JP4163526B2 - Displacement measurement system and method using MR tagging images - Google Patents

Description

と表される。ただし、｛θ_ｉ｝，ｉ∈｛１，２，３｝は初期位相とする。ここで、座標変換
【数２】

を行えば、
【数３】

と表される。ただし、ＦＦＴが可能なディジタル画像では、｛ｌ_ｉ，ｍ_ｉ，ｎ_ｉ｝，ｉ∈｛１，２，３｝は整数をとる。
【００１５】
タグＴ（ｐ）である
【数４】

の個々の要素が１−１ＳＰＡＭＭに代表される単一周波数で生成され、原画像Ｇ^０（ｐ）に付加されているとすれば、タグは基準縞画像そのものとなる。すなわち、ｉ∈｛１，２，３｝において、
【数５】

である。
【００１６】
一方、タグＴ（ｐ）が箱型関数で生成されている場合、ＤＣ成分を除き基準縞画像の周波数を最小にした整数倍の周波数成分の重ね合わせと見ることができる。すなわち、
【数６】

ここで、ｊは有限の周波数成分を表し、｛Ａ_ｉｊ｝は各周波数成分の強度を与える定数であるとする。以降、一般論として１−１ＳＰＡＭＭによるタグも、ｊ＝｛０，１｝とした上式の一つとして議論する。
【００１７】
Ｔ＋１枚の連続ＭＲ画像が撮像されたとし、フレームｔ∈｛０，１，…，Ｔ｝での、タグの付加されていない連続画像の質点ｐ^ｔの輝度をＧ^ｔ（ｐ^ｔ）とする。フレーム０におけるタグ付きの画像Ｉ^ｔ（ｐ^ｔ；｛θ_ｉ｝，｛Ａ_ｉｊ｝）｜_ｔ＝０は、Ｇ^ｔ（ｐ^ｔ）｜_ｔ＝０に、上記で与えられるタグが付加されたものと見なすことができるから
【数７】

と表される。
ｔ≠０の場合、上記関数で与えられるタグが組織の変位・変形にしたがって歪みを持つ。時相０の任意の質点ｐ^０が、時相ｔでｐ^ｔ＝ｐ^０＋Δｐ，Δｐ＝（Δｕ_１，Δｕ_２，Δｕ_３）^Ｔで表される座標に移動するとき、フレームｔにおけるタグのない画像をＧ^ｔ（ｐ^ｔ）とすれば、タギング画像Ｉ^ｔ（ｐ^ｔ）は
【数８】

で与えられる。
【００１８】
時相ｔの画像Ｉ^ｔ（ｐ^ｔ；｛θ_ｉ｝，｛Ａ_ｉｊ｝） と基準縞画像Ｔ_Ｂｉ（ｕ^ｔ _ｉ；θ_Ｂｉ）との積を、Ｍ^ｔ _ｉ（ｐ^ｔ，Δｐ^ｔ，｛θ_ｉ｝，｛Ａ_ｉｊ｝，θ_Ｂｉ）とすれば、
【数９】

となる。上式において、右辺第１項はバイアス項、第２項はＭＲ画像上に現れたタグそのものを表し、第３項は基準縞画像を表す。また、第４項以降は、個々の周波数成分との積であり、この中で座標ｕ^ｔ _ｉに依存しない項はＡ_ｉ１にかかる第２項だけである。一般にΔｕ^ｔ _ｉはｕ^ｔ _ｉの空間変化よりも小さいから、ＬＰＦによってこの要素のみを抽出する。局所的なＬＰF（平均化フィルタ）を適用すれば、Ａ_ｉ０，Ａ_ｉ１も局所的要素として表現できることから、質点ｐ^ｔのそれらをＡ_ｉ０（ｐ^ｔ），Ａ_ｉ１（ｐ^ｔ）として表すことができる。上記の項をＬ^ｔ _ｉとして表し、整理すれば
【数１０】

を得る。
【００１９】
位相シフト法は、位相｛θ_Ｂｉ｝を変化させた基準縞画像Ｔ_Ｂｉを用意し、それらの縞との干渉をとることで、干渉縞の位相を得るための方法である。ここで、基準縞画像Ｔ_Ｂｉの位相θ_Ｂｉを２πｋ／Ｋ＋θ_１，ｋ∈｛０，…，Ｋ−１｝としたＫ枚の低周波数成分画像Ｌ^ｔ _ｉ（ｐ^ｔ，Δｐ^ｔ；｛θ_ｉ’｝，θ_Ｂｉ＝２πｋ／Ｋ）にｓｉｎ（２πｋ／Ｋ）をかけ、ｋに関して足し合わせたものをＳ^ｔ _ｉとすれば、Ｋが４以上の偶数であるとき、
【数１１】

となる。同様にｃｏｓ（２πｋ／Ｋ）をかけ、足し合わせたものをＣ^ｔ _ｉとすれば、
【数１２】

となる。したがって、Δｕ^ｔ _ｉ（ｐ^ｔ）は、
【数１３】

として、求めることができる。ただし、この求めた変位は、［−π，π）の間に限定されている。
【００２０】
時間的に推移した場合、タグが部分的に減衰する。または元来タグがほとんど現れない組織も存在することから、これらの評価が必要になる。このとき、振幅強度Ｗ^ｔ _ｉ（ｐ^ｔ）を
【数１４】

として求めれば、元来輝度が小さいか、タグの基本第一周波数振幅Ａ_ｉ１が局所的に小さくなる質点ｐ^ｔではこの値が小さくなることから、画像輝度に変位の信頼度としての定量的評価を可能にする。一般には、タグの存在する方向に関する局所的フィルタリングはできないので、二次元または三次元の局所的ＬＰＦを用いるとすれば、式（９）の注目方向以外のタグも消去され、
【数１５】

となる。この場合、同様の処理を行なえば、Δｕ^ｔ _ｉ（ｐ^ｔ），Ｗ^ｔ _ｉ（ｐ^ｔ）は式（１２），（１３）の代わりに、
【数１６】

として求められ、注目方向以外のタグの要素が振幅強度に含まれるかどうかの差異が生じる。
マスクを生成する場合、任意のフレームｔに、それぞれの位相に対してＷ^ｔ _ｉ（ｐ）が存在するので、これらを以下のように自乗平均で統合したものを用いるとよい。
【数１７】

【００２１】
本方法は、任意の画像（例えば図３（ｂ））のような変位画像から、歪みや変位のない画像（例えば図３（ａ））からみたときの変位の量を直接的に求めることが可能になる。
また、算出される位相の値域が［−π，π）に限定される問題に対して、時相０と時相１の画像を用いて算出した時相１の位相差に対して、時相０と時相２の画像を用いて算出した時相２の位相差が、［−π，π）となるように設定してやることで、任意の時相における位相算出、すなわち変位算出が可能になる。すなわち、同一質点を与えるフレームｔ−１，ｔの座標をｐ^ｔ−１，ｐ^ｔとすれば、式（１２）は
【数１８】

と表す。ここで、ｎ^ｔ _ｉ（ｐ^ｔ）はΔｕ^ｔ _ｉ（ｐ^ｔ）の整数部分を表す。この方法では、補間を用いる必要があるが、ｐ^ｔと同じ座標のフレームｔ−１の質点Δｕ^ｔ−１ _１ （ｐ^ｔ−１）の周期が緩やかであるとすれば、ｐ’^ｔ−１によって求められる位相Δｕ^ｔ−１ _１（ｐ’^ｔ−１）によって代替することが可能である。
【００２２】
上述したように、本発明は、小さな変位を求めると同時に、その変位を得るために用いたタグの強度を定量的に求めることで、変位の信頼性を算出するのに用いることができる。
本発明は心臓での使用に特に適しており、虚血性心疾患および心筋梗塞の検出および定量化を評価するのに用いることができる。
また、二つのタグ方向に関する振幅画像から心壁，細胞壁，血管，筋肉，腱，血液，髄液等のいずれかの組合せによる組織の違いを見出すことができる。
使用するＭＲタギング画像としては、グラディエントエコー法，スピンエコー法やスパイラル法からなるグループから選択されたＭＲＩパルスシーケンスを用いることができる。
上述では、図９のように変位を矢印で示した図を示しているが、変位を変位等値線で示すこともできる。
なお、本発明の構成は、上記した実施の形態に限定されるものではなく、本発明の要旨を逸脱しない範囲内において、種々変更を加え得ることは勿論である。
【００２３】
【発明の効果】
以上、説明したように本発明では、ＭＲタギング画像を画像処理することにより、タグから対象物の変位（動き）を高精度に出すことができる。また、タグの存在や残存の強度を数値化し、変位（動き）の信頼度として用いることである。
【図面の簡単な説明】
【図１】本発明の実施形態における処理を示すフローチャートである。
【図２】処理の概略を示す模式図である。
【図３】処理される前の画像例を示す図である。
【図４】２方向の縞画像の例を示す図である。
【図５−１】位相の異なる縞画像との積を取った画像例を示す図である。
【図５−２】位相の異なる縞画像との積を取った画像例の続きを示す図である。
【図６−１】図５−１の画像のローパスフィルタ処理後の画像例を示す図である。
【図６−２】図５−２の画像のローパスフィルタ処理後の画像例を示す図である。
【図７】位相画像と振幅強度画像の例を示す図である。
【図８】振幅強度画像から作成したマスク画像（ａ）と、位相画像に対するマスク処理後の画像（ｂ）を示す図である。
【図９】心臓部分を切り取った位相画像（ａ）と、それから得られた変位を示す画像（ｂ）を示す図である。[0001]
BACKGROUND OF THE INVENTION
The present invention relates to measuring displacement (movement) of an object from an MR tagging image.
[0002]
[Technical background]
An MR image is obtained by imaging the resonance energy emitted by spins by aligning spins of atoms in the body with a magnetic field and irradiating the spins with RF pulses. The MR tagging image is an MR image obtained by labeling the tissue with luminance, and SPAMM (spatial modulation of magetization) is a typical method for generating the tagging image. As a method for this, a magnetic field is used as preprocessing for MR image generation. Is spatially modulated to generate a partial spin excitation state, thereby adding a specific luminance pattern to the MR image.
[0003]
It is known to use two sets of tagging planes orthogonal to the image plane for two-dimensional myocardial motion imaging in MRI imaging through spatial magnetization modulation (SPAMM) (Patent Literature). 1-3, refer nonpatent literatures 1 and 2).
It is known to combine features using image processing that detects tag features in relation to MRI, and then using interpolation (interpolation) in motion maps related to placement and distortion (eg, (See Non-Patent Documents 3 and 4). These approaches are not completely automatic because they require some manual manipulation.
It is also known to use optical flow in MR tagging image sequences (see Non-Patent Documents 5 to 7).
In the above approach, a sine wave tag pattern is used instead of a box function (inverse Fourier transform of a sinc function) or a saturation tag pattern that replaces the box function. The temporal luminance change obtained from two image pairs and the spatial luminance gradient of the image are features used to provide a dense motion estimation method known as optical flow.
In such an approach, since only information related to one-dimensional motion with respect to the spatial luminance gradient direction can be acquired, a regularization method (referring to nearby motion) is used in motion calculation.
It is also known to use angle images in MR tagging image sequences (see Patent Document 4). This approach divides two sets of tag elements from a single image in the frequency domain and calculates an angle image for each tag element. By calculating the optical flow for each element from two consecutive images, it is possible to calculate the motion without regularization.
The problem with these motion calculations is that there are no methods to detect the disappearance of tags due to the attenuation of brightness, or the motion detection accuracy is extremely poor at parts where no tags are added. It is. It also means that they must be detected manually.
Optical flow calculation requires the linearity of the spatial change of the target feature, called the gradient equation, and the motion of the pixel where there is a spatial change in brightness or angle where the linearity is not sufficiently established. In calculation, the accuracy decreases.
At the same time, pixels that have little temporal change in luminance have a wide area that satisfies the gradient equation with a small observation error, and the reliability of the calculated motion decreases. In the movement calculation, the accuracy is particularly lowered.
[0004]
[Patent Document 1]
U. S. Pat. No. 5054489
[Patent Document 2]
U. S. Pat. No. 511118
[Patent Document 3]
U. S. Pat. No. 5217016
[Patent Document 4]
U. S. Pat. No. 6453187
[Non-Patent Document 1]
Axel et al., MR Imaging of Motion with Spatial Modulation of Magnetization, Radiology, vol. 171, pp. 841-845, 1989
[Non-Patent Document 2]
Axel et al., Heart Wall Motion: improved Method of Spatial Modulation of Magnetization for MR imaging, Radiology, vol. 172 (1), 349-350, 1989
[Non-Patent Document 3]
Young et al., Three-dimensional motion and deformation with spatial motion of magnetization, Radiology, vol. 185, pp. 241-247, 1992
[Non-Patent Document 4]
McVeigh et al., Noninvasive measurements of transmural gradient in myocardial strain with MR imaging, Radiology, vol. 180, no. 3, pp.677-683, 1991
[Non-Patent Document 5]
Prince et al., Motion estimation from tagged MR image sequences, IEEE Trans.on Medical Imaging, vol. 11, no. 2, pp.238-249, 1992
[Non-Patent Document 6]
Gupta et al., On variable brightness optical flow for tagged MRI, Tech.rep. 95-13, JHU / ECE, 1995
[Non-Patent Document 7]
Gupta et al., Bandpass optical flow for tagged MR Imaging, Proc.IEEE ICIP, 1997
[0005]
[Problems to be solved by the invention]
In the above-described method for searching for the same luminance value, the spin reaches an equilibrium state with time, so that the MR image itself obtained by imaging it is attenuated by continuous imaging. At the same time, the portion where the spin is not directed in one direction (uniformly distributed) by tagging also becomes close to the original brightness of the MR image due to the influence of the magnetic field. Since these functions function to reduce the amplitude of the tag, a method for quantitatively evaluating how much amplitude is required is required. Further, when tracking the luminance, a method that considers the change in luminance is necessary.
An object of the present invention is to perform motion from a tag with high accuracy by performing image processing on an MR tagging image. Also, the presence or remaining strength of the tag is digitized and used as the reliability of the movement.
It is also an object of the present invention to accurately visualize the movement.
[0006]
[Means for Solving the Problems]
In order to achieve the above object, the present invention is a system for measuring displacement using an MR tagging image, and includes an MR tagging image and an interference image of a reference fringe image and a plurality of fringe images having different phases from the reference fringe image. An interference image generating means for generating a phase image, a phase image generating means for generating a phase image from the interference image, and a displacement calculating means for obtaining a displacement from the phase image.
The interference image generation means can generate an interference image by taking the product of the MR tagging image and the fringe image and applying a low-pass filter.
Furthermore, an amplitude image generating means for generating an amplitude image from the interference image can be provided.
Furthermore, a mask generation unit that generates a mask image for extracting a highly reliable displacement portion from the amplitude image is provided, and the displacement calculation unit obtains the displacement after processing the phase image with the mask image. it can.
Further, the reference fringe image is provided for each tag direction, the fringe images having different phases are provided for each tag direction, and the phase image generation unit and / or the amplitude image generation unit is configured to output the phase image and / or the tag image for each tag direction. Alternatively, an amplitude image is generated, and the displacement calculation means can calculate a two-dimensional or three-dimensional displacement.
In addition, the method executed by the system and the program capable of constructing the system in a computer system are also the present invention.
[0007]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings.
The present invention measures the movement (displacement) of an MR image including a pulse sequence that spatially modulates a region of interest of an object by image processing with a computer. The spatially modulated image is obtained by computer processing phase information independent of the spatially modulated frequency by being interfered with a reference fringe having the same fundamental frequency as the tag. This phase information is used for measuring displacement information of the object. At the same time, by acquiring amplitude intensity information based on the phase information, the intensity (reliability) of the tag for which the displacement of the acquired pixel is calculated is quantified by computer processing.
[0008]
Image processing for obtaining a displacement (motion) from an MR tagging image (an image with a tag) will be described with reference to FIGS. FIG. 1 is a flowchart showing image processing according to an embodiment of the present invention, and FIG. 2 is a schematic diagram showing an image processing process. 3 to 9 are examples showing an image processing process of a tagged image near the heart.
The following “phase image” means a phase of an image obtained by extracting a low frequency component when interference is generated with respect to an MR image tagged by spatial modulation. “Image” means the intensity of the amplitude at each pixel of the low frequency component. Note that the spatially modulated image has the same frequency as the fundamental frequency of the tag.
[0009]
First, a tagging image is input from the MR imaging device (S110). As shown in FIG. 2A, an image (t = 0) in which the tag in the initial state is not disturbed and a tagging image (for example, t = 11) at a time when the displacement is desired are required. Here, as an example, “displacement” from t = 0 is calculated. FIG. 3 is an example of a tagged image near the heart. FIG. 3A is an initial image, and FIG.
[0010]
Next, the image is transformed into the frequency domain by Fourier transform (S120). Only the initial image at t = 0 is sufficient. An image in this frequency domain is shown in FIG. Two non-diagonal peak values (not including the peak value) are taken (S130: see FIG. 2C), and a complete sine wave image is generated from the inverse transformation (S140: FIG. 2 ( d)). Since this peak represents the frequency of the tag, if only the peak value is extracted, only one frequency can be extracted. Therefore, as shown in FIG. 2D, two fringe images having a single period can be generated. These images correspond to the tag directions, respectively. 4 (a) and 4 (b) are two striped images obtained from FIG. 3 (a).
When taking the peak value, except for the average component (DC component), in the 2-dimensional image of 1-1 SPAMM, four peaks (each tag has two peaks) appear. If a sinc function type RF pulse is irradiated, each quadrant has several peaks at a frequency higher than the fundamental frequency of the tag, but the fundamental frequency forms the maximum value in each quadrant except for the average. The maximum value may be searched in each quadrant.
Using this fringe image (FIGS. 4A and 4B) as a reference fringe image, a fringe image whose phase is shifted is generated (S150).
[0011]
A product of the shifted image (for example, FIG. 3B) and the fringe image obtained by shifting each phase obtained from the reference fringe image (FIGS. 4A and 4B) is obtained, and moire interference is generated (S160). FIGS. 5A and 5B show images after taking the product of the phase-shifted fringe image from FIG. 3B and FIG. 4A. 5-1 (a) to 5-2 (d) are images obtained by multiplying the fringe image with the phase shifted by ½π. Thereafter, the method described later may be followed as the principle of moire.
A low pass filter is applied to the images shown in FIGS. 5A and 5B to cut a frequency portion higher than the fringe frequency (S170). At this time, an interference fringe is formed, and an image whose luminance value changes with a deviation from the original image (FIG. 3A) is obtained. In the image after filtering, the pixel in which the deviation from the stripe does not occur has a high luminance value, and the luminance value is low in the portion where the deviation is π. Images obtained by applying a low-pass filter to the images shown in FIGS. 5-1 and 5-2 are shown in FIGS.
A phase image and an amplitude intensity image are obtained from this low-pass filtered image (S180). How to obtain these images will be described in detail later. FIGS. 7A and 7B show the phase image and the amplitude intensity image obtained from FIGS. 6-1 (a) to 6-2 (d), respectively.
[0012]
If it changes over time, the tag partially attenuates. Or, there are some organizations that rarely show tags originally, so these evaluations are necessary. In the amplitude intensity image, the amplitude intensity is small at the point where the luminance is originally small or the fundamental frequency amplitude of the tag is locally small, so that the image luminance can be quantitatively evaluated as a reliability of displacement. . From the amplitude intensity image, a mask for extracting only a portion with high reliability is generated. A mask image generated from the amplitude intensity image of FIG. 7B is shown in FIG. 8A, and an image obtained by processing the phase image of FIG. 7A with the mask of FIG. 8A is shown in FIG. (S190).
From FIG. 8B, the displacement can be illustrated so that the displacement can be easily grasped (S200). FIG. 9 (a) shows a heart part cut out from FIG. 8 (b), and FIG. 9 (b) shows the result of calculating the displacement of this part. In FIG. 9B, the displacement obtained from the two-direction stripe image shown in FIG. 4 is synthesized and indicated by a two-dimensional arrow.
[0013]
The basic period of the tag may be started as known. Since such a value depends on the device, if it is known from the beginning, an image of that cycle can be taken, and there is no need to perform the above-described processing for obtaining the above-mentioned fringe image.
In the present invention, a SPAMM pulse sequence can be used as the pulse sequence.
In the above description, an image in the frequency domain is obtained from the MR image. However, an image in the frequency domain called K space is also stored in the MR apparatus as an MR image including a tagging image. Therefore, it is possible to eliminate the need for an operation of extracting a frequency domain spectrum from an image by directly connecting to these devices.
If the displacement is calculated for all the images, the “movement” can be calculated. At this time, since the displacement is expressed as a periodic function of [−π, π), it is necessary to obtain the phase in order from the front and perform processing such as carry or carry down on [−π, π).
[0014]
<Detailed algorithm>
Generalized in a three-dimensional space (i = {1, 2, 3}) and described. In the case of a two-dimensional plane, it can be applied by setting (i = {1, 2}). Similarly, it can be applied to any dimension. Each of the reference fringe images T _Bi and iε {1, 2, 3} is generated by a single frequency. At this time, at an arbitrary mass point p = (x, y, z) on the reference fringe image,
[Expression 1]

It is expressed. However, {θ _i }, i∈ {1, 2, 3} are initial phases. Here, coordinate transformation

If you do
[Equation 3]

It is expressed. However, in a digital image capable of FFT, {l _i , m _i , n _i }, iε {1, 2, 3} takes an integer.
[0015]
The tag T (p)

Are generated at a single frequency represented by 1-1 SPAMM and added to the original image G ⁰ (p), the tag becomes the reference fringe image itself. That is, in i∈ {1, 2, 3},
[Equation 5]

It is.
[0016]
On the other hand, when the tag T (p) is generated by a box function, it can be regarded as a superposition of frequency components of an integral multiple obtained by minimizing the frequency of the reference fringe image except for the DC component. That is,
[Formula 6]

Here, j represents a finite frequency component, and {A _ij } is a constant that gives the intensity of each frequency component. Hereinafter, as a general theory, a tag based on 1-1 SPAMM is also discussed as one of the above equations where j = {0, 1}.
[0017]
And T + 1 sheet continuous MR imaging the image, the frame t∈ {0,1, ..., T} with the luminance of the mass point ^{p t} of the unattached continuous image tags and ^{G t (p} t) . Tagged image I ^t (p ^t ; {θ _i }, {A _ij }) | _{t =} 0 in frame 0 is obtained by adding the tag given above to G ^t (p ^t ) | _{t = 0} Since it can be regarded as something,

It is expressed.
When t ≠ 0, the tag given by the above function has distortion according to the displacement / deformation of the tissue. When an arbitrary mass point p ⁰ in time phase ⁰ moves to a coordinate represented by p ^t = p ⁰ + Δp, Δp = (Δu ₁ , Δu ₂ , Δu ₃ ) ^T in time phase t, the tag of frame t If no image is G ^t (p ^t ), the tagging image I ^t (p ^t ) is

Given in.
[0018]
The product of the image I ^t (p ^t ; {θ _i }, {A _ij }) of the time phase t and the reference fringe image T _Bi (u ^t _i ; θ _Bi ) is ^expressed as M ^t _i (p ^t , Δp ^t , {Θ _i }, {A _ij }, θ _Bi )
[Equation 9]

It becomes. In the above equation, the first term on the right side represents the bias term, the second term represents the tag itself that appeared on the MR image, and the third term represents the reference fringe image. The fourth and subsequent terms are products with individual frequency components, and the only term that does not depend on the coordinates u ^t _i is the second term relating to A _i1 . In general, since Δu ^t _i is smaller than the spatial change of u ^t _i , only this element is extracted by the LPF. If a local LPF (averaging filter) is applied, A _i0 and A _i1 can also be expressed as local elements. Therefore, those of the mass points p ^t are expressed as A _i0 (p ^t ) and A _i1 (p ^t ). Can do. If the above term is expressed as L ^t _i and rearranged,

Get.
[0019]
The phase shift method is a method for obtaining a phase of an interference fringe by preparing a reference fringe image T _{Bi in} which the phase {θ _Bi } is changed and taking interference with those fringes. Here, K low-frequency component images L ^t _i (p ^t , Δp ^t ; {θ) in which the phase θ _Bi of the reference fringe image T _Bi is 2πk / K + θ ₁ , kε {0,..., K−1}. _{i '}, θ Bi = 2πk} / K) to over sin (2πk / K), if those sum with respect to k and ^S _{t i,} when K is 4 or more even number,
## EQU11 ##

It becomes. Similarly over cos (2πk / K), if those sum and ^C _{t i,}
[Expression 12]

It becomes. Therefore, Δu ^t _i (p ^t ) is
[Formula 13]

As can be obtained. However, the obtained displacement is limited to [−π, π).
[0020]
If it changes over time, the tag partially attenuates. Or, there are some organizations that rarely show tags originally, so these evaluations are necessary. At this time, the amplitude intensity W ^t _i (p ^t ) is ^expressed as follows:

Therefore, since this value becomes small at the mass point p ^t where the luminance is originally small or the basic first frequency amplitude A _{i1 of} the tag is locally small, a quantitative evaluation as a reliability of displacement in the image luminance is performed. Enable. In general, since local filtering in the direction in which the tag exists cannot be performed, if a two-dimensional or three-dimensional local LPF is used, tags other than the target direction in Expression (9) are also deleted.
[Expression 15]

It becomes. In this case, if the same processing is performed, Δu ^t _i (p ^t ) and W ^t _i (p ^t ) are replaced by the expressions (12) and (13),
[Expression 16]

And a difference occurs in whether or not tag elements other than the target direction are included in the amplitude intensity.
When generating a mask, W ^t _i (p) is present for each phase in an arbitrary frame t. Therefore, it is preferable to use those obtained by integrating these with the root mean square as follows.
[Expression 17]

[0021]
This method can directly determine the amount of displacement when viewed from an image having no distortion or displacement (for example, FIG. 3A) from a displacement image such as an arbitrary image (for example, FIG. 3B). It becomes possible.
In addition, with respect to the problem that the calculated phase range is limited to [−π, π), the phase difference between the time phase 1 calculated using the images of the time phase 0 and the time phase 1 is By setting the phase difference of time phase 2 calculated using the images of 0 and time phase 2 to be [−π, π], phase calculation in any time phase, that is, displacement calculation becomes possible. . That is, if the coordinates of the frames t−1 and t giving the same mass point are p ^t−1 and p ^t , the equation (12) is expressed as

It expresses. Here, n ^t _i (p ^t ) represents the integer part of Δu ^t _i (p ^t ). In this way, it is necessary to use interpolation, if the period of mass ^{Delta] u} _t-1 1 frame t-1 of the same coordinates as ^{p t (p t-1)} is gentle, p ^'t-1 The phase Δu ^t−1 ₁ (p ′ ^t−1 ) determined by
[0022]
As described above, the present invention can be used to calculate the reliability of displacement by obtaining a small displacement and at the same time quantitatively obtaining the strength of the tag used to obtain the displacement.
The present invention is particularly suitable for use in the heart and can be used to evaluate the detection and quantification of ischemic heart disease and myocardial infarction.
In addition, it is possible to find a difference in tissue due to any combination of heart wall, cell wall, blood vessel, muscle, tendon, blood, cerebrospinal fluid, and the like, from the amplitude images related to the two tag directions.
As an MR tagging image to be used, an MRI pulse sequence selected from the group consisting of a gradient echo method, a spin echo method and a spiral method can be used.
In the above description, the displacement is indicated by an arrow as shown in FIG. 9, but the displacement can also be indicated by a displacement isoline.
It should be noted that the configuration of the present invention is not limited to the above-described embodiment, and various modifications can be made without departing from the scope of the present invention.
[0023]
【The invention's effect】
As described above, according to the present invention, the displacement (motion) of the object can be obtained from the tag with high accuracy by performing image processing on the MR tagging image. Also, the presence or remaining strength of the tag is digitized and used as a reliability of displacement (motion).
[Brief description of the drawings]
FIG. 1 is a flowchart showing processing in an embodiment of the present invention.
FIG. 2 is a schematic diagram showing an outline of processing.
FIG. 3 is a diagram illustrating an example of an image before being processed.
FIG. 4 is a diagram illustrating an example of a stripe image in two directions.
FIG. 5-1 is a diagram illustrating an example of an image obtained by multiplying a product with a fringe image having different phases.
FIG. 5-2 is a diagram illustrating a continuation of the image example obtained by taking the product of the fringe images having different phases.
6A is a diagram illustrating an image example after low-pass filter processing of the image of FIG.
FIG. 6-2 is a diagram illustrating an image example after low-pass filter processing of the image of FIG. 5-2;
FIG. 7 is a diagram illustrating an example of a phase image and an amplitude intensity image.
FIG. 8 is a diagram showing a mask image (a) created from an amplitude intensity image and an image (b) after mask processing for a phase image.
FIG. 9 is a diagram showing a phase image (a) obtained by cutting a heart portion and an image (b) showing displacement obtained therefrom.

Claims (11)

A displacement measurement system using MR tagging images,
An interference image generating means for generating an interference image of the MR tagging image and the reference fringe image and the plurality of fringe images having different phases from the reference fringe image;
Phase image generating means for generating a phase image from the interference image;
A displacement measuring system comprising displacement calculating means for obtaining a displacement from the phase image. The displacement measurement system according to claim 1,
The interference image generating means generates a interference image by taking a product of an MR tagging image and a fringe image and applying a low-pass filter. The displacement measuring system according to claim 1 or 2,
The displacement measurement system further comprises amplitude image generation means for generating an amplitude image from the interference image. The displacement measurement system according to claim 3,
Furthermore, it comprises a mask generating means for generating a mask image for extracting a highly reliable displacement portion from the amplitude image,
The displacement calculation system is characterized in that the displacement calculation means obtains a displacement after processing the phase image with a mask image. In the displacement measuring system according to any one of claims 1 to 4,
The reference stripe image is for each tag direction,
The fringe images with different phases are in each tag direction,
The phase image generation means and / or the amplitude image generation means generates a phase image and / or an amplitude image for each tag direction,
The displacement calculation system is characterized in that the displacement calculation means calculates a two-dimensional or three-dimensional displacement. A displacement measurement method using an MR tagging image,
An interference image generation step for generating an interference image between the MR tagging image and the reference fringe image and the plurality of fringe images having different phases from the reference fringe image;
A phase image generation step of generating a phase image from the interference image;
A displacement calculating method for obtaining a displacement from the phase image. In the displacement measuring method of Claim 6,
The displacement image generating step is characterized in that the interference image is generated by taking a product of an MR tagging image and a fringe image and applying a low-pass filter. In the displacement measuring method according to claim 6 or 7,
The displacement measuring method further includes an amplitude image generation step of generating an amplitude image from the interference image. The displacement measuring method according to claim 8, wherein
Furthermore, a mask generation step for generating a mask image for extracting a highly reliable displacement portion from the amplitude image,
The displacement calculating step is characterized in that the displacement is obtained after processing the phase image with a mask image. In the displacement measuring method in any one of Claims 6-9,
The reference stripe image is for each tag direction,
The fringe images with different phases are in each tag direction,
The phase image generation step and / or the amplitude image generation step generates a phase image and / or an amplitude image for each tag direction,
The displacement calculation method is characterized in that a two-dimensional or three-dimensional displacement is calculated. A computer program for causing a computer system to execute the displacement measuring method according to claim 6.

Images