JP4659974B2 - Ultrasonic diagnostic equipment - Google Patents

Description

【００７２】
ここで、ｋｉは重み付け係数を示し、Σｋｉ＝１もしくは全てのｋｉ＝０（ｉ＝１〜５）である。
【００７３】
本実施形態では、上述した（１）〜（３）の条件を任意に選択・変更して探索中央座標値を推定できるようになっている。通常は、（３）の条件がデフォルトとして設定される。
【００７４】
なお、上述の一般的表現において、ｋ２〜ｋ５＝０（ｋ１＝１）とすれば、前記（１）の条件での推定と等価な推定値ｖα、ｖβを得ることができ、ｋ１＝０とすれば、前記（２）の条件での推定と等価な推定値ｖα、ｖβを得ることができる。
【００７５】
臨床用途に応じて最適な装置条件、すなわち、フレーム（ボリューム）レートの設定（時相間の時間差が変化する）や、演算ピッチの大小の設定が変わるので、各条件に応じて適切な重みｋｉを与えることが好ましい。
【００７６】
例えば、移動ベクトルの移動量に対して、時相差が十分小さいケースではｋ１の重みを大きくして、前記（１）の条件に近い状態で、また演算ピッチが十分に小さいケースではｋ２〜ｋ５の重みを大きくして、前記（２）の条件に近い状態で推定することが好ましい。
【００７７】
なお、上述の例では、簡単のため過去の移動ベクトルとして１点のベクトルを用いる例を示したが、隣接する複数点における過去の移動ベクトル量に適宜な重み付けを行って、これを用いてもよい。
【００７８】
また、全ての係数ｋｉ＝０に設定することもできる。この設定は特別なケースであり、探索中央座標＝演算位置に設定することを意味する。これは前述の各種の連続性を仮定できないような状況の場合に用いる。例えば、演算の対象としている局所領域の受信信号がノイズである場合、演算位置では前記連続性は仮定できず、かかる設定（ｋｉ＝０）を選択できるようにしておくことが望ましい。この設定により、移動ベクトルの演算それ自体の信頼性が低下する（実質的に無意味になる）ようなノイズに対しても、推定誤差がむやみに大きくなるのを避けることができ、推定の信頼性を確保することができる。
【００７９】
演算位置の信号がノイズであるか否かの判定には、類似性の情報や類似性を算出する過程での中間情報を用いられる。例えば、相互相関係数を用いて、相関係数の絶対値があるしきい値以下であるときにはノイズと見なし、この設定（ｋｉ＝０）に切り換えるようにする。あるいは、受信信号のパワー（相互相関係数の演算の途中で生ずる。簡易的には、相互相関係数を求める式における、規格化のための分母部分がそのまま使える）を用いて、信号パワーがあるしきい値以下であるときにはノイズと見なし、この設定（ｋｉ＝０）に切り換えるようにする。相互相関係数及び信号パワーを組み合わせてしきい値処理してもよい。
【００８０】
本実施形態では、後述するステップＳ１２、Ｓ１３に示す如く、各演算位置における演算結果である相互相関係数を用いて、かかるノイズ判定を行うようにしている。これにより、信号値がノイズであると認識された場合、重み係数ｋｉ＝０に設定して、相互相関係数の再計算を行うようにしている。
【００８１】
このようにして推定された移動ベクトルの移動先座標値ｖα、ｖβを用いた確からしい探索中央座標（Ｘｃ，Ｙｃ）ｉｊＬ＝（Ｘｉ＋ｖα、Ｙｊ＋ｖβ）Ｌが決まる。
【００８２】
［指定探索範囲の推定］
次いで、図７を参照して、指定探索範囲ＳＲ_ＳＰＣの適応的推定処理を説明する（ステップＳ６）。
【００８３】
前述のように推定された座標（Ｘｃ，Ｙｃ）ｉｊＬを中心にベストマッチする部位として移動ベクトルを探すので、この探索範囲が小さいほど演算は高遠となるわけである。この探索範囲は、勿論、実際の移動範囲（真値）よりも大きく設定されていなければ、類似性は正確に評価できない。そこで、最適な探索範囲、即ち２ＸｗｉｊＬと２ＹｗｉｊＬの値を推定する。
【００８４】
この場合においても、前述の探索中央座標の推定と同様に、生体の運動の時空間的な連続性が利用される。前述との違いは、確からしい移動ベクトル量の大きさ（移動速さ）の「誤差の上限」を推定する必要があるということである。そこで、過去の移動ベクトル
【数２】

が最大となるｍａｘＶ_ｅｒｒを用いて、探索範囲ＳＲ_ＳＰＣを以下のように設定する。
【００８５】
【数３】

【００８６】
これは、係数ｋａ＝１のとき、推定した座標値ｖα、ｖβの誤差として、±ｍａｘＶ_ｅｒｒの範囲を見込んでいることに相当する。探索範囲に余裕を持たせること、及び、移動の加速成分を考慮する必要があることから、少し大きめ範囲とするために、係数ｋａ≧１と設定しておくことが好ましい。
【００８７】
なお、最適な指定探索範囲ＳＲ_ＳＰＣは、上述の設定法の他にも、種々の設定法を選択的に採り得る。
【００８８】
例えば、過去の移動ベクトル
【数４】

のベクトルのｘ成分の標準偏差ｖｓｘとｙ成分の標準偏差ｖｓｙ（ｖｓ自体はスカラー量）を用いて、以下のように設定してもよい。
【００８９】
【数５】

【００９０】
この場合に、前述の係数ｋａと同様の趣旨により、係数ｋｂ≧１に設定しておくことが好ましい。
【００９１】
なお、臨床用途に応じて、ベクトル情報を選択的に用いることが望ましい。例えば、時間的違続性が小さいと考えられるケースでは、過去の移動ベクトル（ｖｘ，ｖｙ）ｉｊＬ−１を用いないようにし、反対に、空間的違続性が小さいと考えられるケースでは、演算済みの点における移動ベクトル（ｖｉ−１ｊ−１Ｌ，ｖｉｊ−１Ｌ，ｖｉ＋１ｊ−１Ｌ，ｖｉ−１ｊＬ）を用いないようにする。
【００９２】
さらに、空間的連続性が小さい場合には、１時相前の移動ベクトル（ｖｘ，ｖｙ）ｉｊＬ−１と、更にもう１時相前の（ｖｘ，ｖｙ）ｉｊＬ−２とを用いて、
【数６】

により、以下のように最適な指定探索範囲ＳＲ_ＳＰＣを設定してもよい。
【００９３】
【数７】

【００９４】
この場合にも、前述と同様に、係数ｋｃ≧１としておくことが好ましい。
【００９５】
このようにして、適切な探索中央位置と適切な探索領域サイズとを動的に設定することができる。
【００９６】
以上のように、指定探索範囲ＳＲ_ＳＰＣが最適な必要最小限の範囲として設定されると、特徴量演算器１６のＣＰＵ３２は、図６に示す如く、この指定探索範囲ＳＲ_ＳＰＣ内にオブジェクトＲＯＩ：ＯＢ_ＲＯＩをその初期位置に設定する（図５、ステップＳ７）。このオブジェクトＲＯＩ：ＯＢ_ＲＯＩの大きさは、カーネルＲＯＩ：ＫＮ_ＲＯＩのそれと同じに設定される。
【００９７】
次いで、ＣＰＵ３２は、第１フレームＦの与えられた演算位置に在るカーネルＲＯＩ：ＫＮ_ＲＯＩと第２フレームＦ＋１の指定されたオブジェクトＲＯＩ：ＯＢ_ＲＯＩとの間で、類似性を表す特徴としての相互相関係数を演算し、その値をメモリ３３に格納する（ステップＳ８）。この相互相関係数ρ_ｍ，ｎは、カーネルＲＯＩの画素値の信号をＡ、オブジェクトＲＯＩのそれをＢとするとき、
【数８】

に基づき演算される。なお、この演算時において、分母の値（信号パワー値）も中間情報としてメモリに演算毎に格納される。
【００９８】
次いで、ＣＰＵ３２は、オブジェクトＲＯＩ：ＯＢ_ＲＯＩを指定探索範囲ＳＲ_ＳＰＣ内の全ての位置に移動させて相互相関係数を演算し終わったか否かを判断し、指定探索範囲ＳＲ_ＳＰＣ内で未だ終わっていない位置がある場合、オブジェクトＲＯＩ：ＯＢ_ＲＯＩを移動させて同様の演算を繰り返す（ステップＳ９，Ｓ１０）。反対に、必要最小限に適応設定された指定探索範囲ＳＲ_ＳＰＣ内の全ての演算位置で係数演算が終わったときには、ＣＰＵ３２は、それまで演算してメモリ３３に格納してある複数の相互相関係数の中から、代表値として、値が最大の相互相関係数を選択する（ステップＳ１１）。
【００９９】
次いで、ＣＰＵ３２は、この選択された相互相関係数（最大値）の絶対値が所定のしきい値よりも大きいか否かを判断する（ステップＳ１２）。この判断は、前述した信号値がノイズか否かを見極める処理であり、ノイズである、即ち、｜係数｜＞しきい値の判断が下されたときには、前述のように全ての重み係数ｋｉ＝０に設定し、再度、相互相関係数を演算する（ステップＳ１３）。なお、この判断は前述したように、メモリ３３に格納されている演算途中情報としての信号パワー値に基づいて行ってもよい。
【０１００】
反対に、ノイズでは無く、正常な範囲の信号値に基づく相互相関係数であると認識されたときには（ステップＳ１２；ＮＯ）、ＣＰＵ３２は、カーネルＲＯＩ：ＫＮ_ＲＯＩを全ての演算位置に移動させて演算し終わったか否かを判断し、未だ演算が終わっていない位置が在る場合、カーネルＲＯＩ：ＫＮ_ＲＯＩの位置を更新する（ステップＳ１４、Ｓ１５）。そして、指定探索範囲ＳＲ_ＳＰＣ及びそのサイズの適応的設定の処理に戻り、上述の一連の処理が所望する全ての演算位置に対して実行される。
【０１０１】
このように、指定探索範囲ＳＲ_ＳＰＣを各演算位置（カーネルＲＯＩ：ＫＮ_ＲＯＩの位置）毎に適応的に最適設定しながら、各演算位置における相互相関係数が求められる。このため、上述した一連のパターンマッチングの処理が終わると、メモリ３３には、所望断層像の類似性を表す特徴量（ここでは相互相関係数）の２次元分布データが得られる。この特徴量の２次元分布データは、ＤＳＣ１７により、特徴量画像として断層像に並置、又は、断層像に重畳され、この画像が表示モードで表示器１８に表示される。
【０１０２】
このように、本実施形態によれば、心筋などの動く臓器の時系列的な類似性を観察するためのパターンマッチングを行うに際し、指定探索範囲ＳＲ_ＳＰＣは、各演算位置毎に、必要最小限の範囲に適応的に推定・設定される。従って、類似性を表す特徴量は、その必要最小限の領域のみを探索して演算すれば足りる。このため、従来のように、データに含まれる最大の移動速度を検出するために大きめに定められたサーチＲＯＩの範囲を各演算位置毎にくまなく探索する構成に比べて、特徴量を得るまでに要する演算量が大幅に少なくなり、その演算が著しく高速化される。
【０１０３】
なお、上述した超音波診断装置では、類似性を表す特徴量を求めるために、カーネルＲＯＩとオブジェクトＲＯＩとの間で相互相関係数を逐一演算するように構成したが、これは更に以下のようにＳＡＤ（Ｓｕｍ−Ａｂｓｏｌｕｔｅ−Ｄｉｆｆｅｒｅｎｃｅ：相互差分の絶対値和）を用いた演算方式に変形可能である。
【０１０４】
ＳＡＤε_ｍ，ｎは、両画像の画素値の信号をＡ及びＢとするとき、
【数９】

で表される。
【０１０５】
ＳＡＤ法によって得られる空間的最小値の位置は、類似性の高い位置であると考えられるので、このＳＡＤ法を用いることで移動ベクトルを検出可能である。相互相関法はパターンマッチングの分野でゴールドスタンダードと言われているが、ＳＡＤ法は、この相互相関法よりも演算ステップが少ないので、処理が高速という利点がある。しかし、ＳＡＤ法は、移動ベクトルを検出できるが、これだけを単独で用いても、類似性の定量化の点では信号源の大きさに左右されるので規格化が困難であり、物理的に意味のある一定の指標は得にくい。
【０１０６】
そこで、ＳＡＤの高速の利点を生かし、まずＳＡＤで移動ベクトルを探して、検出された移動ベクトル位置にて相互相関係数を求めるようにする。これにより、時間の掛かる相互相関演算を１回に抑えるので、相互相関演算で移動ベクトルを検索する場合よりも高速化できると共に、相互相関係数という、類似性に関して一定の意義を有する物理量を得ることができる。
【０１０７】
このＳＡＤ法と相互相関法とを組み合わせて実行する場合、例えば前述した図５の処理においては、ステップＳ８〜Ｓ１０の代わりに、カーネルＲＯＩと指定探索範囲との間でＳＡＤを演算する処理を行い、次いで、ＳＡＤで求めた移動ベクトルの移動先を中心とするオブジェクトＲＯＩとカーネルＲＯＩとの間で相互相関係数を演算する２つのステップを置けばよい。
【０１０８】
前述した図５の処理とは別に、カーネルＲＯＩと初期探索範囲と間でＳＡＤを演算する処理を行い、次いで、ＳＡＤで求めた移動ベクトルの移動先を中心とするオブジェクトＲＯＩとカーネルＲＯＩとの間で相互相関係数を演算する２つのステップを設けるようにしてもよい。
【０１０９】
ところで、本願発明は、パターンマッチングの演算で得られる移動ベクトルの情報を効果的に利用した３次元の時空間における関心部位の移動追跡（トラッキング）及びその表示法を併せて提供することができる。この表示に関する実施形態を以下に説明する。
【０１１０】
（第４の実施形態）
図８〜９を参照して、本発明の第４の実施形態を説明する。
【０１１１】
本実施形態では、前述した図４と同様に構成した特徴量演算器１６において、そのＣＰＵ３２には、制御器３４及び操作器３５と協働して、前述した特徴量の演算に加えて、関心部位の移動追跡（トラッキング）の演算を行う機能が与えられている。
【０１１２】
このため、ＣＰＵ３２は図８に概略示す一連の処理を行う。具体的には、初期時相Ｓ及び終了時相Ｅを設定し、「時相」の変数＝Ｓを設定し、そして初期時相に該当するフレームの受信信号をフレームメモリ３１から呼び出して、その受信信号が形成する空間領域にオペレータが関心を寄せている関心点を設定する（ステップＳ２１〜Ｓ２３）。
【０１１３】
次いで、移動ベクトル＝０に初期設定した後、「関心点」の変数＝関心点＋Ｖ（移動ベクトル）に設定し、更に、時相＝時相＋１にインクリメントする（ステップＳ２４〜Ｓ２６）。
【０１１４】
次いで、ＣＰＵ３２は、上述の「時相＝Ｓ＋１」から「時相−１＝Ｓ」に渡る２フレームの受信信号の相互間で、前記関心点の移動ベクトルＶを演算する（ステップＳ２７）。この移動ベクトルＶは、従来周知の方法で演算してもよいし、前述した第３の実施形態で説明した高速化した探索領域設定の中で求める方法でもよい。次いで、ＣＰＵ３２は、移動ベクトルＶの軌跡データをＤＳＣ１７に出力し、この軌跡の表示を更新させる（ステップＳ２８）。
【０１１５】
この後、ＣＰＵ３２により、時相＜終了時相Ｅか否かの判断と伴に、終了時相Ｅまで時相を更新しながら上述した関心部位の移動追跡処理が繰り返される（ステップＳ２５〜Ｓ２９）。時相＝終了時相Ｅに到達すると、他に関心部位がオペレータから指定されたか否かを判断し、指定された場合にはステップＳ２２まで遡って移動追跡処理が繰り返される（ステップＳ３０）。反対に、他の関心部位が指定されなければ、移動追跡は終わる。
【０１１６】
以上の移動追跡処理により、ある初期時相のフレームの受信信号が構築する空間画像上に任意の１つ以上の関心部位を指定する手段と、任意の関心点での移動ベクトルを算出する手段と、その移動ベクトルの移動先位置に演算点（位置）を移動させて移動ベクトルを算出する手段とが機能的に構成される。これらの手段による移動追跡処理は、時相経過毎に終了時相まで逐次繰り返されるので、関心部位の時空間的な移動が追跡される。
【０１１７】
この移動に伴う軌跡は、例えば図９又は図１０に示す如く、ＤＳＣ１７によって、初期時相から指定された終了時相までの追跡位置として画像上の位置に対応される。
【０１１８】
ＤＳＣ１７は、操作器３５を介してオペレータが与えた所望の表示モードの指令信号を制御器３４から受け、この指令信号に応じた移動軌跡の情報を画像上に表示すべく画像データの合成処理を行う。図９及び図１０は共に、画像上の構造物（心筋など）上に軌跡情報を線で重ね、その位置変化を示すもので、図９は２次元画像上に、図１０は３次元画像上にそれぞれ示す。関心部位の移動軌跡を表示するための時相区間（初期時相から終了時相までの間隔）は１心周期が好適であるので、図９及び図１０も１心周期における軌跡の表示画像として示されている。いずれの画像も、ある時相における画像（２次元の場合には、ある時相における任意断面の画像）を表示し、初期時相から終了時相までの時相期間内の全軌跡をその画像上で線を用いて表示し、位置を対応付けている。３次元の場合には、表示断面の裏側に移動する場合、例えば点線で表される。
【０１１９】
なお、２次元や３次元の画像データとしては、構造を表す情報（２次元では通常、Ｂモードと呼ばれる）であってもよいし、類似性を表す特徴量（相互相関係数など）の情報であってもよい。
【０１２０】
（表示の変形例その１）
なお、ＤＳＣ１７は、背景となる画像を表示させるに際し、特定の時相の画像のみを表示する構成に限定されるものでは無く、初期時相から終了時相までの各時相と移動軌跡とを逐一対比させる表示を採用してもよい。この例を２次元画像について示すと、図１１〜１３のようである。
【０１２１】
図１１の表示例は、断層像と関心部位の移動軌跡とを共に更新しながら、移動軌跡については、過去の軌跡に現在の軌跡を追加しながら表示するものである（イメージ的には、イメージメモリをめくって観察する状態を想定した例である）。図中、×印は、オペレータが指定した関心部位を示す。
【０１２２】
また、図１２の表示例では、断層像は更新し、移動軌跡はその全体を最初の時相から表示しつつ、現在表示されている時相に相当する軌跡部分（図中、黒丸で示す）を他とは差別化して示される（この場合も、イメージ的には、イメージメモリをめくって観察する状態を想定した例である）。
【０１２３】
さらに、図１３（ａ），（ｂ）の表示例では、特徴的な２時相（例えば拡張期末期及び収縮期末期）の断層像と全体の移動軌跡とを並べて表示させ、移動軌跡については各時相部分（図中、黒丸で示す）を差別化して示すものである。
【０１２４】
このようにして様々な態様で関心部位の時空間的な軌跡が表現されるので、心周期にわたる局所的な運動の様子を視覚的に確実に観察することができる。
【０１２５】
（表示の変形例その２）
また、別の表示法として、移動ベクトルの速さ情報を一緒に付加して表示する態様が挙げられる。この速さ情報は時相間の時間と移動距離とから演算されるもので、これを特徴量演算器１６のＣＰＵ３２又はＤＳＣ１７に何れで求めるようにしてもよいし、皿には、別途、演算器を設けるようにしてもよい。
【０１２６】
図１４（ａ）の表示例は、異なる速さには異なる色相を対応させた速さインジケータＶ_ＩＮＤを設け、線ＭＬで表した移動軌跡の色を速さインジケータＶ_ＩＮＤに基づいて変えたものである。同図（ｂ）の表示例は、異なる速さには異なる輝度を対応させた速さインジケータＶ_ＩＮＤを設け、線ＭＬで表した移動軌跡の輝度を速さインジケータＶ_ＩＮＤに基づいて変えたものである。なお、図１４（ａ）及び（ｂ）に示す表示例では、色や輝度の違いをハッチングの違いで表すために、移動軌跡の線を少し誇張して示している。
【０１２７】
更に、同図（ｃ）の表示例は、異なる速さには異なる太さ（線幅）を対応させた速さインジケータＶ_ＩＮＤを設け、線ＭＬで表した移動軌跡の太さを速さインジケータＶ_ＩＮＤに基づいて変えたものである。同図（ｄ）の表示例は、移動ベクトルの移動軌跡を線分矢印ＡＲで表示し、求められた速さを線分の長さに反映させたものである。
【０１２８】
（表示の変形例その３）
さらに、別の表示法として、移動ベクトルの移動軌跡を心時相と対応付けて表示する例が挙げられる。この対応付け表示は、例えば、心電計により収集した心電図（ＥＣＧ）信号を装置に取り込み、ＣＰＵ３２に収集時相と心電図信号との対応させる一方で、心電図信号をＤＳＣ１７に表示用として送ることでなされる。
【０１２９】
図１５（ａ）の表示例によれば、異なるＥＣＧ時相には異なる色相を対応させたＥＣＧインジケータＥＣＧ_ＩＮＤとＥＣＧ波形とが表示され、線ＭＬで表した移動軌跡の色相がＥＣＧインジケータＥＣＧ_ＩＮＤに基づいて変えられる。また、同図（ｂ）の表示例によれば、異なるＥＣＧ時相には異なる輝度を対応させたＥＣＧインジケータＥＣＧ_ＩＮＤとＥＣＧ波形とが表示され、線ＭＬで表した移動軌跡の輝度がＥＣＧインジケータＥＣＧ_ＩＮＤに基づいて変えられる。さらに、同図（ｃ）の表示例によれば、異なるＥＣＧ時相には異なる太さ（線幅）を対応させたＥＣＧインジケータＥＣＧ_ＩＮＤとＥＣＧ波形とが表示され、線ＭＬで表した移動軌跡の太さがＥＣＧインジケータＥＣＧ_ＩＮＤに基づいて変えられる。同図（ｄ）の表示例は、移動ベクトルの移動軌跡を線ＭＬで表すとともにＥＣＧ波形を表示し、その一方で、その両方に時相を表す矢印ＡＲ１、ＡＲ２を付す。この両矢印ＡＲ１、ＡＲ２は共に連動しながら常に移動するように表示してもよいし、常時は静止していて、何れか一方を所望位置又は所望時相に移動させたときに、他方が一方に連動して移動するようにしてもよい。
【０１３０】
（表示の変形例その４）
さらに、別の表示法として、移動ベクトルの速さ情報の付加とその移動軌跡の心時相との対応付けとを同時に行う例が挙げられる。
【０１３１】
例えば、前述した図１４（ｄ）の表示と図１５（ａ）の表示とを組み合わせた表示法や、図１４（ａ）の表示と図１５（ｄ）の表示とを組み合わせた表示法などを採ることができる。
【０１３２】
さらに、このような組み合わせ表示において、移動軌跡線ＭＬには速さ情報を付加せずに、時相情報のみを識別可能に与え、その一方で、心時相との対応付けや速さ情報を別のグラフ又は数値として表示するようにしてもよい。図１６の表示例では、速さ情報をグラフＧＲとして与え、ＥＣＧ波形との間で心時相との対応付けを点線ＤＬで行っている。
【０１３３】
（表示の変形例その５）
その他の表示例として、ストレスエコー検査におけるストレス付与前後の同一部位の移動軌跡を同時に表示する例（図１７参照）が挙げられる。この例では、両移動軌跡がＥＣＧ波形と共に表示され、心時相と軌跡線ＭＬとの間の対応付けが黒丸印でなされている。
【０１３４】
（表示の変形例その６）
更に別の表示例として、複数のセグメント毎の移動軌跡を同時に表示し比較可能にする例（図１８参照）が挙げられる。この例では、心筋に設定した４箇所の関心部位での移動軌跡線ＭＬと、各軌跡線ＭＬの速さ情報のグラフと、ＥＣＧ波形とが示されている。
【０１３５】
このような種々の表示法によって、心筋の局所的な壁運動の時空間的軌跡に移動の程度が加味され、かつ、心持相と対比付けて示されるので、壁運動の４次元的な様相を表現することができ、壁運動評価の診断支援に有効になる。
【０１３６】
（移動ベクトルを取得するための別の例）
前述した図８では、時相を更新しながら、移動ベクトルを逐一演算するように構成したが、この移動ベクトル取得の別の例を図１９に示す。この例は、既に演算して記憶してある移動ベクトルを読み出して利用する処理を行う。
【０１３７】
図１９の処理は、例えば特徴量演算器１６のＣＰＵ３２又はＤＳＣ１７で行われ、類似性画像生成ルーチンと軌跡表示ルーチンとから成る。類似性画像生成ルーチンでは、例えば特開平８−１６４１３９号に記載の手法に拠って類似性を示す特徴量（例えば相互相関係数）の画像を初期時相Ｓから終了時相Ｅまで、対象画像で演算して例えばメモリ３３に記憶する。なお、この類似性画像生成ルーチンにおいて、前述した第３の実施形態及びその変形形態で説明した高速な特徴量の演算法を用いてもよい。この演算が終わると、軌跡表示ルーチンに移行し、移動ベクトルを記憶したメモリから時相経過毎に逐次、移動位置を読み出して追跡表示させる。これにより、パターンマッチングの演算結果を有効に利用して、関心部位のトラッキングを行うことができ、少ない演算量で充実したアプリケーションを備えた超音波診断装置を提供することができる。
【０１３８】
なお、本発明は上述した実施形態や変形形態の構成に限定されるものではなく、特許請求の範囲に要旨に基づき、さらに種々の形態に変形可能なことは勿論である。例えば、特徴量演算器は、前述のようなＣＰＵによるソフトウェア機能を持たせて構成する他に、ＡＳＩＣなどの専用のハードウェアで構成してもよい。
【０１４０】
【発明の効果】
以上説明したように、本発明に係る超音波診断装置によれば、パターンマッチング演算の高速化によって、検査時間の短緒と、関心部位のより信頼性の高い３次元的な運動評価とを容易に両立させることができる。
【図面の簡単な説明】
【図１】本発明の第１の実施形態に係る超音波診断装置の概要を示すブロック図。
【図２】本発明の第２の実施形態に係る超音波診断装置の概要を示すブロック図。
【図３】本発明の変形した実施形態に係る超音波診断装置の概要を示すブロック図。
【図４】本発明の第３の実施形態に係る超音波診断装置に搭載した特徴量演算器の概要を示すブロック図。
【図５】第３の実施形態の特徴量演算器のＣＰＵで実行されるパターンマッチングの処理概要を示すフローチャート。
【図６】異なる時相の２枚のフレーム画像におけるカーネルＲＯＩ、初期探索範囲、指定探索範囲、及びオブジェクトＲＯＩの位置関係を説明する図。
【図７】指定探索範囲を適応的に設定する手法を模式的に説明する図。
【図８】第４の実施形態の特徴量演算器のＣＰＵで実行される関心部位のトラッキング（移動追跡）の処理概要を示すフローチャート。
【図９】２次元の画像データに対するトラッキングの表示例を示す画像図。
【図１０】３次元の画像データに対するトラッキングの表示例を示す画像図。
【図１１】２次元の画像データに対するトラッキングの別の表示例を示す画像図。
【図１２】２次元の画像データに対するトラッキングの更に別の表示例を示す画像図。
【図１３】２次元の画像データに対するトラッキングの更に別の表示例を示す画像図。
【図１４】トラッキングの更に別の表示例を示す図。
【図１５】トラッキングの更に別の表示例を示す図。
【図１６】トラッキングの更に別の表示例を示す図。
【図１７】トラッキングの更に別の表示例を示す図。
【図１８】トラッキングの更に別の表示例を示す図。
【図１９】トラッキングの更に別の表示例を示す図。
【符号の説明】
１１ 超音波プローブ
１２ 送信系
１３ 受信ビームフォーマ
１４ 信号処理器
１５ 固定エコー除去器
１６ 特徴量演算器
１７ ＤＳＣ
１８ 表示器
２１ バンドパスフィルタ
３１（３１〜３１Ｎ） フレームメモリ
３２ ＣＰＵ
３３ メモリ
３４ 制御器
３５ 操作器[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an ultrasonic diagnostic apparatus, and more particularly to an ultrasonic diagnostic apparatus having a function of performing a time-series cross-correlation operation on a three-dimensional local region with high performance.
[0002]
[Prior art]
An ultrasonic diagnostic apparatus is an apparatus that obtains an image from a received signal obtained by transmitting / receiving an ultrasonic signal to / from a subject, and is used in various modes using the non-invasive nature of an ultrasonic signal. Yes. There are various types of ultrasonic diagnostic apparatuses, but the mainstream is an apparatus that obtains a tomographic image of a soft tissue of a living body using an ultrasonic pulse reflection method.
[0003]
In recent years, this ultrasonic diagnostic apparatus has been remarkably developed, and various arithmetic functions effective for clinical use have become rich. One of them is known to have a function of calculating a time-series similarity of a two-dimensional or three-dimensional local region, that is, a so-called pattern matching function.
[0004]
As this pattern matching function, for example, a function described in JP-A-8-164139 is known. According to the ultrasonic diagnostic apparatus described in this publication, the reception signal output from the reception beamformer is received separately from the system that generates the tomographic image data, and the first frame kernel ROI and the second frame object are received. Computation means for obtaining two-dimensional distribution information of feature quantities (for example, cross-correlation coefficients) representing the similarity between the received signals and the ROI is provided. Thereby, for example, the degree of deformation of the myocardium can be quantified between two frames having different cardiac phases and represented as an image, and this image can be displayed in parallel with the tomographic image or superimposed on the tomographic image.
[0005]
[Problems to be solved by the invention]
However, the pattern matching function provided in the ultrasonic diagnostic apparatus described in the above-mentioned gazette is that the received signal received by the probe is subjected to phasing addition processing by the receiving beamformer, and then the time-series feature amount using this received signal. Therefore, the reception signal containing various artifact components from the living body is directly used for the pattern matching calculation, and the performance of this calculation deteriorates frequently.
[0006]
For example, if there is a fixed reflector such as a chest wall at a short distance or artifacts of fixed echo due to multiple reflections at that point, the correlation coefficient between frames of the received signal is always high, so the myocardium at a short distance (A typical case is when an external apex approach is performed on the apex). In particular, the apex is said to be easily affected by all three major coronary arteries (left coronary artery, right coronary artery, circumflex branch). Despite being an important part for diagnosis, fixed echo artifacts are likely to occur.
[0007]
In the ultrasonic diagnostic apparatus described in the above publication, evaluation based on a tomographic image is the center. Therefore, when body movement occurs in a direction perpendicular to the tomographic plane, a time-series similarity can be obtained even with pattern matching. There is a problem that it cannot be evaluated accurately.
[0008]
In order to solve the problem of inaccuracy in similarity evaluation, it is desired to perform three-dimensional pattern matching (evaluation). However, each dimension of the kernel area used for pattern matching and the area to be searched is 2 Since the number of dimensions increases from three to three, the amount of computation (calculation time) becomes enormous, and there is a problem that high-speed processing cannot be performed.
[0009]
Furthermore, in the case of the ultrasonic diagnostic apparatus described in the above-mentioned publication, pattern matching is only performed and the result is displayed as an image. This is visually related to the movement of the position of interest of an organ such as the myocardium. There was a request to enhance the display function for monitoring.
[0010]
The present invention has been made to overcome the reality of such conventional pattern matching, and eliminates or suppresses the influence of body movement in the direction perpendicular to the artifact and the tomographic plane included in the received signal from the living body. One object of the present invention is to provide an ultrasonic diagnostic apparatus having a pattern matching function that can be processed at a high speed with a small amount of calculation and high performance.
[0011]
Another object of the present invention is to make it possible to visually monitor a movement of a region of interest of an organ with a simple operation simultaneously with or separately from the above-described object.
[0013]
[Means for Solving the Problems]
  To achieve the above purposeThe ultrasonic diagnostic apparatus according to the present invention is, CoveredScanning means for periodically performing two-dimensional or three-dimensional scanning of the diagnostic region of the specimen with an ultrasonic signal;SaidA signal collecting means for obtaining a reception signal having an electric quantity corresponding to an ultrasonic reflection signal accompanying scanning, and first and second spatial regions formed by two sets of reception signals having different scanning timings from each other. Similarity acquisition means for obtaining spatial distribution information of features representing the similarity between the second local regions;,In the ultrasonic diagnostic apparatus comprising:The feature amount is calculated by moving the second local region within a search range, and the size of the search range is changed based on vector information obtained based on the ultrasonic reflection signal. TheSearch range setting means to setMoreHave.
[0017]
  Furthermore, an ultrasonic diagnostic apparatus according to the present invention isAnotherAs a mode of the present invention, a scanning unit that periodically performs two-dimensional or three-dimensional scanning of the diagnostic region of the subject with an ultrasonic signal;SaidA signal collecting means for obtaining a reception signal having an electric quantity corresponding to an ultrasonic reflection signal accompanying scanning, and first and second spatial regions formed by two sets of reception signals having different scanning timings from each other. Similarity acquisition means for obtaining two-dimensional distribution information of feature quantities representing similarity between second local regions;,Diagnostic device withInThe similarity acquisition means calculates an absolute value sum (SAD) of a mutual difference between signal values in the first and second spatial regions.TKuttleinformationA means of seekingSaidvectorBased on informationMeans for obtaining a cross-correlation coefficient of the signal value as the feature quantity;The similarity acquisition means calculates the feature amount based on the second local region moved within the search range and the vector information, and the size of the search range is calculated as follows: Search range setting means for changing and setting based on the vector informationTheMoreHave.
[0021]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments according to the present invention will be described below with reference to the accompanying drawings.
[0022]
(First embodiment)
The ultrasonic diagnostic apparatus according to the first embodiment will be described with reference to FIG.
[0023]
This ultrasonic diagnostic apparatus will be described by adopting a sector electronic scanning method with a shallow shallow field of view and a wide deep field suitable for cardiac diagnosis, but if it can scan a two-dimensional or three-dimensional region of a subject, You may employ | adopt a linear electronic scanning method, a convex electronic scanning method, etc.
[0024]
This ultrasonic diagnostic apparatus includes an ultrasonic probe (hereinafter simply referred to as “probe”) 11. The probe 11 includes a transducer array for sector scanning in which a plurality of transducers that reversibly convert acoustic signals and electrical signals are arranged at the tip thereof. The probe 11 is electrically connected to the transmission system 12 during transmission, and is electrically connected to the reception beamformer 13 during reception.
[0025]
Although not shown, the transmission system 12 includes a clock generator, a rate pulse generator, and a delay circuit and a pulser provided for each transducer (each channel). The clock generator generates a clock pulse that is sent to the rate pulse generator. The rate pulse generator divides the clock pulse into, for example, 5 kHz rate pulses. The rate pulse is distributed for each channel, sent to each delay circuit, and subjected to delay processing for a desired time. That is, each delay circuit gives the rate pulse a delay time necessary to focus the ultrasonic wave into a beam and determine the direction of the ultrasonic beam (called the scanning line direction or azimuth direction).
[0026]
The delayed rate pulse is sent to the pulsar for each channel. Thus, the pulser applies a pulse voltage to the corresponding vibrator at the reception timing. Thereby, an ultrasonic pulse is emitted from the probe 11 in the azimuth direction according to the delay time.
[0027]
A part of the emitted ultrasonic pulse is reflected at the boundary of the acoustic impedance in the subject and becomes a reflected wave. This reflected wave is received by the transducer of the probe 11, converted into an electric signal, and taken into the reception beam former 13.
[0028]
The reception beamformer 13 is configured by a digital beamformer system as shown in Japanese Patent Publication No. 6-14934. The reception beamformer 13 includes a preamplifier, an A / D converter, a digital delay circuit, and an adder (not shown). An analog reception signal sent from the transducer for each channel is amplified by a preamplifier, converted to a digital signal by an A / D converter, and sent to a digital delay circuit. In this circuit, in order to obtain reception directivity, a delay time opposite to that at the time of transmission is given for each channel, and then added by an adder. This added digital signal is hereinafter referred to as a received signal.
[0029]
By sequentially controlling the delay time at the time of transmission / reception described above for each repetition of the ultrasonic pulse, the two-dimensional region of the subject can be sector-scanned to obtain one frame (one set) of received signals. Furthermore, by repeating such sector scanning periodically, a received signal of a plurality of time-series frames can be obtained.
[0030]
As shown in the figure, a signal processor 14, a fixed echo remover 15, and a feature amount calculator 16 are provided in parallel on the output side of the reception beamformer 13.
[0031]
For this reason, the reception signal generated by the reception beamformer 13 is sent to the signal processor 14 as one of the transmission destinations. For this reason, the received signal is detected and logarithmically amplified by the signal processor 14 and generated as B-mode image (tissue tomographic image) data. This image data is sent to a DSC (digital scan converter) 17.
[0032]
The reception signal generated by the reception beamformer 13 is also sent to the fixed echo remover 15. The remover 15 is constituted by a time series HPF including a plurality of frame memories, a plurality of multipliers, a multiplication coefficient ROM, an adder, and a filter controller, as described in, for example, JP-A-8-107896. . As a result, the time series HPF is formed in the inter-frame HPF filter, and the cutoff frequency that removes the fixed echo component and passes the other signal echo components is removed from the signal value of the received signal at the same position in each frame. Have.
[0033]
For this reason, the stationary echo remover 15 suppresses the stationary echo component having no motion from the received signal to such an extent that it can be removed or almost ignored, and a received signal mostly composed of the signal echo component is generated. This received signal is sent to the feature quantity calculator 16 at the next stage.
[0034]
The feature quantity computing unit 16 is composed of a computing unit described in, for example, Japanese Patent Laid-Open No. 8-164139. The computing unit 16 includes a plurality of frame memories capable of writing reception signals for at least two frames having different scanning time phases (timing) of the tomographic plane, and reception signals for two frames having different time phases read from the frame memory. And an arithmetic unit for performing pattern matching.
[0035]
Specifically, the computing unit calculates a kernel ROI (local region) designated in the spatial region composed of the received signal of one frame and an object ROI (local region) of the spatial region composed of the received signal of the other frame. It is composed of a feature amount calculator that calculates a feature amount (for example, a cross-correlation coefficient or SAD) that represents the similarity between them. The object ROI is moved in a predetermined search ROI, and the feature amount is calculated for each movement. For this reason, since a plurality of feature amounts are calculated for one kernel ROI, a representative value of the feature amount (for example, the maximum value of the cross-correlation coefficient) is selected from these, and similarity information for the kernel ROI is selected. Is remembered as
[0036]
Next, the kernel ROI is moved to an adjacent position on the set space area, and the representative value of the feature value is obtained in the same manner as described above. For this reason, by changing the position of the kernel ROI and obtaining the representative value of the feature value at each position, the two-dimensional distribution information of the feature value can be obtained. The two-dimensional distribution information of the feature amount is converted into image data and sent to the DSC 17.
[0037]
The DSC 17 receives tomographic image data from the signal processor 14, and the feature amount image data from the feature amount calculator 16. For this reason, the DSC 17 converts the scan format into the standard TV system, and synthesizes both image data into one frame of image data in a display mode such as overlapping or juxtaposing as exemplified in JP-A-8-164139. To do.
[0038]
The composite image data is read at a predetermined timing for each frame and sent to the display 18. The display 18 converts the composite image data into an analog signal and displays it on the monitor screen. As a result, the B-mode tomographic image and the feature amount image as the pattern matching result are displayed in an appropriate manner on the monitor screen.
[0039]
According to the present embodiment, the fixed echo remover 15 is provided in the preceding stage of the feature quantity calculator 16, and the fixed echo from the stationary part is reliably removed or suppressed before the feature quantity calculation, that is, the pattern matching. In other words, most of the received signals used for pattern matching are signal components from a moving part.
[0040]
That is, when performing myocardial pattern matching, artifacts such as fixed echoes from fixed reflectors such as a chest wall at a short distance and multiple echoes are reliably removed in advance. Also, when diagnosing the apex, this artifact is reliably removed. For this reason, in particular, the performance of myocardial evaluation at a short distance and myocardial evaluation at the apex can be significantly improved as compared with the prior art.
[0041]
Conventionally, since the relational teaching is high in the stationary phase of the myocardial wall motion, the signal value is enhanced even in a normal region, and the evaluation may be hindered. However, in the present embodiment, signal components with high correlation due to the fact that they do not move at all even in such a normal part are removed in advance as fixed echoes by removing the fixed echoes in advance. For this reason, the case where the signal of a normal part is enhanced is reduced. Therefore, on the contrary, a highly correlated part (often abnormal in terms of the function of the myocardium) is enhanced only in the time phase in which the myocardium moves in the cardiac cycle (initial stage of contraction and initial stage of dilation). Accordingly, there is an advantage that when a myocardium is imaged, a specific site where an abnormality exists in the wall motion can be selectively provided.
[0042]
(Second Embodiment)
An ultrasonic diagnostic apparatus according to the second embodiment will be described with reference to FIG. Note that the same components as those in the apparatus of the first embodiment are denoted by the same reference numerals, and the description thereof is omitted or simplified.
[0043]
Similar to the first embodiment, this ultrasonic diagnostic apparatus is characterized by a configuration that eliminates or suppresses artifacts of a received signal sent to the feature amount calculator 16. In this embodiment, a technique called tissue harmonic imaging (THI) is used to realize such a feature.
[0044]
Specifically, as shown in FIG. 2, a band-pass filter (BPF) 21 which is a main element for performing tissue harmonic imaging is inserted in the reception path between the probe 11 and the reception beam former 13. This filter 21 includes an A / D converter (not shown) and a filter circuit, and receives an electrical echo signal received by the probe 11. The echo signal may be sent to the band-pass filter 21 after first being preamplified by a preamplifier.
[0045]
In the band pass filter 21, the echo signal is converted into a digital signal by an A / D converter and sent to a filter circuit. The filter circuit uses the center frequency f of the transmitted ultrasonic wave.₀In this example, band-pass filtering is performed to extract only signal components in a predetermined band centered on a harmonic frequency that is twice as an example. The extracted harmonic signal is sent to the reception beamformer 13 to generate a digital reception signal based on the harmonic signal, as described above. That is, the received signal based on this harmonic signal is sent to the signal processor 14 for generating the B-mode slice and the feature amount calculator 16 for performing pattern matching, as in the first embodiment. The same processing is performed.
[0046]
Thus, according to the present embodiment, the harmonic signal is extracted from the echo signal at the initial stage of reception. That is, after this extraction, generation and display of a B-mode tomographic image and a feature amount image (pattern matching) to which tissue harmonic imaging is applied. According to tissue harmonic imaging, by extracting the harmonic components, the received beam becomes very sharp and thin, so the artifact removal effect and visibility of the myocardium and intima are better than when using the fundamental component ing.
[0047]
Accordingly, since the artifact component included in the received signal based on the harmonic signal is greatly reduced, pattern matching is performed with higher accuracy using this received signal. In general, similarity cannot be accurately evaluated unless a sufficiently strong signal is obtained from the myocardium to be evaluated. However, according to the present embodiment, by using a harmonic signal, Since a reception signal with excellent visibility can be obtained, the similarity evaluation accuracy is also improved in this respect.
[0048]
Furthermore, by using tissue harmonic imaging in combination, the influence of individual differences on image quality is reduced, and it is possible to evaluate similarity with high accuracy for more patients.
[0049]
Note that the tissue harmonic imaging as the pre-processing of the pattern matching is not necessarily limited to the filter method using the bandpass filter 21 described above, and the harmonic component from the echo signal including the fundamental component and the harmonic component. For example, a pulse inversion method may be used.
[0050]
In the configuration of FIG. 2, the transmission system 12 suppresses or removes the fundamental wave component of the transmission ultrasonic wave, for example, by resonating or blocking, and generates a drive pulse that mostly depends only on the harmonic component of the transmission ultrasonic wave. The drive pulse can be supplied to the probe 11. With this configuration, since the transmitted wave is a harmonic signal, the side lobe of the transmitted ultrasonic wave is significantly reduced, and biological artifacts due to the side lobe are reduced accordingly, so pattern matching can be performed with higher accuracy. It can be carried out.
[0051]
Further, FIG. 3 shows an ultrasonic wave using the fixed echo removal method (configuration shown in FIG. 1) according to the first embodiment and the tissue harmonic imaging method (configuration shown in FIG. 2) according to the second embodiment. An example of a diagnostic device is shown.
[0052]
When only the tissue harmonic imaging method is used, fixed artifacts due to multiple reflections generated by a fixed reflector such as a chest wall at a short distance may remain. However, as described above, the fixed echo removal method is also used, so that fixed echoes that may not be removed only by the tissue harmonic imaging method are also removed. Thus, the similarity can be evaluated with high accuracy and stability.
[0053]
(Third embodiment)
An ultrasonic diagnostic apparatus according to a third embodiment of the present invention will be described with reference to FIGS.
[0054]
This ultrasonic diagnostic apparatus is intended to calculate feature amounts representing similarity between local regions, that is, to speed up pattern matching. As an overview of this pattern matching, we assume local spatiotemporal continuity of the movement of the living body, and make the movement vector in the stationary time phase and stationary part, that is, the best match search area as small as possible. It is to become. At least one of the information of “past movement vector amount” and “movement vector amount of an adjacent point that has already been calculated” or both of them are effectively used for the calculation position where matching is performed.
[0055]
In this ultrasonic diagnostic apparatus, the pattern matching process described below is executed, but this process is also preferably executed in any of the feature quantity calculators 16 shown in FIGS.
[0056]
Of course, this pattern matching processing is replaced with pattern matching processing described in the publication in an ultrasonic diagnostic apparatus other than those described in FIGS. 1 to 3, for example, an ultrasonic diagnostic apparatus described in Japanese Patent Laid-Open No. 8-164139. May be executed.
[0057]
As shown in FIG. 4, the feature amount calculator 16 described in any of FIGS. 1 to 3 includes a plurality of frame memories 31 </ b> A to 31 </ b> N, a processor 32 that calculates feature amounts, and a calculation result (for example, A memory 33 for storing cross-correlation coefficients at each calculation position and intermediate calculation information (for example, a signal power value during calculation of the cross-correlation coefficient at each calculation position). Various kinds of designation information are sent from the controller 34 to the calculator 16. The controller 34 is further connected to an operating device 35, and operation information given by the operator via the operating device 35 is sent to the controller 34.
[0058]
The controller 34 sends information designating two different time phases desired by the operator to the frame memories 31A to 31N and the processor 32, and sends information designating a pattern matching acceleration method and search range information to the processor 32. .
[0059]
FIG. 5 shows a pattern matching algorithm executed by the feature quantity calculator 16.
[0060]
The processor 32 of the feature quantity calculator 16 reads two time phases L and L + 1 having different two-dimensional scanning timings commanded by the operator via the controller 34 (step S1). In response to this, the processor 32 scans the two-dimensional cross-section scanned in the time phase that matches the designated time phases L and L + 1 from the received signal group of the plurality of frames stored in the plurality of frame memories 31A to 31N. Two sets of received signals are read and written individually in the work area (step S2). Note that the reception signals for two frames of the designated time phases L and L + 1 may be performed substantially in real time while collecting echo signals.
[0061]
Thereby, in the work area of the processor 32, as schematically shown in FIGS. 6A and 6B, the first two-dimensional section (space) formed by the received signal of the first frame F in the time phase L is shown. (Region) SC1 and a second two-dimensional section (spatial region) SC2 formed by the received signal of the second frame F + 1 at time phase L + 1 are constructed on the memory.
[0062]
Next, the processor 32 initializes a search area instructed by the operator via the controller 34 (step S3). That is, for example, when the operator sets a desired range of ROI on the same image while viewing the B-mode tomographic image collected at one time phase L + 1 and displayed on the display 18, this ROI information is stored in the initial search area. To the processor 32 (see FIG. 6B).
[0063]
The processor 32 then executes the kernel ROI: KN_ROIIs initially set, and the center position (Xi, Yi) is designated as the calculation position (step S4, see FIG. 6A).
[0064]
Next, the processor 32 adaptively (dynamically) estimates the search center coordinates (Xc, Yc) and the optimal search range sizes 2Xw, 2Yw, thereby specifying the search range SR._SPCIs set (steps S5 and S6).
[0065]
(Adaptive estimation of search center coordinates)
[Estimation of search center coordinates (Xc, Yc)]
First, the adaptive estimation process of the search center coordinates (Xc, Yc) will be described with reference to FIG. 7 (step S5).
[0066]
Now, let the current time phase be L, and for the sake of simplicity, let the current calculation position be (Xi, Yj) L, taking 2D as an example, as shown in FIG. In order to obtain the movement vector vijL = (vx, vy) ijL, first, it is necessary to set the search center coordinates (Xc, Yc) ijL for pattern matching and the search range. If the X-axis search width is 2XwijL and the Y-axis search width is 2YwijL, the search range is represented by (Xc ± Xw, Yc ± Yw) ijL.
[0067]
In order to make the search range sufficiently small, it is necessary to first bring the search center coordinates close to the movement vector vijL to some extent. Considering that vxijL and vyijL are estimated under these conditions, and assuming that the estimated values are vα and vβ, these estimated values can be set as follows according to various conditions.
[0068]
(1) When temporal continuity of myocardial motion can be expected, an estimated value can be set using “past movement vector” (vx, vy) ijL−1.
[0069]
(2) If spatial continuity can be expected, an estimated value can be set using “the amount of movement vectors of adjacent points that have already been calculated”. Neighboring points (vi-1j-1L, vij-1L, vi + ij-1L, vi-1jL) are preferable as the adjacent points that can be adapted to this, excluding the end.
[0070]
(3) Furthermore, in an actual living body, spatiotemporal continuity can be expected, so it is desirable to average all of these pieces of information with appropriate weighting to obtain an estimated value. If this is expressed in a general expression using a movement vector that uses one point in the past and four neighboring points, an estimated value can be determined as in the following equation.
[0071]
[Expression 1]

[0072]
Here, ki represents a weighting coefficient, and Σki = 1 or all ki = 0 (i = 1 to 5).
[0073]
In the present embodiment, the search center coordinate value can be estimated by arbitrarily selecting and changing the above-described conditions (1) to (3). Normally, the condition (3) is set as a default.
[0074]
In the above general expression, if k2 to k5 = 0 (k1 = 1), estimated values vα and vβ equivalent to the estimation under the condition (1) can be obtained, and k1 = 0. Then, estimated values vα and vβ equivalent to the estimation under the condition (2) can be obtained.
[0075]
Optimal device conditions according to clinical use, that is, frame (volume) rate setting (time difference between time phases changes) and calculation pitch setting change, so appropriate weight ki is set according to each condition. It is preferable to give.
[0076]
For example, the weight of k1 is increased in the case where the time phase difference is sufficiently small with respect to the movement amount of the movement vector, and in the state close to the above condition (1) and in the case where the calculation pitch is sufficiently small, k2 to k5. It is preferable that the weight is increased and estimation is performed in a state close to the condition (2).
[0077]
In the above example, an example in which a single vector is used as a past movement vector has been shown for the sake of simplicity. However, an appropriate weighting may be applied to past movement vector amounts at a plurality of adjacent points. Good.
[0078]
It is also possible to set all the coefficients ki = 0. This setting is a special case, which means that the search center coordinate is set to the calculation position. This is used in the situation where the above-mentioned various continuity cannot be assumed. For example, when the received signal in the local area to be calculated is noise, the continuity cannot be assumed at the calculation position, and it is desirable to be able to select such a setting (ki = 0). With this setting, it is possible to avoid an unnecessarily large estimation error even for noise that reduces the reliability of the motion vector calculation itself (substantially meaningless). Sex can be secured.
[0079]
In determining whether the signal at the calculation position is noise, similarity information and intermediate information in the process of calculating the similarity are used. For example, using the cross-correlation coefficient, when the absolute value of the correlation coefficient is less than or equal to a threshold value, it is regarded as noise and switched to this setting (ki = 0). Alternatively, using the power of the received signal (which occurs in the middle of the calculation of the cross-correlation coefficient. For simplicity, the denominator part for normalization in the equation for calculating the cross-correlation coefficient can be used as it is), the signal power is When it is below a certain threshold value, it is regarded as noise and switched to this setting (ki = 0). You may threshold-process combining a cross correlation coefficient and signal power.
[0080]
In this embodiment, as shown in steps S12 and S13, which will be described later, such noise determination is performed using a cross-correlation coefficient that is a calculation result at each calculation position. Thus, when the signal value is recognized as noise, the weighting coefficient ki = 0 is set and the cross-correlation coefficient is recalculated.
[0081]
Probable search center coordinates (Xc, Yc) ijL = (Xi + vα, Yj + vβ) L using the movement destination coordinate values vα and vβ of the movement vector thus estimated are determined.
[0082]
[Estimation of specified search range]
Next, referring to FIG. 7, the designated search range SR_SPCThe adaptive estimation process will be described (step S6).
[0083]
Since the movement vector is searched for as a best-matching part with the estimated coordinates (Xc, Yc) ijL as the center as described above, the smaller the search range, the higher the calculation. Of course, the similarity cannot be accurately evaluated unless the search range is set larger than the actual movement range (true value). Therefore, the optimal search range, that is, the values of 2XwijL and 2YwijL are estimated.
[0084]
In this case as well, the spatiotemporal continuity of the movement of the living body is used as in the above-described estimation of the search center coordinates. The difference from the above is that it is necessary to estimate the “upper limit of error” of the size (moving speed) of the probable moving vector. So, past movement vector
[Expression 2]

MaxV that maximizes_errUsing the search range SR_SPCIs set as follows.
[0085]
[Equation 3]

[0086]
When the coefficient ka = 1, the error of the estimated coordinate values vα and vβ is ± maxV_errThis is equivalent to expecting the range of. Since it is necessary to give a margin to the search range and to consider the acceleration component of movement, it is preferable to set the coefficient ka ≧ 1 in order to make the range slightly larger.
[0087]
The optimum designated search range SR_SPCIn addition to the setting method described above, various setting methods can be selectively adopted.
[0088]
For example, past movement vectors
[Expression 4]

Using the standard deviation vsx of the x component and the standard deviation vsy of the y component (vs itself is a scalar quantity), the following may be set.
[0089]
[Equation 5]

[0090]
In this case, it is preferable to set the coefficient kb ≧ 1 for the same purpose as the coefficient ka described above.
[0091]
In addition, it is desirable to selectively use vector information according to clinical use. For example, in the case where the temporal discontinuity is considered to be small, the past movement vector (vx, vy) ijL-1 is not used, and on the contrary, in the case where the spatial discontinuity is considered small, the calculation is not performed. The movement vectors (vi-1j-1L, vij-1L, vi + 1j-1L, vi-1jL) at the points that have already been used are not used.
[0092]
Further, when the spatial continuity is small, using the movement vector (vx, vy) ijL-1 before one time phase and (vx, vy) ijL-2 before another time phase,
[Formula 6]

The optimum designated search range SR as follows:_SPCMay be set.
[0093]
[Expression 7]

[0094]
Also in this case, it is preferable to set the coefficient kc ≧ 1 as described above.
[0095]
In this way, an appropriate search center position and an appropriate search area size can be set dynamically.
[0096]
As described above, the designated search range SR_SPCIs set as the optimum necessary minimum range, the CPU 32 of the feature quantity calculator 16 makes this designated search range SR as shown in FIG._SPCInside object ROI: OB_ROIIs set to the initial position (step S7 in FIG. 5). This object ROI: OB_ROIThe size of the kernel ROI: KN_ROISet to be the same as that.
[0097]
Next, the CPU 32 executes the kernel ROI: KN at the given calculation position of the first frame F._ROIAnd the specified object ROI of the second frame F + 1: OB_ROIThe cross-correlation coefficient as a feature representing similarity is calculated and stored in the memory 33 (step S8). This cross-correlation coefficient ρ_{m, n}When the signal of the pixel value of the kernel ROI is A and that of the object ROI is B,
[Equation 8]

Calculated based on In this calculation, the denominator value (signal power value) is also stored as intermediate information in the memory for each calculation.
[0098]
Next, the CPU 32 determines that the object ROI: OB_ROIDesignated search range SR_SPCTo determine whether the cross-correlation coefficient has been calculated by moving to all the positions within the specified search range SR_SPCIf there is an unfinished position in the object, the object ROI: OB_ROIAnd the same calculation is repeated (steps S9 and S10). On the other hand, the designated search range SR which is adaptively set to the minimum necessary._SPCWhen the coefficient calculation is completed at all the calculation positions, the CPU 32 calculates the maximum correlation value as a representative value from the plurality of cross-correlation coefficients calculated up to that point and stored in the memory 33. A number is selected (step S11).
[0099]
Next, the CPU 32 determines whether or not the absolute value of the selected cross-correlation coefficient (maximum value) is larger than a predetermined threshold value (step S12). This determination is a process for determining whether or not the signal value described above is noise. When it is determined that the signal value is noise, that is, when | coefficient |> threshold is determined, all the weight coefficients ki = The cross correlation coefficient is set again to 0 (step S13). Note that this determination may be made based on the signal power value as the calculation intermediate information stored in the memory 33 as described above.
[0100]
Conversely, when it is recognized that the cross-correlation coefficient is not based on noise but based on a signal value in a normal range (step S12; NO), the CPU 32 determines that the kernel ROI: KN_ROIIs moved to all the calculation positions to determine whether the calculation has been completed. If there is a position where the calculation has not been completed, the kernel ROI: KN_ROIIs updated (steps S14 and S15). And the designated search range SR_SPCThen, returning to the process of adaptive setting of the size, the above-described series of processes is executed for all desired calculation positions.
[0101]
Thus, the designated search range SR_SPCFor each computation position (kernel ROI: KN_ROIThe cross-correlation coefficient at each calculation position is obtained adaptively optimally for each position. For this reason, when the series of pattern matching processes described above is completed, two-dimensional distribution data of feature amounts (here, cross-correlation coefficients) representing the similarity of desired tomographic images is obtained in the memory 33. The two-dimensional distribution data of the feature amount is juxtaposed or superimposed on the tomographic image as a feature amount image by the DSC 17, and this image is displayed on the display 18 in the display mode.
[0102]
Thus, according to the present embodiment, when performing pattern matching for observing time-series similarity of moving organs such as the myocardium, the designated search range SR_SPCIs estimated and set adaptively within the minimum necessary range for each calculation position. Therefore, it is sufficient to calculate the feature quantity representing similarity by searching only the minimum necessary area. For this reason, compared to the conventional configuration in which a search ROI range that is set larger to detect the maximum moving speed included in the data is searched for every calculation position, the feature amount is obtained. The amount of computation required for this is greatly reduced, and the computation is significantly accelerated.
[0103]
In the above-described ultrasonic diagnostic apparatus, the cross-correlation coefficient is calculated one by one between the kernel ROI and the object ROI in order to obtain a feature amount representing similarity. And an arithmetic method using SAD (Sum-Absolute-Difference: sum of absolute values of mutual differences).
[0104]
SADε_{m, n}When the signals of the pixel values of both images are A and B,
[Equation 9]

It is represented by
[0105]
Since the position of the spatial minimum value obtained by the SAD method is considered to be a position with high similarity, the movement vector can be detected by using this SAD method. Although the cross-correlation method is said to be a gold standard in the field of pattern matching, the SAD method has the advantage of high-speed processing because it has fewer calculation steps than the cross-correlation method. However, the SAD method can detect a movement vector, but even if it is used alone, it is difficult to standardize because it depends on the size of the signal source in terms of quantification of similarity, and it is physically meaningful. It is difficult to obtain a certain index.
[0106]
Therefore, taking advantage of the high speed of SAD, first, a movement vector is searched by SAD, and a cross correlation coefficient is obtained at the detected movement vector position. As a result, the time-consuming cross-correlation calculation is suppressed to one time, so that the speed can be increased as compared with the case where the movement vector is searched by the cross-correlation calculation, and a physical quantity having a certain significance with respect to similarity is obtained as a cross-correlation coefficient be able to.
[0107]
When this SAD method and the cross-correlation method are executed in combination, for example, in the process of FIG. 5 described above, a process of calculating SAD between the kernel ROI and the specified search range is performed instead of steps S8 to S10. Then, two steps for calculating a cross-correlation coefficient between the object ROI and the kernel ROI around the movement destination of the movement vector obtained by SAD may be set.
[0108]
In addition to the processing of FIG. 5 described above, processing for calculating SAD is performed between the kernel ROI and the initial search range, and then between the object ROI and the kernel ROI centering on the destination of the movement vector obtained by SAD. Two steps for calculating the cross-correlation coefficient may be provided.
[0109]
By the way, the present invention can also provide a movement tracking (tracking) of a region of interest in a three-dimensional space-time that effectively uses movement vector information obtained by pattern matching calculation and a display method thereof. An embodiment relating to this display will be described below.
[0110]
(Fourth embodiment)
The fourth embodiment of the present invention will be described with reference to FIGS.
[0111]
In the present embodiment, in the feature amount calculator 16 configured in the same manner as in FIG. 4 described above, the CPU 32 cooperates with the controller 34 and the operation unit 35 in addition to the above-described feature amount calculation. A function of calculating the movement tracking of the part (tracking) is given.
[0112]
For this reason, the CPU 32 performs a series of processes schematically shown in FIG. Specifically, the initial time phase S and the end time phase E are set, the “time phase” variable = S is set, and the reception signal of the frame corresponding to the initial time phase is called from the frame memory 31, The point of interest that the operator is interested in is set in the spatial region formed by the received signal (steps S21 to S23).
[0113]
Next, after initially setting the movement vector = 0, the variable of “interest point” = interest point + V (movement vector) is set, and further, the time phase is incremented to time phase + 1 (steps S24 to S26).
[0114]
Next, the CPU 32 calculates the movement vector V of the point of interest between the received signals of two frames extending from “time phase = S + 1” to “time phase−1 = S” (step S27). This movement vector V may be calculated by a conventionally well-known method, or may be obtained by the speed-up search area setting described in the third embodiment. Next, the CPU 32 outputs the trajectory data of the movement vector V to the DSC 17, and updates the display of this trajectory (step S28).
[0115]
Thereafter, the CPU 32 repeats the above-described movement tracking process of the region of interest while updating the time phase until the end time phase E, along with the determination of whether or not time phase <end time phase E (steps S25 to S29). . When time phase = end time phase E is reached, it is determined whether another region of interest has been designated by the operator. If so, the movement tracking process is repeated retroactively to step S22 (step S30). On the other hand, if no other region of interest is specified, the movement tracking ends.
[0116]
Means for designating any one or more regions of interest on a spatial image constructed by a received signal of a frame of a certain initial time phase by means of the above movement tracking processing; and means for calculating a movement vector at any point of interest The means for calculating the movement vector by moving the calculation point (position) to the movement destination position of the movement vector is functionally configured. Since the movement tracking processing by these means is sequentially repeated until the end time phase every time phase elapses, the spatiotemporal movement of the region of interest is tracked.
[0117]
For example, as shown in FIG. 9 or FIG. 10, the trajectory accompanying this movement corresponds to a position on the image as a tracking position from the initial time phase to the designated end time phase by the DSC 17.
[0118]
The DSC 17 receives a command signal of a desired display mode given by the operator via the operation unit 35 from the controller 34, and performs image data synthesis processing to display information on the movement locus according to the command signal on the image. Do. FIG. 9 and FIG. 10 both show trajectory information superimposed on a structure (such as myocardium) on the image and show the change in position. FIG. 9 shows a two-dimensional image, and FIG. 10 shows a three-dimensional image. Respectively. Since the time phase interval (interval from the initial time phase to the end time phase) for displaying the movement trajectory of the region of interest is preferably one cardiac cycle, FIGS. 9 and 10 are also displayed as trajectory display images in one cardiac cycle. It is shown. Each image displays an image in a certain time phase (in the case of two dimensions, an image of an arbitrary cross section in a certain time phase), and shows all trajectories in the time phase period from the initial time phase to the end time phase. It is displayed using a line above, and the position is associated. In the case of the three-dimensional case, when moving to the back side of the display section, for example, it is represented by a dotted line.
[0119]
The two-dimensional or three-dimensional image data may be information representing a structure (usually called B mode in two dimensions), or information on a feature quantity (such as a cross-correlation coefficient) representing similarity. It may be.
[0120]
(Display modification 1)
Note that the DSC 17 is not limited to a configuration that displays only an image of a specific time phase when displaying an image as a background, and shows each time phase and movement trajectory from the initial time phase to the end time phase. You may employ | adopt the display made to compare one by one. This example is shown in FIGS. 11 to 13 for a two-dimensional image.
[0121]
The display example in FIG. 11 displays the moving trajectory while adding the current trajectory to the past trajectory while updating both the tomographic image and the moving trajectory of the region of interest (image is an image. This is an example in which the memory is turned over and observed.) In the figure, a cross indicates a region of interest designated by the operator.
[0122]
In the display example of FIG. 12, the tomographic image is updated, and the movement trajectory is displayed from the first time phase, while the trajectory portion corresponding to the currently displayed time phase (indicated by black circles in the figure). Is differentiated from the others (in this case as well, in this case, it is an example in which an image memory is viewed through the image memory).
[0123]
Further, in the display examples of FIGS. 13A and 13B, the characteristic two-time phase (for example, end-diastolic period and end-systolic period) tomographic images and the entire movement trajectory are displayed side by side. Each time phase part (indicated by a black circle in the figure) is differentiated and shown.
[0124]
In this way, since the spatiotemporal trajectory of the region of interest is expressed in various manners, it is possible to visually and reliably observe the state of local motion over the cardiac cycle.
[0125]
(Display modification 2)
As another display method, there is a mode in which the speed information of the movement vector is added together and displayed. This speed information is calculated from the time between the time phases and the moving distance, and this may be obtained by the CPU 32 or the DSC 17 of the feature amount calculator 16, or a separate calculator May be provided.
[0126]
The display example of FIG. 14A shows a speed indicator V in which different hues correspond to different speeds._INDAnd the color of the movement trajectory represented by the line ML is indicated by the speed indicator V_INDIt is a change based on. The display example of FIG. 6B shows a speed indicator V in which different brightness is associated with different speeds._INDAnd the brightness of the movement locus represented by the line ML is indicated by the speed indicator V._INDIt is a change based on. In the display examples shown in FIGS. 14A and 14B, the lines of the movement trajectory are slightly exaggerated in order to express the difference in color and luminance by the difference in hatching.
[0127]
Furthermore, the display example of FIG. 5C shows a speed indicator V in which different thicknesses (line widths) correspond to different speeds._INDAnd the speed indicator V indicates the thickness of the movement locus represented by the line ML._INDIt is a change based on. In the display example of FIG. 6D, the movement trajectory of the movement vector is displayed by a line segment arrow AR, and the obtained speed is reflected on the length of the line segment.
[0128]
(Display modification 3)
Furthermore, as another display method, there is an example in which the movement trajectory of the movement vector is displayed in association with the cardiac time phase. This association display is performed by, for example, capturing an electrocardiogram (ECG) signal collected by an electrocardiograph into the apparatus and causing the CPU 32 to associate the collected time phase with the electrocardiogram signal while sending the electrocardiogram signal to the DSC 17 for display. Made.
[0129]
According to the display example of FIG. 15A, ECG indicators ECG in which different hues correspond to different ECG time phases._INDAnd the ECG waveform are displayed, and the hue of the movement locus represented by the line ML is the ECG indicator ECG._INDCan be changed based on Further, according to the display example of FIG. 5B, ECG indicators ECG in which different luminances are associated with different ECG time phases._INDAnd the ECG waveform are displayed, and the luminance of the moving locus represented by the line ML is the ECG indicator ECG._INDCan be changed based on Furthermore, according to the display example of FIG. 5C, ECG indicators ECG in which different thicknesses (line widths) are associated with different ECG time phases._INDAnd the ECG waveform are displayed, and the thickness of the moving locus represented by the line ML is the ECG indicator ECG_INDCan be changed based on In the display example of FIG. 4D, the movement trajectory of the movement vector is represented by a line ML and an ECG waveform is displayed, while arrows AR1 and AR2 representing time phases are attached to both of them. These double arrows AR1 and AR2 may be displayed so as to always move while interlocking with each other, or are always stationary, and when one of them is moved to a desired position or a desired time phase, the other is one. You may make it move in conjunction with.
[0130]
(Display modification 4)
Further, as another display method, there is an example in which the addition of the speed information of the movement vector and the association of the movement locus with the cardiac time phase are performed simultaneously.
[0131]
For example, a display method combining the display shown in FIG. 14D and the display shown in FIG. 15A, a display method combining the display shown in FIG. 14A and the display shown in FIG. Can be taken.
[0132]
Furthermore, in such a combination display, the movement trajectory line ML is given without being added speed information so that only time phase information can be identified. On the other hand, correspondence with the heart time phase and speed information are provided. You may make it display as another graph or a numerical value. In the display example of FIG. 16, the speed information is given as a graph GR, and the ECG waveform is associated with the cardiac time phase by the dotted line DL.
[0133]
(Display modification 5)
As another display example, there is an example (see FIG. 17) in which the movement trajectory of the same part before and after applying stress in the stress echo examination is displayed simultaneously. In this example, both movement trajectories are displayed together with the ECG waveform, and the association between the cardiac phase and the trajectory line ML is made by a black circle.
[0134]
(Display modification 6)
As another display example, there is an example (see FIG. 18) in which movement trajectories for a plurality of segments are simultaneously displayed and can be compared. In this example, a movement trajectory line ML at four sites of interest set in the myocardium, a graph of speed information of each trajectory line ML, and an ECG waveform are shown.
[0135]
By such various display methods, the degree of movement is added to the spatiotemporal trajectory of the local wall motion of the myocardium, and it is shown in contrast to the mental phase, so that the four-dimensional aspect of the wall motion can be expressed. It can be expressed and is effective for diagnosis support of wall motion evaluation.
[0136]
(Another example for obtaining a movement vector)
In FIG. 8 described above, the movement vector is calculated one by one while updating the time phase. FIG. 19 shows another example of the movement vector acquisition. In this example, a process of reading and using a movement vector already calculated and stored is performed.
[0137]
The processing in FIG. 19 is performed by, for example, the CPU 32 or the DSC 17 of the feature amount calculator 16 and includes a similarity image generation routine and a locus display routine. In the similarity image generation routine, for example, an image of a feature quantity (for example, a cross-correlation coefficient) indicating similarity according to the method described in Japanese Patent Laid-Open No. 8-164139 is processed from the initial time phase S to the end time phase E. Is calculated and stored in the memory 33, for example. In this similarity image generation routine, the high-speed feature value calculation method described in the third embodiment and its modifications may be used. When this calculation is completed, the process proceeds to a trajectory display routine, and the movement position is sequentially read out from the memory storing the movement vector for each time phase and is displayed for tracking. Accordingly, it is possible to provide an ultrasonic diagnostic apparatus that can track the region of interest by effectively using the calculation result of pattern matching, and includes a substantial application with a small amount of calculation.
[0138]
In addition, this invention is not limited to the structure of embodiment mentioned above or a deformation | transformation form, Of course, based on a summary to a claim, it can change into a various form further. For example, the feature quantity computing unit may be configured with dedicated hardware such as an ASIC in addition to the above-described software function by the CPU.
[0140]
【The invention's effect】
  As explained above, according to the ultrasonic diagnostic apparatus according to the present invention,By increasing the speed of the pattern matching calculation, it is possible to easily achieve both the short inspection time and the more reliable three-dimensional motion evaluation of the region of interest.
[Brief description of the drawings]
FIG. 1 is a block diagram showing an outline of an ultrasonic diagnostic apparatus according to a first embodiment of the present invention.
FIG. 2 is a block diagram showing an outline of an ultrasonic diagnostic apparatus according to a second embodiment of the present invention.
FIG. 3 is a block diagram showing an outline of an ultrasonic diagnostic apparatus according to a modified embodiment of the present invention.
FIG. 4 is a block diagram showing an outline of a feature amount calculator mounted in an ultrasonic diagnostic apparatus according to a third embodiment of the present invention.
FIG. 5 is a flowchart showing an outline of pattern matching processing executed by a CPU of a feature amount calculator according to the third embodiment;
FIG. 6 is a diagram for explaining a positional relationship among a kernel ROI, an initial search range, a designated search range, and an object ROI in two frame images at different time phases.
FIG. 7 is a diagram schematically illustrating a method for adaptively setting a designated search range.
FIG. 8 is a flowchart showing a processing outline of a region of interest tracking (movement tracking) executed by a CPU of a feature amount calculator according to the fourth embodiment;
FIG. 9 is an image diagram showing a display example of tracking for two-dimensional image data.
FIG. 10 is an image diagram showing a display example of tracking for three-dimensional image data.
FIG. 11 is an image diagram showing another display example of tracking for two-dimensional image data.
FIG. 12 is an image diagram showing still another display example of tracking for two-dimensional image data.
FIG. 13 is an image diagram showing still another display example of tracking for two-dimensional image data.
FIG. 14 is a diagram showing still another display example of tracking.
FIG. 15 is a view showing still another display example of tracking.
FIG. 16 is a diagram showing still another display example of tracking.
FIG. 17 is a view showing still another display example of tracking.
FIG. 18 is a diagram showing still another display example of tracking.
FIG. 19 is a diagram showing still another display example of tracking.
[Explanation of symbols]
11 Ultrasonic probe
12 Transmission system
13 Receive beamformer
14 Signal processor
15 Fixed echo canceller
16 feature calculator
17 DSC
18 Display
21 Band pass filter
31 (31-31N) frame memory
32 CPU
33 memory
34 Controller
35 Controller

Claims (5)

Scanning means for periodically performing two-dimensional or three-dimensional scanning of the diagnostic region of the subject with an ultrasonic signal;
A signal acquisition means for obtaining a received signal of an electrical quantity corresponding to the reflected ultrasonic signal due to the scanning,
Similarity for obtaining spatial distribution information of feature amounts representing the similarity between the first and second local regions in the first and second spatial regions formed by two sets of received signals having different scanning timings. and sex acquisition unit, in the ultrasound diagnostic apparatus having a,
The similarity acquisition means calculates the feature amount by moving the second local region within a search range,
An ultrasonic diagnostic apparatus further comprising search range setting means for changing and setting the size of the search range based on vector information obtained based on the ultrasonic reflection signal . The ultrasonic diagnostic apparatus according to claim 1 ,
The similarity acquisition means, wherein moving the second local region within the search range, it calculates the feature quantity for each movement thereof,
Representative value selection means for selecting one representative value from the one or more calculated feature quantities;
Wherein the first local region is moved in said first space region, and repeating means to function the calculation means and selection means for respective moving further ultrasonic diagnostic apparatus characterized by having a. The ultrasonic diagnostic apparatus according to claim 1 or 2 ,
The search range setting means includes vector information between the first local region and the search range in the past time phase and vector information between the first local region and the search range in the current time phase. the search range ultrasonic diagnostic apparatus characterized by and Turkey to adaptively set the center coordinates and size using at least one of vector information. The ultrasonic diagnostic apparatus according to any one of claims 1 to 3 ,
The feature amount calculating means includes means for obtaining the vector information by computing the sum of absolute values of mutual differences of the signal value in the first and second local region between (SAD),
It means for obtaining a cross-correlation coefficient of the signal value as the feature amount on the basis of the vector information,
An ultrasonic diagnostic apparatus further comprising : Scanning means for periodically performing two-dimensional or three-dimensional scanning of the diagnostic region of the subject with an ultrasonic signal;
A signal acquisition means for obtaining a received signal of an electrical quantity corresponding to the reflected ultrasonic signal due to the scanning,
Similarity for obtaining two-dimensional distribution information of feature amounts representing the similarity between the first and second local regions in the first and second spatial regions formed by two sets of received signals having different scanning timings. An ultrasonic diagnostic apparatus comprising: an acquisition unit;
The similarity acquisition means includes means for obtaining a vector information by computing the sum of absolute values of mutual differences of the signal value in the first and second spatial regions between (SAD),
And means for determining a cross-correlation coefficient of the signal value as the feature amount on the basis of the vector information,
The similarity acquisition means calculates the feature amount based on the second local region moved within a search range and the vector information,
An ultrasonic diagnostic apparatus further comprising search range setting means for changing and setting the size of the search range based on the vector information .

Images