施の形態では、空間的歪みによって生じる周回軸方向の物体の大きさが実際よりも大きくなったり小さくなったりする現象を防ぐことができる。Hereinafter, embodiments of the present invention will be described with reference to the drawings.
FIG. 1 is a schematic configuration diagram of a helical scanning tomography apparatus according to an embodiment of the present invention.
This helical scanning tomography apparatus is a multi-row detector tomography apparatus, and includes an X-ray generator 1 that generates radiation (X-rays), a collimator 2 that collimates the generated X-rays, and a patient table 3. A scanner body having a multi-row X-ray detector 5 for detecting X-rays irradiated on the subject 4, an X-ray generator 1 and the multi-row X-ray detector 5 is continuously rotated around the rotation axis. And a scanner driving device 6 for relatively continuously moving the scanner body and the patient table 3 in the axial direction of the orbiting axis, a condition for controlling the scanner, and a scanner controller 7 for controlling the scanner, A collimator controller 8 for controlling the collimator 2 installed in the scanner controller 7, and image processing for pre-processing / image reconstruction processing and various types of analysis processing A location 9, a high voltage generator 10 for supplying a high voltage to the X-ray generator 1, and a display device 11 Prefecture. The image processing device 9 includes a steer step artifact correcting means 9a for executing the steer step artifact correction according to the present invention.
The X-ray generator 1 irradiates a region of interest of the subject 4 on the patient table 3 with a conical or pyramidal X-ray having a three-dimensional extent via a collimator 2. X-rays that have passed through the subject 4 are detected by the multi-row X-ray detector 5 and output to the image processing device 9 as projection data. The image processing apparatus 9 reconstructs a tomographic image from the input projection data using a weighted spiral correction reconstruction method, and corrects spatial distortion for each tomographic image having a different slice position to thereby produce a steer step artifact. To obtain a good corrected image. Details of this correction will be described later.
FIG. 2 is a flowchart showing an image creation procedure according to the technique disclosed in the above-mentioned US Pat. No. 5,802,134 (corresponding to International Publication No. 98/44847 and Special Table 2000-515411). This employs an algorithm different from the weighted spiral correction method which is the basic algorithm in the present invention.
First, in step S1, measurement parameters are set from the input device, and then a spiral scan is performed based on the measurement parameters set in step S2. In step S3, an inclined plane is defined, projection data corresponding to the inclined plane is extracted in step S4, the weighting function extracted in step S5 is weighted to generate spiral corrected projection data, and the inclined plane is reconstructed in step S6. In step S7, an inclined surface is interpolated to generate an axial surface.
3A, 3B, and 3C are schematic diagrams relating to the principle of the weighted spiral correction reconstruction method and the generation of spatial distortion. In the figure, reference numeral 50 denotes a rotating shaft. FIG. 3C shows a state in which the local plane 24 of the image having spatial distortion is locally shifted from the target tomographic image 22 by φ [rad]. In the figure, S ₀ and S ₂ indicate X-ray sources, and S ₁ indicates 22 and 24 reconstruction positions.
In the weighted spiral correction reconstruction method, for each beam having a cone angle as shown in FIG. 3A, the cone angle is ignored as shown in FIG. The reconstructed image is treated as a beam that is orthogonal at the intersecting position and arranged in parallel in the direction of the rotation axis. Therefore, the reconstructed image has different distortions at each position on the rotation, that is, the rotation phase (view) of the focus, and the table feed speed. And spatial distortion due to non-uniformity of weight function, effective slice thickness, image distortion, and the like. Therefore, the plane of the tomographic image obtained by reconstruction is determined by the contribution ratio of all the X-ray beams used for the reconstruction of the tomographic image. When the volume images obtained by stacking the respective tomographic images thus reconstructed are, for example, MPR (Multi-Planer Reconstruction) or volume-rendered images, the spatial distortion in the circumferential rotation axis direction of each tomographic image. Discontinuities appear as steer step artifacts.
In this embodiment, in order to correct such a steer step artifact, the tomographic image obtained from the projection data array obtained by the helical scan by the weighted helical correction reconstruction is compared with the measurement condition and the weight function. An effective spatial distortion amount is calculated, and an image in which the steer step artifact is corrected is obtained by correcting the tomographic image based on the effective spatial distortion amount (accurately rearranged in the Euclidean space). ing.
FIG. 4 is a flowchart showing a processing procedure for correcting a steer step artifact in a tomographic image (reconstructed image) reconstructed by the weighted spiral correction reconstruction method.
After obtaining a reconstructed image by the weighted spiral correction reconstruction method in steps S11 to S14, the steer step artifact correction means 9a shown in steps S15 to S19 performs the steer step artifact correction.
Based on the conditions set in advance in step S15, the inclination angle of the reconstructed image is calculated as a spatial distortion amount by the calculation means. In subsequent steps S16 to S19, the correction means calculates the spatial distortion calculated by the calculation means. Correct the image.
As shown in the figure, the correction of the spatial distortion by the correction means is performed by determining the z position of the tomographic image (hereinafter referred to as the target tomographic image) to be finally obtained by correcting the spatial distortion in step S16, and in step S17. By calculating the position on the rotation axis (z axis) of the distortion image including each voxel (x, y, z) on the target tomographic image, the slice position in the circumferential rotation direction is specified. Next, the actual position on the z-axis of the distorted image (two or more reconstructed images sandwiching the identified slice position) close to the previously identified slice position in step S18 is calculated. In step S19, the 2 The pixel value of each voxel (x, y, z) on the target tomographic image is calculated by interpolating the pixel value on each position (x, y) of the reconstructed images in the position relation on the z axis.
Spatial distortion is regarded as a set of local planes inclined with respect to the axial plane for each part on the distortion image. Hereinafter, a method for calculating the distortion of the reconstructed image will be described.
5 (a) and 5 (b), if the phase around the rotation axis of the X-ray source (focal point) that irradiates X-rays is θ [rad], and the distance between the focus and the rotation axis is sod, The focal position S (θ) = (X _S (θ), Y _S (θ), Z _S (θ)) is expressed by Equation (1).

In Equation (1), “table” is a relative movement amount [mm / 2π] between the scanner and the patient table 3 when the focal point is rotated 360 degrees along the spiral locus 30. The z-direction component in Equation (1) is represented by a straight line proportional to the phase θ as shown in FIG.
Considering a certain local plane inclined by φ [rad] with respect to a plane perpendicular to the rotation axis (z-axis), the phase of the focus around the rotation axis is defined as θ [rad]. Further, when the distance from the rotation axis is r and the phase is θ _i , a plane I _slant (x, y, z) = I _slant (r, θ _i ) obtained by extending a certain inclined local plane is expressed by the following equation (2). It is represented by

The z-direction component in Equation (2) is represented by a curve that changes sinusoidally with respect to the phase around the rotation axis as shown in FIG. 10B, where φ and r are constants.
From the above formulas (1) and (2), the absolute error Z _{err in the} rotation axis direction of the focal plane (θ = θ _i ) and a certain local plane (r = sod) is expressed by formula (3).

FIG. 7A shows how a distortion image can be grasped as a part. In the figure, a certain distorted image is divided into a matrix of 4 rows and 4 columns and divided into 16 parts. Since local planes inclined by the number of divisions are created, increasing the number of divisions also improves the accuracy of later interpolation and improves the image quality of the target tomographic image. The shape of these portions is not limited, and is not limited to a quadrilateral, but may be other polygons or circles.
Here, when the phase range of effective projection data that contributes to the respective portions at the time of generation of the distorted image is ± α [rad], Expression (4) is obtained.

Referring to FIG. 7B, the phase ranges of effective projection data that contributed to the image reconstruction of the portions 101 and 102 are −α ₁₀₁ to α ₁₀₁ and −α ₁₀₂ to α ₁₀₂ , respectively. It becomes. Reconstructing the portion 101 requires a phase angle from −α ₁₀₁ to α ₁₀₁ (180 ° or more), but reconstructing the portion 102 requires a phase angle from −α ₁₀₂ to α ₁₀₂ (180 °). Less than). In this way, α varies from part to part. Note that the portion here corresponds to the local plane of the distorted image.
The inclination angle φ [rad] of the local plane is expressed by Expression (5) by modifying Expression (4).

In actual application, if the focal locus 30 and the local plane intersect at three points α = ± π / 2 and α = 0, the inclination angle φ [rad] of the local plane is expressed by the following formula ( 6).

Here, φ is obtained without considering the weight function, but φ may be calculated more accurately by considering the weight function. In this case, it is desirable to consider the beam shape (fan beam, parallel beam) and slice thickness during processing.
FIGS. 8A to 8D are conceptual diagrams showing a correction method for obtaining an image having no spatial distortion from a distortion image having a spatial distortion (a set of inclined local planes). In FIG. 8A, 30 indicates a spiral locus of the X-ray tube 1, and 41, 42, and 43 indicate X-ray beams. 8 (b) and 8 (c), 51 and 52 show the strain image as viewed from the side, 241, 242, and 243 denote inclined local planes that are portions of the strain image 51, and 251, 252, Reference numeral 253 denotes an inclined local plane which is a part of the distorted image 52.
As described above, the local plane of the distorted image obtained by the weighted spiral correction reconstruction method is inclined with respect to the axial plane. However, conventionally, the distortion plane is treated as a single plane parallel to the axial plane. In the volume image created by stacking the images, spatial distortions such as the distortion images 51 and 52 in FIGS. 8B and 8C are generated. FIG. 8D shows a volume image without spatial distortion in which the reconstructed images 222, 221, 22, 223, and 224 having spatial distortion are rearranged as described above.
Next, a method for rearranging the volume images shown in FIGS. 8B and 8C so as to become the volume images 222, 221, 22, 223, and 224 shown in FIG. 8D will be described.
In FIG. 8A, a distortion image generated by irradiation of X-ray beams 41, 42, 43 and the like intersecting on the rotation axis can be partially divided along the vector of the X-ray beam. Here, for the sake of simplicity, it is considered that the distortion image is partially divided into inclined local planes 241, 242, and 243. The finer the partialization, the higher the definition of the target tomographic image, and the rougher the processing speed, the higher the processing speed. Therefore, the degree of partialization is appropriately determined according to the purpose. Since FIG. 8 is a side view, it seems that it is partially divided into three local planes 241, 242, and 243, but in reality, it is partially divided on a two-dimensional plane, and thus is divided into a larger number.
In order to explain this, FIG. 9 shows a view of the side surface of the distorted image viewed from two orthogonal directions. In the figure, the distorted image 51 is a diagram viewed from the same side as FIG. 8B, and the distorted image 56 is a diagram viewed from a side surface orthogonal thereto. As can be seen from these two side views, the distortion image has a three-dimensionally complex wavy shape. Moreover, the swell is various. Therefore, by dividing such a distorted image into a plurality of parts on two-dimensional coordinates and replacing each with a tilted local plane, it becomes possible to correct the target tomographic image in the z-axis direction.
For example, in the case of the side view of FIG. 8B, when the distortion image 51 is partial, the local plane 242 passes through the narrowest part of the distortion image 51, that is, the intersection with the z axis. The position is specified, and the local planes 241 and 243 are determined as line segments that divide the area of the distorted image 51 in the vertical direction starting from the end of the local plane 242. Since the individual local planes that have been segmented do not necessarily have the same shape and area, the lengths of the local planes 241, 242, and 243 do not necessarily match in FIGS. 8B and 8C. The inclined local planes 251, 252, and 253 of the distortion image 52 are determined in the same manner.
In FIG. 8C, the set of local planes 241, 242, and 243 and the set of local planes 251, 252, and 253 obtained in FIG. 8B are extracted and displayed. The plane set is determined in the same manner. For example, in FIG. 8C, when it is desired to obtain the target tomographic image 22, the local plane 241 obtained from at least two distortion images 51 and 52 and the voxel I on the target tomographic image 22 on the local plane 251 are (x , y) to obtain the target data of I in position on the corresponding interpolated from the data of the voxel I _m and I _n. By correcting in the same manner between the local planes 242 and 252 and between the local planes 243 and 253, the interference data is interpolated over the entire surface of the target tomographic image 22 to obtain an image without distortion. Similarly, the target tomographic images 221, 222, 223, and 224 are acquired by interpolating from the local plane of at least two distortion images sandwiching them for each pixel, and finally a volume image without spatial distortion is obtained. be able to.
In the case of the weighted spiral reconstruction method, when generating a certain pixel on the target tomographic image, projection data at a position shifted in the rotation axis direction may be used. For this reason, an image extended in the direction of the rotation axis may be obtained. For example, in the prior art, in creating the target tomographic image 223 in FIG. 8D, the distortion image 52 including the local planes 251, 252, and 253 in FIG. 8B is simulated with the target tomographic image 223. in fact the information on the position of 251 on I _n shown in FIG. 8 (c) would be imaged as if it were located at the left end of 223. According to this embodiment, since the projection data is handled so that an image is formed at a place closer to the actual position in the direction of the rotation axis, such expansion and contraction can be corrected.
FIG. 10 shows a state in which the local plane 253 has an angle of φ with respect to the coordinate axis t orthogonal to the rotation axis and intersects the z axis at the point Z ₀ .
As shown in FIG. 11, the position where the extended surface of the local plane inclined by φ [rad] intersects the z axis with respect to the plane (axial plane) perpendicular to the rotation axis (z axis) is Z ₀ , and this Z ₀ position When the phase of the X-ray focus corresponding to is θ [rad] and the relative movement amount of the scanner and the patient table 3 when the focus is rotated once is table [mm / 2π], the focus phase θ is expressed by the formula ( As shown in 7), it can be expressed by a function having Z ₀ as a variable.

Next, a plane that passes through the rotation axis and is perpendicular to the focal rotation axis direction (the tz plane where the rotation axis direction is the z axis, the direction perpendicular to the z axis is the gradient direction of the local plane is the t axis (see FIG. 10)) When the equation of the inclined straight line formed by the local plane is calculated, Equation (8) is obtained.

Here, as shown in FIG. 12, assuming that the axis in the direction of the focal round axis perpendicular to the t-axis and the z-axis is the v-axis, the position on the xy coordinate is expressed by Equation (9) on the tv coordinate rotated by θ. expressed.

Substituting equation (9) into equation (8) yields equation (10).

Therefore, the position Z ₀ on the z-axis of the local plane passing through the position (t, v, z) can be expressed by Expression (11) obtained by modifying Expression (10).

However, θ (Z ₀ ) in Equation (11) is Z ₀ · 2π / table as shown in Equation (7).

It is obtained by repeatedly calculating multiple times (for example, 5 times) until

Note that the acronym for Z ₀ in Expression (12) indicates the number of repetitions in the iterative calculation.
Next, referring to FIG. 13, the position on the _rotation axis is Z _m0 < Z ₀ <Z _n0 , and a set of pixels I _m (x _m , y _m ) and I _n (x _n , y _n ), respectively. A target tomographic image 22 consisting of a set of pixels I (x, y) is obtained from two local planes 241 and 251 consisting of Here, an example of calculating the pixel value of the target tomographic image 22 by linearly interpolating the pixel values of the two local planes 241 and 251 in order to prevent the description from becoming complicated will be described. It may be calculated by multipoint interpolation using an image, or may be nonlinearly interpolated.
Similar to Equation (11) in which the positions z _m and z _n of the local planes 241 and 251 corresponding to the position (x, y) on the z-axis are obtained as the position Z ₀ of the local plane described above, Equation (13) Calculated by

Here, the coordinates on the local plane 241 above the pixel _{I m} and the local plane 251 above the pixel _{I n (x m, y m} ), and _{(x n, y} n), considering the correspondence between (x, y) Is expressed as the following formula (14).

Position on the object tomogram (x, y, z) the pixel value I in (x, y, z) is the local plane on the pixel _I m, the position _z m on the z-axis of the _{I n,} and _{z n,} object fault Based on the position z on the z-axis of the image, the pixel values I _m (x _m , y _m ), I _n (x _n , y) at the position (x, y) on the local plane passing through the positions z _m and z _{n n} ) is used to interpolate I (x, y, z), and can be calculated according to equation (15).

As described above, pixel values I (x, y, z) at all positions of the target tomographic image are obtained to obtain an image. Similarly, a volume image in which spatial distortion is corrected can be obtained by obtaining all pixel values constituting the tomographic image for each position on the z-axis of the target tomographic image.
Here, when creating the target tomographic image, considering the number of slices required for each slice of the tomographic image, the distortion image in the region of interest (diameter of FOV (Field of View) circle) in the circumferential axis direction in the one-slice image. Since the spread width width_Z is width_Z = FOV · tanφ, if the reconstruction interval on the rotation axis is rPitch and the number of slices necessary for generating the target tomographic image from the local plane by interpolation is nSlice, the number of slices nSlice can be expressed by Equation (16).

As a result of visual verification by adopting such a steer step artifact correcting means 9a, almost no steer step artifact was seen.
The present embodiment can be applied to a multi-row X-ray detector or a planar X-ray detector having any number of columns of 2 or more, and is limited to any helical pitch (an integer value such as 3 or 4). It is also possible to deal with decimal values such as 1.5 and 2.5. In the present embodiment, the steer step artifact correcting means can be mounted on any image processing apparatus or image reconstruction apparatus, and can be provided as software without adding hardware.
Furthermore, in the present embodiment, it is possible to prevent a phenomenon in which the size of the object in the direction of the rotation axis caused by spatial distortion becomes larger or smaller than the actual size.

  As described above, according to the helical scanning tomography apparatus according to the present invention, a tomographic image is created from projection data obtained at any helical pitch, and a three-dimensional image is created from the created tomographic image. Since the spatial distortion of the reconstructed image is corrected, the steer step artifact that occurs depending on the spatial distortion of the reconstructed image can be reduced.

Claims (9)

A radiation source that emits conical or pyramid-shaped radiation having a three-dimensional extent; and a two-dimensional radiation detector that is disposed opposite to the radiation source via an object and detects radiation that passes through the object. A scanner body having
The radiation source and the radiation detector of the scanner main body are simultaneously rotated around the rotation axis, and the scanner main body and the table on which the object is placed are relatively moved in the axial direction of the rotation axis. Scanner driving means for spirally scanning the object,
A tomographic image reconstruction means for reconstructing a distortion image for each slice in the rotation axis direction from the two-dimensional projection data detected by the radiation detector during spiral scanning of the object using a weighted spiral correction reconstruction method; In spiral CT apparatus,
Data corresponding to the plane coordinate position of the pixel from a plurality of local planes that divide the distortion image into an arbitrary plurality, obtain local planes for each part, and sandwich pixels on a predetermined target tomographic image in the direction of the rotation axis A helical CT apparatus comprising an artifact correction unit configured to interpolate the image on the target tomographic image for each pixel and obtain a volume image thereof. The spiral CT apparatus according to claim 1, further comprising an input unit that instructs to distort the distortion image obtained for each slice in the circumferential axis direction into an arbitrary region. 2. The helical CT apparatus according to claim 1, wherein, in obtaining the local plane for each part, a local plane that substantially coincides with a plane that includes the phase range of the spiral trajectory used for reconstruction of each part is selected. The execution program of the artifact correction means according to claim 1. The execution program of the artifact correction means of Claim 3. An image processing apparatus equipped with the artifact correction means according to claim 1. An image processing apparatus equipped with the artifact correction means according to claim 3. An image reconstruction apparatus equipped with the artifact correction means according to claim 1. An image reconstruction apparatus equipped with the artifact correction means according to claim 3.

Images