JP2004344244A - Method and apparatus for picture signal processing - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To accurately detect an effective period of an arterial input function (AIF) which is used to reconstitute a perfusion image (Perfusion Image) using, for example, a MR signal showing the concentration of a contrast agent after the contrast medium is injected through a vein. <P>SOLUTION: The AIF is smoothly processed (S3) and the maximum value of the smoothly processed AIF is acquired (S5). The first threshold value which is for acquiring a substantial start of existence of the contrast agent and the second threshold value which is for acquiring a substantial end of existence of the contrast agent are computed based on the acquired maximum value (S6 and S8). The smoothly processed AIF and the first threshold value are compared to determine the substantial start of the existence of the contrast medium, and the smoothly processed AIF and the second threshold value are compared to determine the substantial end of existence of the contrast medium (S6 and S8). <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

【００３９】
式１は、ｉ≧ｎにおいて、ｉ番目の動脈入力関数（ＡＩＦ）、Ｈ（ｉ＋ｊ）について、ｉ番目の前後−ｎ、＋ｎ個（ｊ）についての移動平均Ｈ（ｉ）’を求めることを意味している。
式２は、ｉ＜ｎにおいて、ｉ番目の動脈入力関数（ＡＩＦ）、Ｈ（ｉ＋ｊ）について、１〜ｎ個（ｊ）についての移動平均Ｈ（ｉ）’を求めることを意味している。
【００４０】
本実施の形態は、ＴＲ＝１〜２秒であり、ｎは、たとえば、４である。すなわち、１〜２秒周期で検出したＭＲ信号に基づく動脈入力関数（ＡＩＦ）について、４個ごとに移動平均処理を行う。
もちろん、ｎ＝４は例示であり、ノイズの重畳状態、ＭＲ信号検出周期などに応じてｎを決定することができる。
【００４１】
ステップ４：傾き計算
データ処理部１７０は、平滑処理した動脈入力関数（ＡＩＦ）の各点を微分して傾きＧ（ｉ）を求める。傾きＧ（ｉ）を求める意味は、後述する平滑処理したＡＩＦの最大値Ｈｍａｘおよび最小値Ｈｍｉｎを見いだすためである。
すなわち、傾きＧ（ｉ）＝０は極大値か極小値であるから、造影剤濃度が増加していく方向にそって最初にＧ（ｉ）＝０となる点を見いだせば、そこが平滑処理したＡＩＦの最大値Ｈｍａｘである。次に、最大値Ｈｍａｘから造影剤濃度が減少していく方向にそってＧ（ｉ）＝０となる点を見いだせば、そこが平滑処理したＡＩＦの最小値Ｈｍｉｎである。
【００４２】
ステップ５：最大値を探索
データ処理部１７０は、平滑処理した動脈入力関数（ＡＩＦ）を微分して求めた傾き（微分値）Ｇ（ｉ）が０となる最初の位置のＡＩＦを最大値Ｈｍａｘとして見つけ出す。
【００４３】
ステップ６：第１しきい値計算、第１パス期間の開始点の決定
データ処理部１７０は、検出した最大値Ｈｍａｘから第１のしきい値Ｈｍａｘ／ｐ１を求める。次いでデータ処理部１７０は、第１しきい値Ｈｍａｘ／ｐ１より大きい、ステップ３において平滑処理された動脈入力関数（ＡＩＦ）でこのＡＩＦの始点に一番近い位置を、実質的に造影剤が検出されたＡＩＦの第１パス期間の開始点と決定する。
本実施の形態においては、第１のしきい値Ｈｍａｘ／ｐ１を固定した絶対値ではなく、検出した最大値Ｈｍａｘを基準として求めた第１しきい値Ｈｍａｘ／ｐ１を参照して開始点を決定しているので、造影剤濃度の状況に則して最適な開始点を決定できるという利点がある。
【００４４】
第１のしきい値Ｈｍａｘ／ｐ１を規定する第１しいき値パラメータｐ１は、第１のしきい値Ｈｍａｘ／ｐ１が、被検部位において予想される最大値Ｈｍａｘの、たとえば、２５％の値とする（ｐ１＝４のとき）。
第１のしきい値Ｈｍａｘ／ｐ１が大きければ、第１パス期間の開始点を検出する信頼性が高いが、開始点の検出が遅れる。逆に、第１のしきい値Ｈｍａｘ／ｐ１が小さいと、第１パス期間の開始点を検出する信頼性が低くなるが、開始点が遅れずに検出できる。このような事情を考慮して経験的に、しいき値パラメータｐ１を決定することができる。
また、第１のしきい値Ｈｍａｘ／ｐ１は被検部位に応じて決定することができる。
さらに、第１のしきい値Ｈｍａｘ／ｐ１は動脈入力関数（ＡＩＦ）の造影剤濃度曲線のプロファイル（形状）に応じて決定することができる。
【００４５】
ステップ７：最小値を探索
データ処理部１７０は、ステップ４において平滑処理した動脈入力関数（ＡＩＦ）を微分して求めた傾きＧ（ｉ）＝０から最小値Ｈｍｉｎを求める。
上記検出した最大値Ｈｍａｘから造影剤濃度が減少していく方向にそって、Ｇ（ｉ）＝０となる点を最初の位置として見いだせば、そこがＡＩＦの最小値Ｈｍｉｎである。
【００４６】
ステップ８：第２しきい値計算、第１パス期間の終了点の決定
データ処理部１７０は、算出した最小値Ｈｍｉｎから第２のしきい値Ｈｍｉｎ／ｐ２を求める。
データ処理部１７０は次いで、ステップ３において平滑処理した動脈入力関数（ＡＩＦ）が正で（ＡＩＦ≧０）、最大値Ｈｍａｘから造影剤濃度が減少していく方向に沿って最初に現れる、第２しきい値Ｈｍｉｎ／ｐ２より小さい平滑処理されたＡＩＦの位置（点）をＡＩＦの第１パス期間の終了点と決定する。
本実施の形態においては、第２のしきい値Ｈｍｉｎ／ｐ２を固定した絶対値ではなく、検出した最小値Ｈｍｉｎを基準として終了点を決定しているので、造影剤濃度の状況に則して終了点を決定できるという利点がある。
【００４７】
第２のしきい値Ｈｍｉｎ／ｐ２を規定する第２しいき値パラメータｐ２は、第２のしきい値Ｈｍｉｎ／ｐ２が、被検部位において予想される最大値Ｈｍａｘの、たとえば、２５％の値とする（ｐ２＝４の時）。
第２のしきい値Ｈｍｉｎ／ｐ２が大きければ、第１パス期間の終了点を検出する信頼性が高いが、終了点の検出が早まり、第１パス期間が狭くなる。逆に、第２のしきい値Ｈｍｉｎ／ｐ２が小さいと、第１パス期間の終了点を検出する信頼性が低くなるが、終了点の検出が早まることはない。このような事情を考慮して、経験的に、しいき値パラメータｐ２を決定することができる。
また、第２のしきい値Ｈｍｉｎ／ｐ２は被検部位に応じて決定することができる。
さらに、第２のしきい値Ｈｍｉｎ／ｐ２は動脈入力関数（ＡＩＦ）の造影剤濃度曲線の状況に応じて決定することができる。
【００４８】
ステップ９：結果保存
データ処理部１７０は、上述した方法で求めた動脈入力関数（ＡＩＦ）の第１パス期間の開始点、終了点、最大値、最小値、および、開始点と終了点との間のＡＩＦを保存する。
保存したこれらデータは、特にＡＩＦは、データ処理部１７０における灌流画像処理などに使用される。
【００４９】
上述した処理において、ステップ５の後ステップ７の処理を行い、時間的な順序の制約がないステップ６と８とを同じ処理工程において行うこともできる。
【００５０】
第２例の処理方法
図６はデータ処理部１７０で行われる、動脈入力関数（ＡＩＦ）の第１パス期間を識別（決定）する第２例としての処理を示すフローチャートである。
図５を参照して述べた処理と、図６に図解した処理との相違は、図６に図解したステップ８Ａにおける、第２しきい値と、終了点の検出方法が異なるだけである。以下、その相違点を中心に述べる。
【００５１】
ステップ８Ａ：第２しきい値計算、第１パス期間の終了点の決定
データ処理部１７０は、算出した最大値Ｈｍａｘから第２のしきい値Ｈｍａｘ／ｐ３を求める。すなわち、図５のステップ８においては、第２のしきい値として最小値Ｈｍｉｎを用いたが、第２例においては最大値Ｈｍａｘを用いる。
さらに、データ処理部１７０は、最大値Ｈｍａｘから造影剤濃度が減少していく方向に沿って最初に現れる、上記第２しきい値Ｈｍａｘ／ｐ３より小さい平滑処理された動脈入力関数（ＡＩＦ）の位置（点）で、しかも、ステップ７において検出した最小値Ｈｍｉｎの位置より前の位置をＡＩＦの第１パス期間の終了点と決定する。
【００５２】
第２例のステップ８Ａの処理を、図５を参照して述べた第１例のステップ８の処理と比較すると、第２のしきい値は振幅が大きく、正確に検出できる最大値Ｈｍａｘを用いるので、しきい値判定の信頼性が高くなるという利点がある。
もちろん、最小値Ｈｍｉｎの位置より前の位置を動脈入力関数（ＡＩＦ）の第１パス期間の終了点としているので、さらに第１パス期間の終了点の検出の信頼性は高い。
【００５３】
第２例においても、第２のしきい値を固定した絶対値ではなく、検出した最大値Ｈｍａｘを基準として終了点を決定しているので、造影剤濃度の状況に則して終了点を決定できるという利点がある。
【００５４】
第２のしきい値Ｈｍａｘ／ｐ３を規定する第２しきい値パラメータｐ３は、第２のしきい値Ｈｍａｘ／ｐ３が、被検部位において予想される最大値Ｈｍａｘの、たとえば、２５％の値とする（ｐ３＝４の時）。
第２のしきい値Ｈｍａｘ／ｐ３が大きければ、第１パス期間の終了点を検出する信頼性が高いが、終了点の検出が早まり、第１パス期間が狭くなる。逆に、第２のしきい値Ｈｍａｘ／ｐ３が小さいと終了点の検出が早まることはない。もちろん、最小値Ｈｍｉｎより前を終了点として検出しているので、終了点の検出の信頼性が低下することはない。このような事情を考慮して、経験的に、しきい値パラメータｐ３を決定することができる。
また、第２のしきい値Ｈｍａｘ／ｐ３は被検部位に応じて決定することができる。
さらに、第２のしきい値Ｈｍａｘ／ｐ３は動脈入力関数（ＡＩＦ）の造影剤濃度曲線の状況に応じて決定することができる。
【００５５】
図６を参照して上述した処理においても、ステップ５の後ステップ７の処理を行い、時間的な順序の制約がないステップ６と８Ａとを同じ処理工程において行うこともできる。
【００５６】
処理内容の総合的な説明
図６に図解したステップ８Ａにおける処理の場合を例示して上述した処理の全体内容を述べる。
ＭＲ−コントラスト灌流画像処理においては、造影剤が静注されると、ＭＲ信号で示される造影剤濃度を示す信号（動脈入力関数（ＡＩＦ））の強度が急激に変化する。したがって、図１、図３に例示したように、造影剤濃度曲線（または動脈入力関数曲線）で示すＡＩＦの傾きが大きくなる。逆に、造影剤が血管または組織から出て行くと、ＡＩＦの信号強度が弱くなる。理想的な状態は元の画像強度に戻る（図１を参照）。しかし、造影剤が僅かに残りながら、一回、静脈を抜けて、再び動脈に入ってくる。これがノイズとなる第２パスを引き起こす。
ＡＩＦの画像強度が大きく変化する部分は、ＡＩＦの傾きは大きく変化する。したがって、本実施の形態においては、基本的に、造影剤濃度曲線におけるＡＩＦの各点について微分してその傾きを求め、その値を評価の指標として第１パス期間を検出する。
しかしながら、ＡＩＦは、非常に小さなＲＯＩ（通常、１０以下のピクセルの場合が多い）を使って算出されているためノイズの影響を大きく受け、ＡＩＦの微分値だけで第１パス期間を判定すると誤判断を招く。
【００５７】
以下、図７（Ａ）、図８、図１０に図解した例示的な動脈入力関数（ＡＩＦ）の造影剤濃度曲線を参照して具体的に述べる。
各ＭＲ信号には各種のノイズが重畳されている。複数のＭＲ信号から計算した動脈入力関数（ＡＩＦ）は部分的にノイズが除去されているが、非常に小さなＲＯＩについて計算して求めたＡＩＦであるので、図７（Ａ）、図８、図１０に例示したように、ノイズの影響は依然として残っている。
【００５８】
図７（Ａ）に図解したグラフは、図１または図３に図解した動脈入力関数（ＡＩＦ）の第１パス期間と第２パス期間とが比較的離れているが、ＭＲ信号に高周波ノイズが重畳されている状態を示す。破線ＣＶ１１で示した曲線がノイズの影響を受けない動脈入力関数（ＡＩＦ）を示し、実線ＣＶ１２で示した曲線がノイズが重畳された動脈入力関数（ＡＩＦ）を示す。
このようなランダムな高周波ノイズは、ステップ３における式１、２に示した移動平均処理によってキャンセルされて、図７（Ｂ）に実線ＣＶ１３で示したように、破線ＣＶ１１で示したノイズの影響を受けない動脈入力関数（ＡＩＦ）と同等のＡＩＦが得られる。
図７（Ｂ）に実線ＣＶ１３で示した平滑処理したＡＩＦについて微分して傾きを算出すれば、ステップ５においてａ１点を最大値Ｈｍａｘとして検出でき、ステップ６において第１しきい値Ｈｍａｘ／ｐ１からａ２点を第１パス期間の開始点と検出でき、ステップ７においてａ３から最小値Ｈｍｉｎを検出でき、ステップ８において第２しきい値Ｈｍａｘ／ｐ３からａ４点を第１パス期間の終了点と検出できる。
【００５９】
以上により、単に高周波ノイズが重畳した場合は問題なく、正確に第１パス期間を検出できることが理解されよう。したがって、単なる高周波ノイズが重畳した場合は問題ないとして、その他のノイズが発生した場合について述べる。
【００６０】
図８の実線で示した曲線ＣＶ２１は最大値の近傍に２つの極大値ｂ１、ｂ２を含む動脈入力関数（ＡＩＦ）の造影剤濃度曲線である。ＡＩＦは非常に小さなＲＯＩ（通常、１０以下のピクセルの場合が多い）を使って算出されているため、ノイズの影響を大きく受けて、曲線ＣＶ２１に例示したＡＩＦが得られる場合もある。この曲線の傾きだけを見ると、ｂ１部分が最大値を示すピーク点として検出されれば、ｂ２部分で微分値が逆転するので、ｂ２点が第１パス期間の終点として検出される。
このようなノイズの影響による誤決定を回避するため、ステップ４におけるＡＩＦを微分してＡＩＦの傾きを求める前に、ステップ３において、ＡＩＦの平滑処理を行う。その結果を破線で示した曲線ＣＶ２２として示す。曲線ＣＶ２２においては、２つの極大値ｂ１、ｂ２の間に新たな１つの極大値ｂ３が内挿され、全体のＡＩＦも平滑になっている。しかしながら、元のＡＩＦが失われたわけではない。したがって、図８に例示したように、第１しきい値Ｈｍａｘ／ｐ１および第２しきい値Ｈｍａｘ／ｐ３を用いて、第１パス期間を正確に決定できる。
【００６１】
図９は図８を参照して述べた前処理をした後、画像再構成した脳頭部の脳血流（ＣＢＦ）についての灌流画像の例を示す。図解の限界により正確な図解ではないが、斜線部分が赤で表示されたＣＢＦを示す。図８の曲線ＣＶ２１のような動脈入力関数（ＡＩＦ）でも、ＣＢＦが鮮明に画像化されている。
図９に図解したＣＢＦをプロトン放射断層画像（Ｐｒｏｔｏｎ Ｅｍｉｓｓｉｏｎ Ｔｏｍｏｇｒａｐｈｙ：ＰＥＴ） によるＣＢＦと比較すると、ＰＥＴによるＣＢＦと非常に高い相関を持つことが分かった。
【００６２】
図１０の実線で示した曲線ＣＶ３１は、ノイズにより動脈入力関数（ＡＩＦ）の立ち上がり部分が瞬間的に０以下に低下しているＡＩＦの造影剤濃度曲線である。このようなノイズは、ステップ３におけるＡＩＦの平滑処理により、破線で示した曲線ＣＶ３２のごとく修正される。ステップ４以降の処理によって、上記同様、正確に第１パス期間を検出することができる。
【００６３】
他の応用例
本発明の画像信号処理方法は、動脈入力関数（ＡＩＦ）の造影剤濃度曲線だけでなく、他の特定期間の検出にも適用できる。
また本発明の画像信号処理方法は、ＡＩＦだけではなく、筋肉組織など他の造影剤濃度曲線にも応用できる。
【００６４】
また上述した実施の形態においては、動脈入力関数（ＡＩＦ）について述べたが、本発明の画像信号処理方法は、基本的に１ピクセルごとの信号についても適用できる。すなわち、本発明の画像信号処理方法によれば、ピクセルごとに（Ｐｉｘｅｌ ｂｙ Ｐｉｘｅｌで）第１パス期間を正確に検出できる。
【００６５】
さらに本発明は、ＭＲＩ装置に限らず、造影剤を用いたＣＴ−灌流画像処理における前処理として、第１パス期間の検出に適用でき、ＣＴ−灌流画像処理においても、上述したＭＲ−灌流画像処理におけると同様の問題、すなわち、第１パス期間の決定の不正確さ、および、それに起因する脳血流（ＣＢＦ）マップの精度の不足を改善することができる。
【００６６】
【発明の効果】
本発明によれば、ＭＲコントラスト灌流画像信号処理方法、ＣＴコントラスト灌流画像信号処理方法などに用いる原検出信号が各種のノイズの影響を受けている場合でも、平滑処理によりノイズの影響を排除して、さらに、適切なしきい値を設定し、正確に、かつ、自動的に、第１パス期間を検出できる。
さらに本発明によれば、このように前処理して正確に検出した第１パス期間の信号を用いることにより、灌流画像の精度を向上させることができる。
【図面の簡単な説明】
【図１】図１は造影剤濃度・時間変化グラフである。
【図２】図２は動脈内血液の流速、静脈内血液の流速、筋肉組織内血液の流速を図解したグラフである。
【図３】図３は動脈、静脈、筋肉組織など脳全体の平均組織濃度曲線ＣＶＡＶと、動脈入力関数（ＡＩＦ）の曲線ＣＶＡＩＦ を示すグラフである。
【図４】図４は本発明の第１実施の形態としてのＭＲＩ装置の構成図である。
【図５】図５は図４に図解したデータ処理部で行われる動脈入力関数（ＡＩＦ）の第１パス期間を検出する第１例としての処理を示すフローチャートである。
【図６】図６は図４に図解したデータ処理部で行われる動脈入力関数（ＡＩＦ）の第１パス期間を検出する第２例としての処理を示すフローチャートである。
【図７】図７（Ａ）はＭＲ信号に高周波ノイズが重畳したときの動脈入力関数（ＡＩＦ）の造影剤濃度曲線を示すグラフであり、図７（Ｂ）は本発明の平滑処理したＡＩＦの造影剤濃度曲線を示すグラフである。
【図８】図８は最大値の近傍に２つの極大値ｂ１、ｂ２を含む動脈入力関数（ＡＩＦ）の造影剤濃度曲線である。
【図９】図９は図８の動脈入力関数（ＡＩＦ）を処理して得られた第１パス期間の信号を用いて生成した脳血流（ＣＢＦ）を示す画像を概略的に図解した図である。
【図１０】図１０は動脈入力関数（ＡＩＦ）の立ち上がり部分にノイズによる瞬間的な低下が見られるＡＩＦとその平滑処理結果を示すグラフである。
【符号の説明】
１００・・マグネットシステム
１０２・・主磁場コイル部
１０６・・勾配コイル部
１０８・・ＲＦコイル部
１３０・・勾配コイル駆動部 １４０・・ＲＦコイル駆動部
１５０・・データ収集部 １６０・・制御部
１７０・・データ処理部 １８０・・表示部
１９０・・操作部
３００・・被検体 ５００・・クレードル[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to an image signal processing method and an image signal processing device using the same.
In particular, the present invention provides a magnetic resonance imaging (MRI) apparatus configured to form an image using an MR signal detected by using a magnetic resonance (MR) phenomenon, and an X-ray tomographic imaging signal. The present invention relates to a diagnostic image signal processing device such as a CT (Computed Tomography) device and a diagnostic image signal processing method.
Specifically, the present invention relates to perfusion image processing (Perfusion Imaging) in which, for example, in an MRI apparatus, an image is formed using a change in the intensity of a signal indicating a contrast agent obtained when a contrast agent is injected into a blood vessel. As preprocessing, an image signal processing method for accurately and automatically detecting a first pass (first wave) period of an arterial input function (AIF), and applying such an image signal processing method MRI apparatus. Similarly, according to the present invention, the first pass period is accurately set as a pre-process for performing image processing using information indicating the presence of a contrast agent included in an X-ray tomographic imaging signal obtained when a contrast agent is injected into a blood vessel. In addition, the present invention relates to an image signal processing method for automatically detecting an image signal and a CT apparatus to which such an image signal processing method is applied.
[0002]
[Prior art]
In the MRI apparatus, as a method of generating a perfusion image (Perfusion Image) using an MR signal, (1) a method using a change in the intensity of a signal due to injection of a contrast agent, (2) a pulse is applied from the outside, thereby There are known a method of observing the flow of labeled spin, (3) a method of using oxyhemoglobin and an intrinsic signal intensity change due to a change in oxyhemoglobin (BOLD method), and the like.
In particular, the present invention relates to a method that utilizes a change in signal intensity due to a contrast agent injection.
[0003]
Contrast-enhanced MRA (or MR-Contrast-Perfusion Imaging) is known in which a contrast agent is injected intravenously (intravenously) and the contrast of the blood in the artery due to the T1 shortening effect is imaged. ing.
In contrast-enhanced MRA, an image close to ordinary angiography can be obtained by a T1 shortening effect that depends on the concentration of gadolinium (Gd) used as a contrast agent. Therefore, cerebral blood flow (CBF) can be imaged by contrast-enhanced MRA, and a perfusion image of the whole brain can be obtained. Using such a perfusion image, each part of the brain can be imaged. Various diagnoses have been made.
The MR-contrast perfusion image signal processing method also has the advantage that the subject is not exposed and the examination time is short, and is attracting attention as a quantification method.
[0004]
On the other hand, the MR-contrast perfusion image signal processing method has encountered many problems to be overcome to establish a quantification method, such as insufficient accuracy of the MR-contrast perfusion image.
[0005]
[Non-patent document 1]
Katrin A. Rempp, Etc: Quantification of Regional Cerebral Blood Flow and Volume with Volume Dynamic Susceptibility Contrast-enhanced, Radiology, No. 19, No. 3, No. 19, No. 19, No. 19, No. 34, No. 19, No. 19, No. 19, No. 19, No. 19, No. 19, No. 34, No. 19, No. 19, No. 19, No. 19, No. 19, No. 34, No. 19, No. 19, No. 19, No. 19, No. 19, No. 19, No. 19, No. 19, No. 19, No. 34, No. 19, No. 19, No. 19, No. 19, No. 19, No. 19, No. 19, No. 19, No. 19, No. 19, No. 19, No. 19, Revolution, Etc.
[Non-patent document 2]
Shigeki Aoki: MRI perfusion image. Image Diagnosis, Vol. 19, No10, 1999
[0006]
[Problems to be solved by the invention]
A quantification method will be described as one of the problems of the MR-contrast perfusion image signal processing method.
As a quantification method, a quantification analysis method using a deconvolution method (Deconvolution) is considered to be an effective method. However, in the quantitative analysis of the MR-perfusion image, the determination of the arterial input function (AIF) greatly affects the result of the quantitative analysis. In other words, in the quantitative analysis based on the perfusion image processing, the accuracy of the arterial input function (AIF) greatly affects the quantitative value of the cerebral blood flow (CBF).
[0007]
FIG. 1 shows how an arterial input function (AIF) (or contrast agent concentration) obtained by collecting MR signals indicating the presence of a contrast agent in a cerebral artery for a plurality of pixels when an intravenous contrast agent is injected changes over time. 6 is a contrast agent concentration / time change graph (or a contrast agent concentration curve graph or a tissue concentration curve graph) showing whether the change occurs. The abscissa indicates the passage of time, and the ordinate indicates the arterial input function (AIF) (contrast agent concentration).
When a predetermined time elapses from the time point t = 0 when the contrast medium is intravenously injected, the contrast medium once passes through the artery of the brain as a first pass, and then a small amount of the contrast medium passes through the vein and the heart. Again, it enters the brain artery as a second pass. The minute amount of the contrast agent of the second wave becomes noise on the image signal.
[0008]
In the quantitative analysis using the perfusion image, the tissue concentration is usually considered using the arterial input function (AIF) of the first pass period for each pixel or a plurality of pixels. Therefore, in order to accurately perform quantitative analysis based on perfusion image processing, it is necessary to exclude the AIF in the second pass period as noise and accurately detect the AIF in the first pass period. In other words, the first pass (first wave signal) period and the second pass (second wave signal) period in the tissue concentration curve (contrast agent concentration curve) are divided (identified), and the second pass period It is known that eliminating the influence of AIF improves the accuracy of quantitative analysis results.
[0009]
As a method of identifying the first pass period, as shown in FIG. 1, an arterial input function (AIF) for each pixel or a plurality of pixels is used as a tissue concentration curve (contrast agent concentration curve). This is a method in which a doctor, a technician, or the like (hereinafter, referred to as an operator) who displays a tissue concentration curve on a display device and uses the MRI apparatus observes the displayed tissue concentration curve and visually determines the first pass period.
However, such a method cannot avoid the variation caused by the individual difference of the operator, the degree of training (or the degree of proficiency), etc. in the determination of the start position and the end position of the first pass period. I have encountered the problem of variation. In particular, if the first pass period and the second pass period are close to each other, it becomes difficult to identify them.
In addition, various high-frequency noises are superimposed on the MR signal and the waveform is disturbed, and the operator discriminates the level at which the contrast agent does not substantially exist and the position at which the contrast agent concentration reaches the maximum value. Can be difficult.
[0010]
As a method of eliminating the influence of noise superimposed on each MR signal described above, an average tissue density curve of the entire brain is obtained by averaging the MR signal for the entire brain and canceling the noise. Attempts have also been made to manually determine the first pass period. However, such an approach raises the following problems.
As illustrated in FIG. 2, actually, the flow rate of the blood in the artery illustrated in the curve CV1 is different from the flow rate of the blood in the vein illustrated in the curve CV2, the flow rate of the blood in the muscle tissue illustrated in the curve CV3, and the like. Faster than the organization. That is, the first path to the arterial input function (AIF) arrives much earlier than the first path of other tissues, such as veins and muscle tissue. Therefore, the second pass period of the contrast agent flowing through the artery also arrives earlier than the second pass period of another tissue. Therefore, when the quantitative analysis is performed using the first pass period of the vein or muscle tissue, the second pass period of the AIF is also included in the first pass period. As a result, it has a serious adverse effect on the results of the quantitative analysis.
[0011]
FIG. 3 is a graph showing an average tissue concentration curve CVAV of the whole brain such as arteries, veins, and muscle tissues, and a curve CVAIF of an arterial input function (AIF).
When the first pass period is determined for the arterial input function (AIF) using the average tissue concentration curve CVAV of the whole brain, the determined first pass period is not suitable for determining the first pass period of the AIF.
As described above, the whole brain average is not suitable for determining the first pass period of the AIF, and if it is used, it is impossible to avoid an error in the quantification value due to an artificial cause.
[0012]
As a method of overcoming the above-mentioned problem caused by noise, signal processing using a computer or the like can be considered. However, various noises are superimposed on the MR signal, and there is a need for a solution that can cope with the various types of noise even when performing signal processing by a computer. Various noise modes will be described in the embodiments.
[0013]
As described above, the perfusion image processing for the arterial input function (AIF) of the brain part in the MRI apparatus has been mainly described. However, the same problem as described above is encountered in other test sites.
[0014]
Although the identification of the first pass period in the MRI apparatus has been described by way of example, the above-described problem is not limited to the MRI apparatus. For example, the brain part and other test sites are injected using a contrast agent by using a CT apparatus. The same applies to the case of CT scanning.
Of course, a similar problem is encountered in image preprocessing in another diagnostic apparatus that performs image reconstruction by intravenous injection of a contrast agent other than the MRI apparatus and the CT apparatus.
[0015]
An object of the present invention is to generate an image in which the state is detected by injecting a contrast agent, when a signal in a second pass (second wave) period becomes a noise component in a time period, or when noise is included in each signal. Is superimposed accurately and automatically for each pixel of the image or for each minute unit period to be diagnosed, such as a plurality of pixels, of the first pass of the contrast agent substantially. It is an object of the present invention to provide an image signal processing method and an image signal processing device which can identify (determine) a period from the start to the end of the existence (first pass period).
[0016]
It is another object of the present invention to accurately identify the first pass period as described above, and to improve the accuracy of quantitative analysis using a reconstructed image result by performing image processing using a signal using the first pass period. is there.
[0017]
[Means for Solving the Problems]
According to the first aspect of the present invention, (a) a contrast agent concentration signal indicating a contrast agent concentration into which a contrast agent is intravenously injected is continuously input, and (b) the continuously input image is reconstructed. The contrast agent density signal for one pixel or a plurality of pixels at that time is subjected to smoothing processing in accordance with the lapse of the continuously input time, and (c) the maximum value of the smoothed contrast agent density signal is obtained; With reference to the maximum value of the determined contrast agent concentration signal, a first threshold value for determining the start of substantial presence of the contrast agent included in the continuously input contrast agent concentration signal, and Calculating a second threshold value for determining the end of the substantial presence of the contrast agent contained in the continuously input contrast agent concentration signal; and (e) calculating the smoothed contrast agent concentration signal and the first contrast agent concentration signal. Determine the onset of substantial presence of the contrast agent by comparing the thresholds of The determining termination of a substantial presence of the contrast agent by comparing the smoothed contrast agent concentration signal and the second threshold value, the image signal processing method is provided.
[0018]
According to a second aspect of the present invention, a means for continuously inputting a contrast agent concentration signal indicating a contrast agent concentration into which a contrast agent has been intravenously injected, and one pixel for reconstructing the continuously input image Means for smoothing the contrast agent density signal for one minute or a plurality of pixels in accordance with the lapse of the continuously input time, means for obtaining the maximum value of the smoothed contrast agent density signal, and the obtained contrast agent density A first threshold value for determining the start of substantial presence of the contrast agent contained in the continuously input contrast agent concentration signal with reference to the maximum value of the signal, and the continuously input contrast agent Means for calculating a second threshold value for determining the end of substantial presence of the contrast agent contained in the density signal, and comparing the smoothed contrast agent density signal with the first threshold value Determine the onset of substantial presence of the contrast agent , And means for determining the end of the substantial presence of the contrast agent by comparing the smoothed contrast agent concentration signal and the second threshold value, the image signal processing apparatus is provided.
[0019]
BEST MODE FOR CARRYING OUT THE INVENTION
An MRI apparatus and a signal processing method in the MRI apparatus will be described as an example of the image signal processing apparatus and the image signal processing method of the present invention.
[0020]
FIG. 4 is a configuration diagram of the MRI apparatus according to the first embodiment of the present invention.
The MRI apparatus illustrated in FIG. 4 includes a magnet system 100, a data collection unit 150, an RF coil driving unit 140, a gradient coil driving unit 130, a control unit 160, a data processing unit 170, a display unit 180, An operation unit 190.
[0021]
The magnet system 100 includes a main magnetic field coil unit 102, a gradient coil unit 106, and an RF (Radio-Frequency) coil unit 108, and is installed in a scan room that is shielded from electromagnetic waves and magnetism.
The main magnetic field coil unit 102, the gradient coil unit 106, and the RF (Radio-Frequency) coil unit 108 are formed in a substantially cylindrical shape, and a human body 300 as a subject is placed in the internal space (bore) of the magnet system 100. Is mounted on the cradle 500, and is moved by a carrying means (not shown) together with the cradle 500 so as to be able to carry in and out in accordance with a portion to be inspected.
[0022]
The main magnetic field coil unit 102 forms a static magnetic field in the bore of the magnet system 100. The direction of the static magnetic field is substantially parallel to the body axis direction of the human body 300 as the subject, and forms a horizontal magnetic field.
In the present embodiment, an example is shown in which main magnetic field coil section 102 is formed of a permanent magnet such as a superconducting magnet. Therefore, a drive unit for the main magnetic field coil unit 102 is not provided.
[0023]
The gradient coil driving unit 130 drives the gradient coil unit 106 under the control of the control unit 160 to make the static magnetic field strength formed in the magnet system 100 have a gradient (incline). There are three types of gradient magnetic fields generated by the gradient coil unit 106: a slice gradient magnetic field, a readout gradient magnetic field, and a phase concord gradient magnetic field. It has a coil.
[0024]
The RF coil unit 108 has a transmission coil and a reception coil. The transmission coil and the reception coil may each be provided with two dedicated coils, or both may be shared and only one coil may be provided. The RF coil driving unit 140 drives (exciting) the transmission coil of the RF coil unit 108 under the control of the data processing unit 170, and moves the human body 300, which is the subject, into the static magnetic field space of the magnet system 100. To form a high-frequency magnetic field for exciting the spins. The receiving coil of the RF coil unit 108 detects a magnetic resonance (MR) signal, which is an electromagnetic wave generated by the excited spin.
[0025]
The data collection unit 150 inputs (collects) the MR signal detected by the reception coil of the RF coil unit 108 and outputs the data to the data processing unit 170 under the control of the control unit 160.
[0026]
The data processing unit 170 has a computer, and performs various operations related to the MRI process according to various programs stored in a memory of the computer. The data processing unit 170 stores, for example, the MR signals collected by the data collection unit 150 in a memory in the data processing unit 170 in which a data space of a two-dimensional Fourier space is defined, and uses the MR signals stored in the memory. Then, for example, various signal processings including perfusion image processing by the maximum brightness projection method (MIP) or the like are performed.
Hereinafter, in the present specification, the data processing unit 170 will only describe processing related to perfusion image processing.
The reconstructed image obtained by processing in the data processing unit 170 is displayed on the display unit 180.
[0027]
The control unit 160 cooperates with the data processing unit 170 to control the gradient coil driving unit 130, the RF coil driving unit 140, and the data collection unit 150, and to perform imaging of a test site of the human body 300 as a subject. Perform control.
[0028]
The operation unit 190 is used by a doctor, a technician, or the like (hereinafter, an operator) using the MRI apparatus to instruct a desired operation process. The data processing unit 170 and the control unit 160 cooperate with each other to process the content specified by the operation unit 190.
[0029]
Schematic operation of MRI system
An operator who operates the MRI apparatus instructs a desired MRI operation from the operation unit 190.
The data processing unit 170, together with the control unit 160 in response to an instruction from the operation unit 190, generates a slice gradient magnetic field and a readout gradient magnetic field within the static magnetic field generated by the main magnetic field coil unit 102 via the gradient coil driving unit 130. , Generate a phase concord gradient magnetic field and the like. In addition, the data processing unit 170 excites the transmission coil via the RF coil driving unit 140 together with the control unit 160 in response to an instruction from the operation unit 190, and causes the static magnetic field space of the magnet system 100 to move the inside of the human body 300, A high-frequency magnetic field for exciting spins is formed.
The data processing unit 170 and the control unit 160 drive the transporting means (not shown) in accordance with the part to be examined of the human body 300 as the subject to move the cradle 500 into the bore of the magnet system 100.
[0030]
For example, a doctor injects (intravenously) an MR contrast agent, for example, a contrast agent of a gadolinium (Gd) compound into a human body 300 as a subject intravenously. Although Gd is not directly reflected on the MR image, it promotes the relaxation of hydrogen protons (protons) in the tissue, and its presence is indirectly reflected on the MR image. Therefore, the MR signal detected after injecting the contrast agent indirectly indicates the position and concentration of the contrast agent. Typically, MR signals for very small adjacent ROTs of 10 pixels or less are indicative of the arterial input function (AIF).
[0031]
The receiving coil of the RF coil unit 108 continuously detects an MR signal which is an electromagnetic wave generated by the excited spin.
The data collection unit 150 continuously inputs the MR signal continuously detected by the reception coil of the RF coil unit 108 at, for example, a period of 1 to 2 seconds (TR = 1 to 2 seconds), and sends the MR signal to the data processing unit 170. Output continuously.
As described above, the data processing unit 170 operates various programs related to the MRI processing stored in the memory of the computer that constitutes the data processing unit 170, and converts the MR signals collected by the data collection unit 150 into data processing units. The data is stored in a memory in which the data space of the two-dimensional Fourier space in 170 is defined, and using the MR signal stored in the memory, for example, perfusion image processing is performed by a method such as maximum intensity projection (MIP). And performs various signal processing. The data processing unit 170 displays the perfusion image on the display unit 180.
[0032]
In the present embodiment, basically, the first pass period of the contrast agent density for one pixel of the reconstructed image is determined.
However, in practice, the arterial input function (AIF) is used, so in the present embodiment, the MR signal continuously input to the data processing unit 170 will be described below. The AIF is calculated for a very small ROI for the pixel to be stored, stored in the memory of the data processing unit 170, and the first pass period is determined for the obtained AIF.
[0033]
Processing method of the first example
Hereinafter, as the preprocessing of the perfusion image processing performed in the data processing unit 170, the processing content regarding the determination of the first pass period of the arterial input function (AIF) will be described.
FIG. 5 shows image pre-processing performed on the MR signals continuously collected by the data collection unit 150 and continuously input to the data processing unit 170, that is, the arterial input function (AIF) performed by the data processing unit 170. 9 is a flowchart illustrating a process as a first example of identifying (determining) the first pass period of FIG.
[0034]
Step 1: MR signal input
The data processing unit 170 continuously collects data at a predetermined period, for example, at a period of TR = 1 to 2 seconds by the data collection unit 150, and shows the MR concentration indicating the contrast agent concentration continuously output to the data processing unit 170. The signal is input and stored in a memory in the data processing unit 170.
[0035]
Step 2: AIF calculation
For the MR signals continuously stored in the memory of the data processing unit 170, the arterial input function (AIF) is calculated for a very small ROI for adjacent pixels of 10 pixels or less, and the memory of the data processing unit 170 is calculated. To remember again.
[0036]
When the data processing unit 170 continuously receives the MR signals from the data collection unit 150 and stores the MR signals in a memory in the data processing unit 170, the data processing unit 170 performs the above-described calculation of the arterial input function (AIF) to perform data processing. The information may be stored in the memory of the unit 170. That is, the processing of step 1 and the processing of step 2 can be performed continuously as one step.
[0037]
Step 3: AIF smoothing process
The data processing unit 170 performs a moving average process on the obtained arterial input function (AIF) by applying Expressions 1 and 2, and performs an AIF smoothing process.
[0038]
(Equation 3)

[0039]
Equation 1 indicates that, for i ≧ n, a moving average H (i) ′ for the i-th preceding and following −n, + n (j) for the i-th artery input function (AIF), H (i + j) is obtained. Means.
Equation 2 means that, for i <n, a moving average H (i) ′ for 1 to n (j) is obtained for the i-th artery input function (AIF), H (i + j).
[0040]
In the present embodiment, TR = 1 to 2 seconds, and n is 4, for example. That is, a moving average process is performed on the arterial input function (AIF) based on the MR signal detected at a period of 1 to 2 seconds for every four.
Of course, n = 4 is an example, and n can be determined according to the superimposed state of the noise, the MR signal detection cycle, and the like.
[0041]
Step 4: Tilt calculation
The data processing unit 170 obtains a gradient G (i) by differentiating each point of the smoothed artery input function (AIF). The meaning of obtaining the gradient G (i) is to find the maximum value Hmax and the minimum value Hmin of the AIF subjected to the smoothing process described later.
That is, since the gradient G (i) = 0 is a maximum value or a minimum value, if a point where G (i) = 0 is first found along the direction in which the contrast agent concentration increases, the point is smoothed. This is the maximum value Hmax of the obtained AIF. Next, when a point where G (i) = 0 is found along the direction in which the contrast agent concentration decreases from the maximum value Hmax, that point is the minimum value Hmin of the smoothed AIF.
[0042]
Step 5: Search for maximum value
The data processing unit 170 finds, as the maximum value Hmax, the AIF at the first position where the gradient (differential value) G (i) obtained by differentiating the smoothed arterial input function (AIF) becomes 0.
[0043]
Step 6: Calculate the first threshold value and determine the start point of the first pass period
The data processing unit 170 obtains a first threshold value Hmax / p1 from the detected maximum value Hmax. Next, the data processing unit 170 detects the position closest to the start point of the AIF in the artery input function (AIF) smoothed in step 3 that is larger than the first threshold value Hmax / p1, and the contrast agent substantially detects the position. It is determined as the start point of the first pass period of the AIF.
In the present embodiment, the starting point is determined with reference to the first threshold value Hmax / p1 obtained based on the detected maximum value Hmax, instead of the absolute value where the first threshold value Hmax / p1 is fixed. Therefore, there is an advantage that an optimum starting point can be determined according to the situation of the contrast agent concentration.
[0044]
The first threshold value parameter p1 that defines the first threshold value Hmax / p1 is such that the first threshold value Hmax / p1 is, for example, a value of 25% of the maximum value Hmax expected at the test site. (When p1 = 4).
If the first threshold value Hmax / p1 is large, the reliability of detecting the start point of the first pass period is high, but the detection of the start point is delayed. Conversely, if the first threshold value Hmax / p1 is small, the reliability of detecting the start point of the first pass period is low, but the start point can be detected without delay. The threshold parameter p1 can be empirically determined in consideration of such circumstances.
In addition, the first threshold value Hmax / p1 can be determined according to the test site.
Further, the first threshold value Hmax / p1 can be determined according to the profile (shape) of the contrast agent concentration curve of the arterial input function (AIF).
[0045]
Step 7: Search for minimum value
The data processing unit 170 obtains the minimum value Hmin from the gradient G (i) = 0 obtained by differentiating the artery input function (AIF) smoothed in step 4.
If the point where G (i) = 0 is found as the first position along the direction in which the contrast agent concentration decreases from the detected maximum value Hmax, that point is the minimum value Hmin of the AIF.
[0046]
Step 8: Calculate the second threshold, determine the end point of the first pass period
The data processing unit 170 calculates a second threshold value Hmin / p2 from the calculated minimum value Hmin.
Next, the data processing unit 170 determines that the artery input function (AIF) smoothed in step 3 is positive (AIF ≧ 0), and appears first along the direction in which the contrast agent concentration decreases from the maximum value Hmax. The position (point) of the AIF subjected to the smoothing processing smaller than the threshold value Hmin / p2 is determined as the end point of the first pass period of the AIF.
In the present embodiment, the end point is determined not based on the absolute value of the second threshold value Hmin / p2 but on the basis of the detected minimum value Hmin. There is an advantage that the end point can be determined.
[0047]
The second threshold value parameter p2 that defines the second threshold value Hmin / p2 is such that the second threshold value Hmin / p2 is, for example, a value of 25% of the maximum value Hmax expected at the test site. (When p2 = 4).
If the second threshold value Hmin / p2 is large, the reliability of detecting the end point of the first pass period is high, but the end point is detected earlier and the first pass period becomes narrower. Conversely, when the second threshold value Hmin / p2 is small, the reliability of detecting the end point of the first pass period is reduced, but the end point is not detected earlier. In consideration of such circumstances, the threshold value parameter p2 can be empirically determined.
Further, the second threshold value Hmin / p2 can be determined according to the test site.
Further, the second threshold value Hmin / p2 can be determined according to the state of the contrast agent concentration curve of the arterial input function (AIF).
[0048]
Step 9: Save results
The data processing unit 170 stores the start point, the end point, the maximum value, the minimum value of the first pass period of the arterial input function (AIF) obtained by the above-described method, and the AIF between the start point and the end point. I do.
These stored data, especially AIF, is used for perfusion image processing in the data processing unit 170 and the like.
[0049]
In the above-described processing, the processing of step 7 can be performed after step 5, and steps 6 and 8, which have no restriction on the temporal order, can be performed in the same processing step.
[0050]
Processing method of the second example
FIG. 6 is a flowchart illustrating a process performed by the data processing unit 170 as a second example of identifying (determining) the first pass period of the arterial input function (AIF).
The only difference between the processing described with reference to FIG. 5 and the processing illustrated in FIG. 6 is that the second threshold value and the method of detecting the end point in step 8A illustrated in FIG. 6 are different. Hereinafter, the differences will be mainly described.
[0051]
Step 8A: Calculation of Second Threshold, Determination of End Point of First Pass Period
The data processing unit 170 calculates a second threshold value Hmax / p3 from the calculated maximum value Hmax. That is, in step 8 of FIG. 5, the minimum value Hmin is used as the second threshold value, but in the second example, the maximum value Hmax is used.
Further, the data processing unit 170 calculates the smoothed arterial input function (AIF) smaller than the second threshold value Hmax / p3, which first appears along the direction in which the contrast agent concentration decreases from the maximum value Hmax. The position (point) and the position before the position of the minimum value Hmin detected in step 7 is determined as the end point of the first pass period of the AIF.
[0052]
Comparing the processing of step 8A of the second example with the processing of step 8 of the first example described with reference to FIG. 5, the second threshold uses a maximum value Hmax that has a large amplitude and can be detected accurately. Therefore, there is an advantage that the reliability of the threshold value determination is increased.
Of course, since the position before the position of the minimum value Hmin is set as the end point of the first pass period of the arterial input function (AIF), the detection of the end point of the first pass period is more reliable.
[0053]
Also in the second example, the end point is determined based on the detected maximum value Hmax instead of the absolute value in which the second threshold value is fixed. Therefore, the end point is determined based on the state of the contrast agent concentration. There is an advantage that you can.
[0054]
The second threshold value parameter p3 that defines the second threshold value Hmax / p3 is such that the second threshold value Hmax / p3 is, for example, a value of 25% of the maximum value Hmax expected at the test site. (When p3 = 4).
If the second threshold value Hmax / p3 is large, the reliability of detecting the end point of the first pass period is high, but the end point is detected earlier and the first pass period is narrowed. Conversely, if the second threshold value Hmax / p3 is small, the detection of the end point is not accelerated. Of course, since the end point is detected before the minimum value Hmin, the reliability of detection of the end point does not decrease. Considering such circumstances, the threshold parameter p3 can be empirically determined.
Further, the second threshold value Hmax / p3 can be determined according to the test site.
Further, the second threshold value Hmax / p3 can be determined according to the state of the contrast agent concentration curve of the arterial input function (AIF).
[0055]
In the processing described above with reference to FIG. 6 as well, the processing in step 7 after step 5 can be performed, and steps 6 and 8A with no restriction on the temporal order can be performed in the same processing step.
[0056]
Comprehensive explanation of processing contents
The entire contents of the above-described processing will be described by exemplifying the case of the processing in step 8A illustrated in FIG.
In MR-contrast perfusion image processing, when a contrast agent is injected intravenously, the intensity of a signal (arterial input function (AIF)) indicating the concentration of the contrast agent indicated by the MR signal rapidly changes. Therefore, as illustrated in FIGS. 1 and 3, the slope of the AIF indicated by the contrast agent concentration curve (or the arterial input function curve) increases. Conversely, as the contrast agent leaves the blood vessel or tissue, the signal intensity of the AIF decreases. The ideal state returns to the original image intensity (see FIG. 1). However, once the contrast agent slightly remains, it passes through the vein once and reenters the artery. This causes the second path to become noise.
In a portion where the image intensity of the AIF changes greatly, the inclination of the AIF changes greatly. Therefore, in the present embodiment, basically, each point of the AIF in the contrast agent concentration curve is differentiated to obtain its slope, and the first pass period is detected using the value as an index for evaluation.
However, since the AIF is calculated using a very small ROI (usually, the number of pixels is usually 10 or less), the AIF is greatly affected by noise, and it is erroneous to determine the first pass period using only the differential value of the AIF. Invite judgment.
[0057]
Hereinafter, a specific description will be given with reference to the contrast agent concentration curves of the exemplary arterial input function (AIF) illustrated in FIGS. 7A, 8, and 10.
Various noises are superimposed on each MR signal. Although the arterial input function (AIF) calculated from a plurality of MR signals is partially noise-free, but is an AIF calculated by calculating a very small ROI, FIG. 7A, FIG. As illustrated in FIG. 10, the effects of noise still remain.
[0058]
The graph illustrated in FIG. 7A shows that the first pass period and the second pass period of the arterial input function (AIF) illustrated in FIG. 1 or FIG. 3 are relatively separated, but high frequency noise is included in the MR signal. This shows a state in which they are superimposed. A curve indicated by a broken line CV11 indicates an artery input function (AIF) that is not affected by noise, and a curve indicated by a solid line CV12 indicates an artery input function (AIF) on which noise is superimposed.
Such random high-frequency noise is canceled by the moving average processing shown in Equations 1 and 2 in Step 3, and as shown by the solid line CV13 in FIG. An AIF equivalent to the artery input function (AIF) that is not received is obtained.
If the slope is calculated by differentiating the smoothed AIF indicated by the solid line CV13 in FIG. 7B, the point a1 can be detected as the maximum value Hmax in Step 5, and in Step 6, the first threshold value Hmax / p1 is calculated. The point a2 can be detected as the start point of the first pass period, the minimum value Hmin can be detected from a3 in step 7, and the point a4 can be detected as the end point of the first pass period from the second threshold value Hmax / p3 in step 8. it can.
[0059]
From the above, it can be understood that the first pass period can be accurately detected without any problem when the high frequency noise is simply superimposed. Therefore, it is assumed that there is no problem when mere high-frequency noise is superimposed, and a case where other noise occurs is described.
[0060]
A curve CV21 indicated by a solid line in FIG. 8 is a contrast agent concentration curve of the arterial input function (AIF) including two local maximum values b1 and b2 near the maximum value. Since the AIF is calculated using a very small ROI (usually, the number of pixels is usually 10 or less), the AIF illustrated in the curve CV21 may be obtained under the influence of noise. Looking only at the slope of this curve, if the b1 part is detected as the peak point indicating the maximum value, the differential value is reversed at the b2 part, so the b2 point is detected as the end point of the first pass period.
In order to avoid such an erroneous decision due to the influence of noise, the AIF is smoothed in step 3 before the AIF is differentiated in step 4 to obtain the slope of the AIF. The result is shown as a curve CV22 indicated by a broken line. In the curve CV22, one new maximum value b3 is interpolated between the two maximum values b1 and b2, and the entire AIF is also smoothed. However, the original AIF was not lost. Therefore, as illustrated in FIG. 8, the first pass period can be accurately determined using the first threshold value Hmax / p1 and the second threshold value Hmax / p3.
[0061]
FIG. 9 shows an example of a perfusion image of the cerebral blood flow (CBF) of the brain reconstructed after performing the preprocessing described with reference to FIG. Although not an accurate illustration due to the limitation of the illustration, a hatched portion indicates the CBF displayed in red. Even in the arterial input function (AIF) such as the curve CV21 in FIG. 8, the CBF is clearly imaged.
Comparing the CBF illustrated in FIG. 9 with the CBF based on Proton Emission Tomography (PET), it was found that the CBF had a very high correlation with the CBF based on PET.
[0062]
A curve CV31 indicated by a solid line in FIG. 10 is an AIF contrast agent concentration curve in which the rising portion of the arterial input function (AIF) is instantaneously reduced to 0 or less due to noise. Such noise is corrected by the AIF smoothing process in step 3 as shown by the curve CV32 indicated by the broken line. By the processing after step 4, the first pass period can be accurately detected as described above.
[0063]
Other application examples
The image signal processing method of the present invention can be applied not only to the contrast agent concentration curve of the arterial input function (AIF), but also to detection of other specific periods.
Further, the image signal processing method of the present invention can be applied not only to AIF but also to other contrast agent concentration curves such as muscle tissue.
[0064]
Although the arterial input function (AIF) has been described in the above-described embodiment, the image signal processing method of the present invention can be basically applied to a signal for each pixel. That is, according to the image signal processing method of the present invention, the first pass period can be accurately detected for each pixel (in a pixel by pixel).
[0065]
Further, the present invention is not limited to the MRI apparatus, and can be applied to the detection of the first pass period as preprocessing in CT-perfusion image processing using a contrast agent. The same problem as in the processing, that is, the inaccuracy of the determination of the first pass period and the resulting lack of accuracy of the cerebral blood flow (CBF) map can be improved.
[0066]
【The invention's effect】
According to the present invention, even when the original detection signal used in the MR contrast perfusion image signal processing method, the CT contrast perfusion image signal processing method, and the like is affected by various kinds of noise, the influence of the noise is eliminated by smoothing processing. Furthermore, an appropriate threshold value can be set, and the first pass period can be detected accurately and automatically.
Further, according to the present invention, the accuracy of the perfusion image can be improved by using the signal of the first pass period that is accurately detected by performing the preprocessing.
[Brief description of the drawings]
FIG. 1 is a graph showing a contrast agent concentration / time change graph.
FIG. 2 is a graph illustrating the flow rate of blood in an artery, the flow rate of blood in a vein, and the flow rate of blood in a muscle tissue.
FIG. 3 is a graph showing an average tissue concentration curve CVAV of the whole brain such as arteries, veins, and muscle tissues, and a curve CVAIF of an arterial input function (AIF).
FIG. 4 is a configuration diagram of an MRI apparatus as a first embodiment of the present invention.
FIG. 5 is a flowchart showing a process as a first example of detecting a first pass period of an arterial input function (AIF) performed by the data processing unit illustrated in FIG. 4;
FIG. 6 is a flowchart illustrating a process as a second example of detecting the first pass period of the arterial input function (AIF) performed by the data processing unit illustrated in FIG. 4;
7A is a graph showing a contrast agent concentration curve of an arterial input function (AIF) when high frequency noise is superimposed on an MR signal, and FIG. 7B is a graph showing a smoothed AIF of the present invention. 3 is a graph showing a contrast agent concentration curve of FIG.
FIG. 8 is a contrast agent concentration curve of an arterial input function (AIF) including two maximum values b1 and b2 near the maximum value.
FIG. 9 is a diagram schematically illustrating an image showing a cerebral blood flow (CBF) generated by using a signal of a first pass period obtained by processing the arterial input function (AIF) of FIG. 8; It is.
FIG. 10 is a graph showing an AIF in which an instantaneous decrease due to noise is observed in a rising portion of an arterial input function (AIF), and a result of the smoothing process.
[Explanation of symbols]
100 magnet system
102 ・ ・ Main magnetic field coil
106..Gradient coil section
108 RF coil part
130 ・ ・ Gradient coil driver 140 ・ ・ RF coil driver
150 data collection unit 160 control unit
170 data processing unit 180 display unit
190 ・ ・ Operation unit
300 subject 500 cradle

Claims (14)

Contrast agent concentration signal indicating the contrast agent concentration intravenously injected contrast agent is continuously input,
Smoothing the continuously input contrast agent density signal for one pixel or a plurality of pixels when reconstructing an image according to the continuously input time lapse;
Find the maximum value of the smoothed contrast agent concentration signal,
With reference to the maximum value of the determined contrast agent concentration signal, a first threshold value for determining the start of substantial presence of the contrast agent included in the continuously input contrast agent concentration signal, and Calculating a second threshold value for determining the end of the substantial presence of the contrast agent contained in the continuously input contrast agent concentration signal;
Comparing the smoothed contrast agent concentration signal with the first threshold to determine the onset of substantial presence of the contrast agent, and determining the smoothed contrast agent concentration signal with the second threshold; An image signal processing method for comparing thresholds to determine the end of substantial presence of the contrast agent. Reconstructing an image using the contrast agent concentration signal between the start and end of the substantial presence of the determined contrast agent,
The image signal processing method according to claim 1. The continuously input contrast agent concentration signal is an MR signal detected in the MRI apparatus after intravenous injection of the contrast agent,
The contrast agent concentration signal for the plurality of pixels is a signal indicating an arterial input function (AIF).
The image signal processing method according to claim 1. Performing a smoothing process of the continuously input contrast agent density signal for one pixel or a plurality of pixels when reconstructing the image based on the following equation:
The image signal processing method according to claim 3.

The first threshold value is defined by Hmax / p1 (where Hmax is a maximum value and p1 is a first threshold value parameter).
The image signal processing method according to claim 1. The second threshold value is defined by Hmax / p3 (where Hmax is a maximum value and p3 is a second threshold value parameter).
An image signal processing method according to claim 5. The continuously input contrast agent concentration signal is an XG tomographic imaging signal detected by the CT device after intravenous injection of the contrast agent.
The image signal processing method according to claim 1. Means for continuously inputting a contrast agent concentration signal indicating a contrast agent concentration obtained by intravenously injecting a contrast agent,
Means for smoothing the contrast agent density signal for one pixel or a plurality of pixels when the continuously input image is reconstructed in accordance with the lapse of the continuously input time;
Means for determining the maximum value of the smoothed contrast agent concentration signal,
A first threshold value for determining the start of substantial presence of a contrast agent contained in the continuously input contrast agent concentration signal with reference to the maximum value of the determined contrast agent concentration signal; and Means for calculating a second threshold value for determining the end of the substantial presence of the contrast agent included in the contrast agent concentration signal obtained,
Comparing the smoothed contrast agent concentration signal with the first threshold to determine the onset of substantial presence of the contrast agent, and determining the smoothed contrast agent concentration signal with the second threshold; Means for comparing thresholds to determine the end of substantial presence of the contrast agent. Means for reconstructing an image using the contrast agent concentration signal between the start and end of the substantial presence of the determined contrast agent,
The image signal processing device according to claim 8. The continuously input contrast agent concentration signal is an MR signal detected in the MRI apparatus after intravenous injection of the contrast agent,
The contrast agent concentration signal for the plurality of pixels is a signal indicating an arterial input function (AIF).
The image signal processing device according to claim 8. Performing the smoothing processing of the contrast agent density signal for one pixel or a plurality of pixels when reconstructing the continuously input image based on the following equation:
The image signal processing device according to claim 10.

The first threshold value is defined by Hmax / p1 (where Hmax is a maximum value and p1 is a first threshold value parameter).
The image signal processing device according to claim 8. The second threshold value is defined by Hmax / p3 (where Hmax is a maximum value and p3 is a second threshold value parameter).
The image signal processing device according to claim 11. The continuously input contrast agent concentration signal is an XG tomographic imaging signal detected by the CT device after intravenous injection of the contrast agent.
The image signal processing device according to claim 8.

Images