JP3730143B2 - Ultrasonic image processing device - Google Patents

Description

標準偏差演算部６４では、シフトレジスタ６０から読み出されたデータそれぞれが合成回路８０にて平均値Ｍを減算された後、その差分値が乗算回路８２にて二乗される。さらに、それら各二乗値を加算回路８４で合計し、この合計値を除算回路８６にてデータ数Ｎで除算した後、平方根演算回路８８にて平方根を求め、次式で表される標準偏差値Ｓを出力する。
【００３２】
【数３】

図４は、オパシティ曲線選択部３４にてオパシティ曲線を選択するアルゴリズムを説明する模式図である。オパシティ曲線選択部３４は、平均値Ｍと標準偏差値Ｓとをパラメータとした二次元のルックアップテーブル（ＬＵＴ）９０を有している。ここで、平均値Ｍ及び標準偏差値Ｓそれぞれは複数の範囲に区分され、それらＭ及びＳの範囲に基づいてＭ−Ｓ平面が格子状に区分される。ＬＵＴ９０はこの各格子領域ごとにオパシティ曲線を対応付けるものである。図４に示す例では、Ｍ及びＳそれぞれが２つの範囲に区分され、Ｍ−Ｓ平面が４つの格子領域Ａ₁〜Ａ₄に区分される。一方、オパシティ曲線は同じエコーデータに対して大中小の異なる不透明度を与える３種類の曲線Ｃ₁〜Ｃ₃を用いる。
【００３３】
例えば、Ｍ，Ｓが共に小さい場合（すなわち領域Ａ₁に属する場合）、それらＭ，Ｓが得られたカーネルは、一様な液体部にあると推定される。そこで、そのカーネルを代表する対象ボクセルの不透明度には小さな値を与えることが好適であると考え得る。そこで、この場合には、オパシティ曲線Ｏｐ３を対応付けることができる。また、Ｍが大きくＳが小さい場合（すなわち領域Ａ₂に属する場合）、カーネルは、一様な固体部にあると推定される。この場合にも、対象ボクセルには小さな不透明度を与えることが好適であり、オパシティ曲線Ｏｐ３を対応付けることとする。一方、Ｍが小さくＳが大きい場合（すなわち領域Ａ₃に属する場合）、カーネルは、液体部と固体部との境界にまたがっていると推定される。そこで、そのカーネルを代表する対象ボクセルの不透明度には大きな値を与えることが好適であると考え得る。そこで、この場合には、オパシティ曲線Ｏｐ１を対応付けることができる。Ｍ，Ｓが共に大きい場合（すなわち領域Ａ₄に属する場合）、カーネルは、固体部内の境界にまたがっていると推定され、対象ボクセルには、オパシティ曲線Ｏｐ１及びＯｐ３の中間の不透明度を与えるオパシティ曲線Ｏｐ２を対応付ける。
【００３４】
図５は、統計演算部２８、オパシティ曲線選択部３４及びオパシティ曲線番号メモリ部３６の処理を説明するものであり、１つの透視線に対するオパシティ曲線の割り当て処理を説明する模式図である。ここでは、カーネルサイズＮを３に設定する。統計演算部２８のシフトレジスタ６０には、透視線上のボクセル列１００のエコーデータが順に入力される。例えば、ボクセルＶ₁，Ｖ₂，Ｖ₃，…Ｖ_mの順にそれぞれのエコーデータが入力されるとする。統計演算部２８は、例えば、シフトレジスタ６０にＶ₃のエコーデータが入力されたタイミングにて、ボクセルＶ₁〜Ｖ₃からなるカーネル１０-1に対して平均値Ｍ及び標準偏差値Ｓを算出する。オパシティ曲線選択部３４は、そのＭ，Ｓに基づいてＬＵＴ９０を検索し、カーネルの中心ボクセルＶ₂のオパシティ曲線を例えばＯｐ３と定める。その結果、オパシティ曲線番号メモリ部３６には、ボクセルＶ₂に対するオパシティ曲線番号Ｏｐ(Ｖ₂)として“３”が格納される。統計演算部２８は、シフトレジスタ６０にボクセルＶ₄のエコーデータが入力されると、カーネルをシフトさせる。そして、カーネル１０-1に対すると同様の処理が行われる。例えば、ボクセルＶ₂〜Ｖ₄からなるカーネル１０-2に対してオパシティ曲線Ｏｐ１が選択されると、オパシティ曲線番号メモリ部３６には、カーネル１０-2の中心ボクセルＶ₃に対するオパシティ曲線番号Ｏｐ(Ｖ₃)として“１”が格納される。以下、同様に新しいボクセルのエコーデータが入力される度に、カーネルをシフトさせ、そのカーネルの中心ボクセルのオパシティ曲線番号が決定されてオパシティ曲線番号メモリ部３６に格納される。
【００３５】
なお、透視線の端部のボクセルＶ₁，Ｖ_mに対するオパシティ曲線は、例えば隣接するボクセルＶ₂、Ｖ_m-1のオパシティ曲線に等しいとする等の処理により定めることが可能である。
【００３６】
図６は、上述した本装置のボリュームレンダリング処理をまとめたフロー図である。ユーザはトラックボール等の入力機器３２を操作してカーネルサイズを指定する（Ｓ１５０）。統計演算部２８は、各透視線に対する初期状態として、その透視線に対応する超音波ビームの先頭にカーネルを位置させる（Ｓ１５５）。統計演算部２８は、設定したカーネル内のエコーデータの平均値Ｍ及び標準偏差値Ｓを計算する（Ｓ１６０）。得られたＭ及びＳに基づいて、オパシティ曲線選択部３４は１つのオパシティ曲線を選択し、そのオパシティ曲線番号を当該カーネルの代表ボクセル（例えば中心ボクセル）に割り当てる。割り当てられたオパシティ曲線番号はオパシティ曲線番号メモリ部３６に記録される（Ｓ１６５）。統計演算部２８は透視線に沿ってカーネルを順に移動させ（Ｓ１７０）、上記ステップＳ１６０及びＳ１６５が繰り返される。カーネルがある透視線の終わりまで移動した時点にて（Ｓ１７５）、オパシティ曲線番号メモリ部３６にはその透視線上の各ボクセルのオパシティ曲線番号が格納されている。ボリュームレンダリング演算部３８は、ラインメモリ部２６に格納されている１透視線分の各ボクセルのエコーデータを用いたレイキャスティング処理を、オパシティ曲線番号メモリ部３６に格納されている同じ透視線の各ボクセルのオパシティ曲線番号を参照しながら実行し、ボリュームレンダリングを行う（Ｓ１８０）。ステップＳ１８０により得られる当該透視線に対応する画素値はフレームバッファメモリ４０に記録される（Ｓ１８５）。以上のステップＳ１５５〜Ｓ１８５は、透視線を移動させて繰り返され（Ｓ１９０）、全ての透視線上でステップＳ１５５〜Ｓ１８５のデータ処理が終わると（Ｓ１９５）、各透視線に対応する画素値がフレームバッファメモリ４０に格納される。このフレームバッファメモリ４０に格納された超音波画像データが表示部４２を介してユーザに提示される（Ｓ２００）。
【００３７】
図７は、本装置の動作を説明するタイミング図である。図７において、期間２１０-1では第１の透視線に対するオパシティ曲線番号の決定処理（図６に示すステップＳ１５５〜Ｓ１６５）が実行され、これに引き続く期間２１２-1では第１の透視線に対するボリュームレンダリング演算部３８での演算（図６に示すステップＳ１８０〜Ｓ１８５）が実行される。また、第２の透視線に対するオパシティ曲線番号の決定処理の期間２１０-2は、第１の透視線に対するボリュームレンダリング演算の期間２１２-1にオーバーラップして設定することができる。ちなみに、ラインメモリ部２６に格納されたエコーデータは、ボリュームレンダリング演算の進行につれて、順に先頭部分から不要になり、その部分を次の透視線、すなわち次の超音波ビームについて順に送受信部２２から出力されるエコーデータで上書きすることができるので、ラインメモリ部２６は１透視線分の容量で足りる。第２の透視線に対するオパシティ曲線番号の決定処理期間２１０-2に引き続く期間２１２-2にて第２の透視線に対するボリュームレンダリング演算が実行される。このようにして、各透視線についての処理が順に実行される。
【００３８】
［実施形態２］
図８は、本発明の第２の実施形態である超音波診断装置の概略のブロック構成図である。上記第１の実施形態は、エコーデータ及びオパシティ曲線番号を１透視線分格納した上で処理を行うものであり、そのためにラインメモリ部２６及びオパシティ曲線番号メモリ部３６を備えている。これに対して本実施形態は、受信信号に対応して送受信部２２から順次出力されるエコーデータに対して、オパシティ曲線の決定及びレイキャスティングを逐次処理にて行い、ボリュームレンダリングされた超音波画像を生成するものである。以下の説明において、上記第１の実施形態と同様の機能を有する構成要素には同一の符号を付し、説明の簡略化を図る。
【００３９】
本装置は、統計演算部２８が或る対象ボクセルに対応したカーネルについて平均値Ｍ及び標準偏差値Ｓを出力すると、オパシティ曲線選択部３００はＬＵＴ９０を検索してそのＭ及びＳに対応するオパシティ曲線を決定し、その情報を直ちにボリュームレンダリング演算部３０２へ出力する。遅延時間部３０４は、或るボクセルに対応するオパシティ曲線の情報がオパシティ曲線選択部３００からボリュームレンダリング演算部３０２に与えられるタイミングに合わせて、当該ボクセルのエコーデータをボリュームレンダリング演算部３０２に供給するために設けられている。すなわち、遅延時間部３０４は統計演算部２８及びオパシティ曲線選択部３００での処理時間に応じて、送受信部２２からボリュームレンダリング演算部３０２へのエコーデータの入力を遅延させる。
【００４０】
オパシティ曲線選択部３００はオパシティ曲線に関する情報として、上記第１の実施形態と同様にオパシティ曲線番号をボリュームレンダリング演算部３０２に与えることができる。この場合、エコーデータに対応する不透明度関数値を求める処理はボリュームレンダリング演算部３０２にて行われる。一方、オパシティ曲線選択部３００は、オパシティ曲線を定め、それを用いてエコーデータに対応する不透明度を計算し、この不透明度をボリュームレンダリング演算部３０２に与えるように構成することもできる。
【００４１】
図９は、上述した本装置のボリュームレンダリング処理をまとめたフロー図である。この図で上記第１の実施形態の図６と同様のステップには同一の符号を付している。本装置の処理が第１の実施形態の装置と異なる点は、レイキャスティング処理の実行タイミングである。すなわち第１の実施形態では透視線１本分のオパシティ曲線番号が得られた後に、ボリュームレンダリングのレイキャスティング処理が実行された（Ｓ１８０）。これに対し本装置では、各対象ボクセルについてオパシティ曲線が決定されると、それに基づいて当該ボクセルに対するレイキャスティング処理が逐次実行される（Ｓ４００）。その結果、カーネルが透視線の終わりまで移動した時点で、その透視線についてのレイキャスティング処理は完了しているので、ステップＳ１８０に相当する処理は不要であり、当該透視線に対応する画素値をフレームバッファメモリ４０に格納する処理（Ｓ１８５）が直ちに実行される。
【００４２】
図１０は、本装置の動作を説明するタイミング図である。図１０において、期間４１０-1では第１の透視線の第１のボクセルに対するオパシティ曲線の決定処理が実行され、これに引き続く期間４１２-1では当該ボクセルに対するボリュームレンダリング演算部３０２での演算が実行される。また、第１の透視線の第２のボクセルに対するオパシティ曲線決定処理の期間４１０-2は、第１のボクセルに対するボリュームレンダリング演算の期間４１２-1にオーバーラップして設定することができる。以下、同様にして、オパシティ曲線の決定とボリュームレンダリング演算とがボクセル単位で逐次実行される。また、期間４１０-kにて実行される第１の透視線の最後のボクセルであるｋ番目のボクセルのオパシティ曲線の決定が完了すると、第２の透視線についてのボクセルごとの逐次処理が開始される。なお、図１０において、期間４１４-i及び４１６-iがそれぞれ第２の透視線のｉ番目のボクセルに対するオパシティ曲線決定期間とボリュームレンダリング演算期間とを表している。
【００４３】
【発明の効果】
本発明の超音波画像処理装置によれば、三次元空間にて得られたエコーデータのそれぞれの位置での統計的性質に基づいて、各ボクセルに不透明度が割り当てられるので、不透明度の設定の自動化が図られ、ひいてはユーザの操作が簡略化され、検査時間が短縮される。
【図面の簡単な説明】
【図１】 エコーデータが得られる三次元空間と、ボリュームレンダリングの透視線及び投影面と、複数の統計量が求められるサンプル領域であるカーネルとを示す模式図である。
【図２】 本発明の第１の実施形態である超音波診断装置の概略のブロック構成図である。
【図３】 統計演算部の概略のブロック構成図である。
【図４】 オパシティ曲線選択部にてオパシティ曲線を選択するアルゴリズムを説明する模式図である。
【図５】 １つの透視線に対するオパシティ曲線の割り当て処理を説明する模式図である。
【図６】 第１の実施形態の超音波診断装置のボリュームレンダリング処理の概略のフロー図である。
【図７】 第１の実施形態の超音波診断装置のボリュームレンダリング処理の動作を説明するタイミング図である。
【図８】 本発明の第２の実施形態である超音波診断装置の概略のブロック構成図である。
【図９】 第２の実施形態の超音波診断装置のボリュームレンダリング処理の概略のフロー図である。
【図１０】 第２の実施形態の超音波診断装置のボリュームレンダリング処理の動作を説明するタイミング図である。
【符号の説明】
２０ 超音波探触子、２２ 送受信部、２６ ラインメモリ部、２８ 統計演算部、３０ カーネルサイズ発生部、３４，３００ オパシティ曲線選択部、３６ オパシティ曲線番号メモリ部、３８，３０２ ボリュームレンダリング演算部、４０ フレームバッファメモリ、４２ 表示部、３０４ 遅延時間部。[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an ultrasonic image processing apparatus that generates a three-dimensional image by volume rendering from echo data obtained in a three-dimensional space, and more particularly to setting opacity in volume rendering processing to improve image quality.
[0002]
[Prior art]
In order to generate a three-dimensional image from echo data in a three-dimensional space captured by an ultrasonic diagnostic apparatus, a volume rendering method is used. In the volume rendering method, a three-dimensional space from which echo data is obtained is divided into small volume elements called voxels, and the echo data is assigned to each voxel. For rendering, the ray casting method is mainly used. In ray casting, a parallel perspective line is set corresponding to each pixel of an image, and a calculation similar to the generation and absorption of light is performed along each perspective line in each voxel through which it passes. That is, in the voxels arranged along the perspective line, the light coming from the previous voxel is transmitted according to the transparency of the voxel, and the light emission according to the echo data of the voxel is added to the voxel behind the voxel. Passed. Here, the light emission amount is calculated using a parameter called opacity. The opacity at the i-th voxel along the perspective line is expressed as α_iThen the transparency is (1-α_i) And echo data_iIn other words, the input light amount C passed from the voxel before the above-described voxel._INiAnd the output light quantity C passed to the voxel behind_OUTiIs expressed by the following equation.
[0003]
[Expression 1]
C_OUTi= (1-α_i) C_INi+ Α_i・ E_i                      ……… (1)
This calculation is performed along the perspective line while integrating the opacity. When the integrated value of opacity becomes 1 (that is, the transparency is 0), the calculation for the perspective line is terminated. And the output light C from the last voxel_OUTIs a pixel value corresponding to the perspective line. In volume rendering, an image is projected on a two-dimensional screen by performing such ray casting for each perspective line. It is known that the ultrasonic image obtained in this manner has both a personality as a fluoroscopic image and a personality as a stereoscopic image. That is, the tissue in the living body can be expressed in a watermark like an X-ray photograph, while it can be expressed with a sense of depth like an ultrasonic three-dimensional image. Thus, for example, simultaneous observation of the surface and inside of the fetus can be performed, and a three-dimensional grasp of the tissue can be easily performed in disease diagnosis.
[0004]
The image quality of the ultrasonic image changes according to the opacity, and for example, it is possible to enhance the transparency or the stereoscopic effect. Alternatively, the tissue surface can be emphasized, or the inside of the tissue can be emphasized.
[0005]
Conventionally, opacity α_iIs the echo data e_iIs defined as a function whose parameter is an echo value e._iOpacity α for increasing_iNonlinear functions with increasing values are preferred. Opacity α_iVoxel echo data e_iIn addition, there is a method of assigning in consideration of a gradient value based on echo data of surrounding voxels.
[0006]
The opacity is adjusted by the user operating the dial or the like of the ultrasonic diagnostic apparatus, and the echo data e_iThis is done by changing the opacity function with a parameter.
[0007]
[Problems to be solved by the invention]
However, this function is uniformly applied to all voxel echo data, that is, the entire three-dimensional space. That is, by changing the function, for example, even if a desired image quality such as transparency or stereoscopic effect is achieved in a certain area, the desired image quality is not necessarily obtained in another area. For this reason, it is difficult for the user to optimize the overall image quality of the image, and it takes a lot of time to adjust the image, and there is a problem in that a skilled technique is required.
[0008]
The present invention has been made to solve the above-described problems, and provides an ultrasonic image processing apparatus that can automate the setting of opacity, simplify user operations, and shorten inspection time. Objective.
[0009]
[Means for Solving the Problems]
  The ultrasonic image processing apparatus according to the present invention, for each target voxel, a sample region setting means for setting a sample region composed of a plurality of voxels in a three-dimensional space with reference to the target voxel,For each target voxel, the target voxel is used as a reference.Statistical calculation means for obtaining a set of statistics consisting of a plurality of statistics having different properties from each other using echo data of each voxel in the sample region;A plurality of opacity functions that define the correspondence between the echo data and the opacity are set in advance according to the statistic set, and among the plurality of opacity functions, the opacity function in the sample region based on the target voxel is set. Function selecting means for selecting an opacity function corresponding to the target voxel based on the statistic set, and a function for obtaining opacity corresponding to echo data of the target voxel based on the selected opacity function Computing means;HaveIn a preferred aspect of the present invention, the plurality of opacity functions include an opacity function corresponding to the liquid portion. In a preferred aspect of the present invention, the plurality of opacity functions include an opacity function corresponding to a solid part. In a preferred aspect of the present invention, the plurality of opacity functions include an opacity function corresponding to a boundary between the liquid part and the solid part. In a preferred aspect of the present invention, the plurality of opacity functions include an opacity function corresponding to a boundary in the solid part.
[0010]
According to the present invention, the opacity of the target voxel is determined according to the statistic set obtained from the echo data in the sample region corresponding to the target voxel. For example, even if the echo data of each of two target voxels has the same value, different opacity can be assigned if the values of the statistics that make up the statistics set are different between the target voxels. That is, it is possible to assign different opacity to different places in the three-dimensional space, and the image quality of the entire image can be made suitable. It is possible to predetermine a suitable opacity corresponding to the statistical properties of the echo data, and based on the correspondence between the statistical properties and the opacity, the opacity determination means automatically determines from the statistics set. It is possible to determine the opacity.
[0011]
In a preferred aspect of the present invention, the ultrasonic image processing apparatus is characterized in that each of the plurality of statistics is an n-order product factor having a predetermined order n. For example, the nth-order product rate is a set of an average of echo data that is the primary product rate and a variance of echo data that is the secondary product rate.
[0012]
Another preferred aspect of the present invention is the ultrasonic image processing apparatus, wherein the sample region includes the target voxel.
[0013]
Another preferred aspect of the present invention is the ultrasonic image processing apparatus, wherein the sample region is composed of a plurality of voxels arranged in a line in the perspective line direction.
[0014]
In another ultrasonic image processing apparatus according to the present invention, the opacity determination means includes a function selection means for selecting an opacity function that defines a correspondence relationship between echo data and opacity based on the statistic set. Function calculating means for obtaining opacity corresponding to echo data of the target voxel based on the selected opacity function.
[0015]
According to the present invention, the opacity function using the echo data as a parameter is switched according to the statistic set. That is, it is possible to obtain the opacity corresponding to the echo data with a suitable function for each voxel, instead of a uniform opacity function for the three-dimensional space.
[0016]
DETAILED DESCRIPTION OF THE INVENTION
Next, embodiments of the present invention will be described with reference to the drawings.
[0017]
[Embodiment 1]
FIG. 1 is a schematic diagram showing a three-dimensional space in which echo data is obtained, a volume rendering perspective line and a projection plane, and a kernel that is a sample region for which a plurality of statistics are obtained.
[0018]
The three-dimensional space 2 from which echo data is obtained is divided into voxels, and echo data is assigned to each voxel. In the three-dimensional space 2, a plurality of parallel perspective lines 8 corresponding to each pixel 6 of the image formed on the projection surface 4 are set. For each voxel on each perspective line 8, a calculation based on equation (1) is performed. The opacity α used in this calculation is determined based on the statistic of echo data in the kernel 10 set corresponding to the target voxel on which the calculation is performed. The kernel 10 has, for example, a rectangular parallelepiped shape in which Nx, Ny, and Nz voxels are arranged in the height, width, and depth directions, and the depth direction z coincides with the line-of-sight direction. The kernel 10 includes a plurality of voxels in order to calculate a statistic, and generally, the kernel 10 is set so that a predetermined position, for example, a central voxel becomes a target voxel. Here, in order to simplify the processing, a case where Nx = Ny = 1, that is, the kernel 10 is constituted by a plurality of voxels arranged in a line along the perspective line 8 will be described.
[0019]
Ray casting along the perspective line 8 is performed using the opacity α determined for each voxel, and the pixel value of the pixel 6 is determined. By performing this ray casting process for each perspective line 8, a projection image on the projection plane 4 is generated.
[0020]
FIG. 2 is a schematic block diagram of the ultrasonic diagnostic apparatus according to the first embodiment of the present invention. The ultrasonic probe 20 has, for example, a linear array type ultrasonic transducer, and a scanning plane is formed by electronic scanning (linear scanning, sector scanning) of the ultrasonic transducer. By, for example, mechanically scanning the scanning surface in the normal direction of the surface, echo data in a three-dimensional space is acquired. Here, the mechanical scanning of the ultrasonic probe 20 is performed by a drive unit (not shown).
[0021]
The transmission / reception unit 22 supplies a transmission signal to the ultrasonic probe 20 and processes a reception signal output from the ultrasonic probe 20. The transmission / reception unit 22 converts the received signal into a digital signal (echo data) and then outputs the digital signal to the line memory unit 26 and the statistical calculation unit 28.
[0022]
The line memory unit 26 stores echo data for one perspective line. In this apparatus, since the perspective line coincides with the ultrasonic beam, the echo data output from the transmission / reception unit 22 can be stored in the line memory unit 26 without coordinate conversion.
[0023]
The statistical calculation unit 28 calculates a predetermined statistical quantity for each kernel. Specifically, the statistical calculation unit 28 calculates a statistic with respect to continuous Nz echo data output from the transmission / reception unit 22. Here, the statistical calculation unit 28 calculates an average value M, which is a primary product rate, and a standard deviation value S, which is a secondary product rate, as statistics. Note that the statistical calculation unit 28 calculates higher-order product moments, for example, data distribution distortion and other statistical quantities that are third-order product moments instead of or in addition to the average value M and the standard deviation value S. Can be configured to.
[0024]
The kernel size generation unit 30 receives a value N designated by the user by operating the input device 32 such as a trackball, and sets this in the statistical calculation unit 28 as a size Nz of the perspective direction of the kernel. That is, Nz = N is set.
[0025]
The opacity curve selection unit 34 selects one of a plurality of preset opacity curves based on the combination of the average value M and the standard deviation value S calculated for each kernel by the statistical calculation unit 28. And assigned to the representative voxel included in the kernel. The opacity curve is an opacity function that gives opacity with echo data as an argument. As will be described later, in response to the statistical calculation unit 28 moving the kernel along the perspective line, the opacity curve selection unit 34 sequentially determines an opacity curve corresponding to each voxel on the perspective line.
[0026]
The opacity curve is identified by a number, and the opacity curve number memory unit 36 stores the opacity curve number (opacity curve number) assigned to each voxel on one perspective line.
[0027]
The volume rendering operation unit 38 stores echo data of each voxel for one perspective line stored in the line memory unit 26 and the opacity of each voxel for one perspective line stored in the opacity curve number memory unit 36 for the same perspective line. Using the curve number, the ray casting process represented by equation (1) is performed to calculate the pixel value corresponding to the perspective line. In other words, the opacity function specified by the opacity curve number is stored in the volume rendering operation unit 38, and when the echo data and the opacity curve number for a certain voxel are obtained, the opacity function specified by the opacity curve number is obtained. The echo data is given as an argument to the transparency function, and the opacity corresponding to the voxel is calculated. Then, the volume rendering calculation unit 38 performs the calculation of equation (1) using this opacity. The volume rendering operation unit 38 determines a corresponding pixel value for each ultrasonic beam and sequentially stores it in the frame buffer memory 40. As a result, the frame buffer memory 40 stores image data for one ultrasonic image formed by the volume rendering process.
[0028]
The image data read from the frame buffer memory 40 is displayed on the display unit 42.
[0029]
As described above, the operation of each unit is controlled by the control unit 44. For example, the control unit 44 supplies a transmission synchronization signal to the transmission / reception unit 22, and supplies an R / W signal for controlling reading / writing of echo data or opacity curve number to the line memory unit 26 and the opacity curve number memory unit 36. . In addition, when the opacity curve number for one perspective line is stored in the opacity curve number memory unit 36, the control unit 44 gives a calculation start signal to the volume rendering calculation unit 38.
[0030]
Hereinafter, a more detailed description of the main part will be given. FIG. 3 is a schematic block diagram of the statistical calculation unit. The statistical calculation unit 28 sequentially inputs the echo data input from the transmission / reception unit 22 to the shift register 60. The input echo data is forwarded through the shift register 60. Of the predetermined number of echo data held in the shift register 60, N consecutive data (e₁, E₂, ..., e_NIs read by the average calculation unit 62 and the standard deviation calculation unit 64, and the average value and the standard deviation value are calculated based on the N pieces of echo data. The average calculator 62 sums the echo data read from the shift register 60 by the adder circuit 70, then divides this total value by the number of data N added by the divider circuit 72, and is expressed by the following equation. The average value M is output.
[0031]
[Expression 2]

In the standard deviation calculation unit 64, the data read from the shift register 60 is subtracted from the average value M by the synthesis circuit 80, and then the difference value is squared by the multiplication circuit 82. Further, each square value is summed by the adder circuit 84, and the sum value is divided by the number of data N by the divider circuit 86. Then, the square root is obtained by the square root arithmetic circuit 88, and the standard deviation value expressed by the following equation is obtained. S is output.
[0032]
[Equation 3]

FIG. 4 is a schematic diagram for explaining an algorithm for selecting an opacity curve by the opacity curve selection unit 34. The opacity curve selection unit 34 has a two-dimensional lookup table (LUT) 90 using the average value M and the standard deviation value S as parameters. Here, each of the average value M and the standard deviation value S is divided into a plurality of ranges, and the MS plane is divided into a lattice shape based on the ranges of the M and S. The LUT 90 associates an opacity curve for each lattice area. In the example shown in FIG. 4, M and S are each divided into two ranges, and the MS plane has four lattice regions A.₁~ A_FourIt is divided into. On the other hand, the opacity curve has three types of curves C that give different opacity of large, medium and small to the same echo data.₁~ C_ThreeIs used.
[0033]
For example, when both M and S are small (ie, region A₁), It is presumed that the kernel from which M and S are obtained is in a uniform liquid part. Therefore, it can be considered that it is preferable to give a small value to the opacity of the target voxel representing the kernel. Therefore, in this case, the opacity curve Op3 can be associated. Further, when M is large and S is small (that is, region A)₂The kernel is assumed to be in a uniform solid part. Also in this case, it is preferable to give a small opacity to the target voxel, and the opacity curve Op3 is associated. On the other hand, when M is small and S is large (ie, region A_ThreeThe kernel is assumed to straddle the boundary between the liquid part and the solid part. Therefore, it can be considered that it is preferable to give a large value to the opacity of the target voxel representing the kernel. Therefore, in this case, the opacity curve Op1 can be associated. When both M and S are large (ie, region A_FourThe kernel is presumed to straddle the boundary in the solid part, and the target voxel is associated with an opacity curve Op2 that gives an opacity intermediate between the opacity curves Op1 and Op3.
[0034]
FIG. 5 is a schematic diagram illustrating the processing of the statistical calculation unit 28, the opacity curve selection unit 34, and the opacity curve number memory unit 36, and illustrates the processing of assigning an opacity curve to one perspective line. Here, the kernel size N is set to 3. The echo data of the voxel string 100 on the perspective line is sequentially input to the shift register 60 of the statistical calculation unit 28. For example, voxel V₁, V₂, V_Three, ... V_mAssume that the respective echo data are input in the order of. For example, the statistical calculation unit 28 adds V to the shift register 60._ThreeVoxel V at the timing when the echo data of₁~ V_ThreeAn average value M and a standard deviation value S are calculated for a kernel 10-1 consisting of The opacity curve selection unit 34 searches the LUT 90 based on the M and S, and the center voxel V of the kernel.₂The opacity curve is defined as, for example, Op3. As a result, the opacity curve number memory unit 36 stores the voxel V.₂Opacity curve number for Op (V₂) Is stored as “3”. The statistical calculation unit 28 adds the voxel V to the shift register 60._FourWhen the echo data is input, the kernel is shifted. Then, the same processing as that for the kernel 10-1 is performed. For example, voxel V₂~ V_FourWhen the opacity curve Op1 is selected with respect to the kernel 10-2 consisting of, the opacity curve number memory unit 36 stores the center voxel V of the kernel 10-2._ThreeOpacity curve number for Op (V_Three) Is stored as “1”. Similarly, every time echo data of a new voxel is input, the kernel is shifted, and the opacity curve number of the center voxel of the kernel is determined and stored in the opacity curve number memory unit 36.
[0035]
The voxel V at the end of the perspective line₁, V_mThe opacity curve for is, for example, the adjacent voxel V₂, V_m-1It can be determined by a process such as equaling to the opacity curve.
[0036]
FIG. 6 is a flowchart summarizing the volume rendering processing of the apparatus described above. The user designates a kernel size by operating the input device 32 such as a trackball (S150). The statistical calculation unit 28 positions the kernel at the head of the ultrasonic beam corresponding to each perspective line as an initial state for each perspective line (S155). The statistical calculation unit 28 calculates the average value M and the standard deviation value S of the echo data in the set kernel (S160). Based on the obtained M and S, the opacity curve selection unit 34 selects one opacity curve and assigns the opacity curve number to the representative voxel (for example, the center voxel) of the kernel. The assigned opacity curve number is recorded in the opacity curve number memory unit 36 (S165). The statistical calculation unit 28 sequentially moves the kernel along the perspective line (S170), and steps S160 and S165 are repeated. When the kernel moves to the end of a perspective line (S175), the opacity curve number memory unit 36 stores the opacity curve number of each voxel on the perspective line. The volume rendering calculation unit 38 performs ray casting processing using echo data of each voxel for one perspective line stored in the line memory unit 26 for each of the same perspective lines stored in the opacity curve number memory unit 36. The volume rendering is performed by referring to the opacity curve number of the voxel (S180). The pixel value corresponding to the perspective line obtained in step S180 is recorded in the frame buffer memory 40 (S185). The above steps S155 to S185 are repeated by moving the perspective line (S190). When the data processing of steps S155 to S185 is completed on all the perspective lines (S195), the pixel values corresponding to the respective perspective lines are converted into frame buffers. Stored in the memory 40. The ultrasonic image data stored in the frame buffer memory 40 is presented to the user via the display unit 42 (S200).
[0037]
FIG. 7 is a timing chart for explaining the operation of the present apparatus. In FIG. 7, the opacity curve number determination process (steps S155 to S165 shown in FIG. 6) for the first perspective line is executed in the period 210-1, and the volume for the first perspective line is subsequently executed in the period 212-1. Calculations in the rendering calculation unit 38 (steps S180 to S185 shown in FIG. 6) are executed. Further, the period 210-2 for determining the opacity curve number for the second perspective line can be set so as to overlap with the period 212-1 for the volume rendering operation for the first perspective line. Incidentally, the echo data stored in the line memory unit 26 becomes unnecessary from the head part in order as the volume rendering operation proceeds, and the part is output from the transmitting / receiving part 22 in order for the next perspective line, that is, the next ultrasonic beam. Therefore, the line memory unit 26 is sufficient for the capacity of one perspective line. The volume rendering operation for the second perspective line is executed in a period 212-2 following the opacity curve number determination process period 210-2 for the second perspective line. In this way, the processing for each perspective line is executed in order.
[0038]
[Embodiment 2]
FIG. 8 is a schematic block diagram of an ultrasonic diagnostic apparatus according to the second embodiment of the present invention. In the first embodiment, the echo data and the opacity curve number are stored after one perspective line is stored, and the line memory unit 26 and the opacity curve number memory unit 36 are provided for this purpose. On the other hand, in the present embodiment, volume-rendered ultrasonic images are obtained by sequentially determining opacity curves and ray casting for echo data sequentially output from the transmission / reception unit 22 corresponding to received signals. Is generated. In the following description, components having the same functions as those in the first embodiment are denoted by the same reference numerals, and the description is simplified.
[0039]
When the statistical calculation unit 28 outputs the average value M and the standard deviation value S for the kernel corresponding to a certain target voxel, the opacity curve selection unit 300 searches the LUT 90 and the opacity curve corresponding to the M and S. And the information is immediately output to the volume rendering operation unit 302. The delay time unit 304 supplies the echo data of the voxel to the volume rendering calculation unit 302 in accordance with the timing at which the opacity curve information corresponding to a certain voxel is given from the opacity curve selection unit 300 to the volume rendering calculation unit 302. It is provided for. That is, the delay time unit 304 delays the input of echo data from the transmission / reception unit 22 to the volume rendering calculation unit 302 according to the processing time in the statistical calculation unit 28 and the opacity curve selection unit 300.
[0040]
The opacity curve selection unit 300 can give the opacity curve number to the volume rendering calculation unit 302 as information about the opacity curve, as in the first embodiment. In this case, the processing for obtaining the opacity function value corresponding to the echo data is performed by the volume rendering operation unit 302. On the other hand, the opacity curve selection unit 300 may be configured to determine an opacity curve, calculate opacity corresponding to echo data using the opacity curve, and give the opacity to the volume rendering operation unit 302.
[0041]
FIG. 9 is a flowchart summarizing the volume rendering processing of the apparatus described above. In this figure, the same steps as those in FIG. 6 of the first embodiment are denoted by the same reference numerals. The difference between the processing of this apparatus and the apparatus of the first embodiment is the execution timing of the ray casting process. That is, in the first embodiment, after the opacity curve number for one perspective line is obtained, the volume rendering ray casting process is executed (S180). In contrast, in the present apparatus, when an opacity curve is determined for each target voxel, ray casting processing for the voxel is sequentially executed based on the determined opacity curve (S400). As a result, when the kernel moves to the end of the perspective line, the ray casting process for the perspective line is completed, so the processing corresponding to step S180 is unnecessary, and the pixel value corresponding to the perspective line is changed. The process of storing in the frame buffer memory 40 (S185) is immediately executed.
[0042]
FIG. 10 is a timing chart for explaining the operation of the present apparatus. In FIG. 10, in the period 410-1, the opacity curve determination process for the first perspective line for the first voxel is executed, and in the subsequent period 412-1 the calculation in the volume rendering calculation unit 302 is executed for the voxel. Is done. Further, the period 410-2 of the opacity curve determination process for the second voxel of the first perspective line can be set so as to overlap with the period 412-1 of the volume rendering operation for the first voxel. Hereinafter, similarly, determination of an opacity curve and volume rendering calculation are sequentially performed in units of voxels. When the determination of the opacity curve of the k-th voxel that is the last voxel of the first perspective line executed in the period 410-k is completed, sequential processing for each voxel with respect to the second perspective line is started. The In FIG. 10, periods 414-i and 416-i represent an opacity curve determination period and a volume rendering calculation period for the i-th voxel of the second perspective line, respectively.
[0043]
【The invention's effect】
According to the ultrasonic image processing apparatus of the present invention, opacity is assigned to each voxel based on statistical properties at each position of echo data obtained in a three-dimensional space. Automation is achieved, and the operation of the user is simplified, and the inspection time is shortened.
[Brief description of the drawings]
FIG. 1 is a schematic diagram showing a three-dimensional space from which echo data can be obtained, a volume rendering perspective line and projection plane, and a kernel that is a sample area for which a plurality of statistics are obtained.
FIG. 2 is a schematic block configuration diagram of the ultrasonic diagnostic apparatus according to the first embodiment of the present invention.
FIG. 3 is a schematic block diagram of a statistical calculation unit.
FIG. 4 is a schematic diagram illustrating an algorithm for selecting an opacity curve by an opacity curve selection unit.
FIG. 5 is a schematic diagram illustrating an opacity curve assignment process for one perspective line.
FIG. 6 is a schematic flowchart of volume rendering processing of the ultrasonic diagnostic apparatus according to the first embodiment.
FIG. 7 is a timing chart for explaining the operation of the volume rendering process of the ultrasonic diagnostic apparatus according to the first embodiment.
FIG. 8 is a schematic block configuration diagram of an ultrasonic diagnostic apparatus according to a second embodiment of the present invention.
FIG. 9 is a schematic flowchart of volume rendering processing of the ultrasonic diagnostic apparatus according to the second embodiment.
FIG. 10 is a timing diagram illustrating an operation of volume rendering processing of the ultrasonic diagnostic apparatus according to the second embodiment.
[Explanation of symbols]
20 ultrasonic probe, 22 transmission / reception unit, 26 line memory unit, 28 statistical calculation unit, 30 kernel size generation unit, 34,300 opacity curve selection unit, 36 opacity curve number memory unit, 38,302 volume rendering calculation unit, 40 frame buffer memory, 42 display section, 304 delay time section.

Claims (9)

Assigning opacity to each voxel arranged in a three-dimensional space, calculating a light emission amount according to echo data of each voxel and a transmitted light amount in each voxel along a perspective line based on the opacity, In an ultrasonic image processing apparatus that generates a three-dimensional image using echo data obtained in the three-dimensional space by volume rendering to obtain a pixel value of a pixel corresponding to the perspective line,
Sample region setting means for setting a sample region composed of a plurality of the voxels in the three-dimensional space with respect to the target voxel for each target voxel;
For each target voxel , using the echo data of each voxel in the sample region with reference to the target voxel, statistical calculation means for obtaining a statistic set consisting of a plurality of statistics having different properties;
A plurality of opacity functions that define the correspondence between the echo data and the opacity are set in advance according to the statistic set, and among the plurality of opacity functions, the opacity function in the sample region based on the target voxel is set. Function selection means for selecting an opacity function corresponding to the target voxel based on the statistic set;
Based on the selected opacity function, function calculation means for obtaining opacity corresponding to the echo data of the target voxel;
An ultrasonic image processing apparatus comprising: The ultrasonic image processing apparatus according to claim 1,
The plurality of opacity functions include an opacity function corresponding to the liquid part.
An ultrasonic image processing apparatus. The ultrasonic image processing apparatus according to claim 2,
The plurality of opacity functions include an opacity function corresponding to a solid part.
An ultrasonic image processing apparatus. The ultrasonic image processing apparatus according to claim 3.
The plurality of opacity functions include an opacity function corresponding to a boundary between the liquid part and the solid part.
An ultrasonic image processing apparatus. The ultrasonic image processing apparatus according to claim 4.
The plurality of opacity functions include an opacity function corresponding to a boundary in the solid part.
An ultrasonic image processing apparatus. The ultrasonic image processing apparatus according to claim 5.
The ultrasonic image processing apparatus according to claim 1, wherein each of the plurality of statistics is an n-order product factor having a predetermined order n. The ultrasonic image processing apparatus according to claim 6.
The ultrasonic image processing apparatus, wherein the n-th product rate is a set of an average and a variance of the echo data in the sample region. The ultrasonic image processing apparatus according to claim 7.
The ultrasonic image processing apparatus, wherein the sample region includes the target voxel. The ultrasonic image processing apparatus according to claim 8.
The ultrasonic image processing apparatus, wherein the sample region includes a plurality of voxels arranged in a line in the perspective line direction.

Images