JP4236352B2 - Biological signal measuring device - Google Patents

Biological signal measuring device Download PDF

Info

Publication number
JP4236352B2
JP4236352B2 JP30413299A JP30413299A JP4236352B2 JP 4236352 B2 JP4236352 B2 JP 4236352B2 JP 30413299 A JP30413299 A JP 30413299A JP 30413299 A JP30413299 A JP 30413299A JP 4236352 B2 JP4236352 B2 JP 4236352B2
Authority
JP
Japan
Prior art keywords
component
signal
independent
noise
detection signal
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Fee Related
Application number
JP30413299A
Other languages
Japanese (ja)
Other versions
JP2001120511A (en
Inventor
茂樹 梶原
啓介 外山
思朗 池田
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shimadzu Corp
Japan Science and Technology Agency
National Institute of Japan Science and Technology Agency
Original Assignee
Shimadzu Corp
Japan Science and Technology Agency
National Institute of Japan Science and Technology Agency
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shimadzu Corp, Japan Science and Technology Agency, National Institute of Japan Science and Technology Agency filed Critical Shimadzu Corp
Priority to JP30413299A priority Critical patent/JP4236352B2/en
Publication of JP2001120511A publication Critical patent/JP2001120511A/en
Application granted granted Critical
Publication of JP4236352B2 publication Critical patent/JP4236352B2/en
Anticipated expiration legal-status Critical
Expired - Fee Related legal-status Critical Current

Links

Images

Landscapes

  • Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)

Description

【0001】
【発明の属する技術分野】
この発明は、脳磁計や脳波計などの生体信号計測装置に係り、特に生体信号測定用のセンサにより得られた検出信号からノイズ成分を除去し、脳内活動を推定するための技術に関する。
【0002】
【従来の技術】
生体内に流れる生体活動電流により微小な生体磁気(生体磁界)が生体から発生する。例えば、脳から発生する生体磁気は脳磁と呼ばれ、生体に刺激を与えることにより発生する誘発脳磁や、α波やてんかんのスパイク波のように自然に発生する自発脳磁などがある。
【0003】
近年、生体から出る微小な生体磁気を測定できる磁束計として、SQUID(Superconducing Quantum Interference Device:超電導量子干渉計)を用いたマルチチャンネルSQUIDセンサが開発されている。マルチチャンネルSQUIDセンサは、デュアーと呼ばれる容器内に多数個のSQUIDセンサを液体窒素などの冷媒に浸漬・収納した構成となっている。
【0004】
このマルチチャンネルSQUIDセンサ(以下、適宜「磁束計」と略記)を備えた生体信号計測装置、つまり生体磁気計測装置の場合、磁束計を被検体の関心部位である例えば頭部の傍らに置くと、頭部内に生じた生体活動電流源から発生する微小な生体磁気が磁束計内の各SQUIDセンサで無侵襲で測定されて磁気検出信号として出力されるとともに、SQUIDセンサからの磁気検出信号に基づいて生体磁気解析が行われて、生体活動電流源の状態(例えば、位置や向き或いは大きさ等)を把握することができるという構成になっている(例えば特開平7−327943号公報参照)。
【0005】
【発明が解決しようとする課題】
しかしながら、従来の生体磁気計測装置には、各SQUIDセンサによって検出された磁気検出信号に含まれるノイズ成分を十分に除去することが難しいという問題がある。測定対象である生体磁気は非常に微弱であることから生体活動電流源以外の別の磁気発生源から出る(環境ノイズとも称するような)ノイズ磁気の混入が避けられない。したがって、各SQUIDセンサからの磁気検出信号には混入するノイズ磁気によるノイズ成分が含まれており、磁気検出信号からノイズ成分を十分に除去しなければ、生体磁気の正確な解析はおぼつかないことになる。
【0006】
そこで、生体磁気測定用のSQUIDセンサとは別のノイズ磁気検出専用の磁気センサでノイズ磁気だけを同時に測定することにより得られたノイズ磁気検出信号を利用し、生体磁気測定用のSQUIDセンサの磁気検出信号に含まれているノイズ成分を除去する補正処理を行うようなことも提案されてはいる。
【0007】
この場合、ノイズ磁気検出専用の磁気センサと生体磁気測定用のSQUIDセンサとは設置位置が異なっており、ノイズ磁気検出専用の磁気センサで得るノイズ磁気検出信号は、生体磁気測定用のSQUIDセンサの磁気検出信号に含まれているノイズ成分と正確に対応しているわけではないので、ノイズ磁気検出専用の磁気センサで得たノイズ磁気検出信号に基づき、生体磁気測定用のSQUIDセンサの磁気検出信号に含まれているノイズ成分を推定することになる。しかし、ノイズ磁気が複雑な様相を呈するものであることなどから、空間的に異なる位置のノイズ成分を正確に求めることは非常に難しく、その結果、生体磁気測定用のSQUIDセンサの磁気検出信号からノイズ成分を十分に除去することは、やはり望めない。
【0008】
そこで、本出願人は、磁気検出信号のノイズ成分に対処すべく検討を重ねた結果、いわゆるICA(独立成分分析)手法に従って、複数個のセンサにより検出された原検出信号(観測信号)を、生体活動の各電流源とその他の各信号源ごとの独立成分に分解した上で、各独立成分について各独立成分自体の状態のみに基づきノイズ成分であるか否かを判定し、ノイズ成分と判定された独立成分は除いた後、残りの非ノイズ独立成分によって検出信号を各々の情報源毎に分離して復元する構成を備えた点を特徴とする生体信号計測装置を、先に特願平11−173839号として提案している。この出願人の先願に係る生体信号計測装置は、ノイズ検出専用のセンサを別途に設けずとも、ノイズ成分が十分に除去された生体信号解析用の復元検出信号に基づいて正確な信号解析が可能である。
【0009】
しかしながら、先願の生体信号計測装置でも、原検出信号の数(即ちセンサの個数)が、各生体活動電流源(信号源)と、その他のノイズ源などを含む各発生源(信号源)とを合計した発生源全個数より大幅に多い場合や、ノイズレベルが大きい(S/N比が低い)場合は、独立成分への分解が適正に行われないという問題がある。センサの個数が発生源全個数より大幅に多い場合は、分解後は一つの独立成分だけに纏まっているはずの成分が複数の独立成分に分割されたり、またノイズレベルが大きい場合は、分解後は単独の成分のみからなるはずの独立成分が複数の独立成分を含んでいて、独立成分への分解が適正には行われず、信号解析の信頼度が悪くなるという不都合がある。
【0010】
この発明は、上記の事情に鑑み、センサの個数が発生源全個数より多い場合や、ノイズレベルが大きい場合であっても、独立成分への分解が適正に行われて正確な信号解析結果を得ることができる生体信号計測装置を提供することを課題とする。
【0011】
【課題を解決するための手段】
前記課題を解決するために、この発明に係る生体信号計測装置は、被検体の診断対象領域において生体活動電流源により生じる微小な生体信号を測定する複数個のセンサを備えた生体信号計測装置において、前記複数個のセンサにより検出された検出信号における独立成分の個数を求出する独立成分個数求出手段と、検出信号に乗算処理を行って、検出信号を(a)複数個のセンサにより検出された検出信号において生体活動電流源による生体信号成分と(b)ノイズ成分との和とすることにより、生体信号成分とノイズ成分との相互依存性を解消する(生体信号成分とノイズ成分との間を無相関化する)成分間相互依存解消手段と、生体信号成分とノイズ成分との相互依存性が解消された検出信号を前記独立成分個数求出手段により求出した個数の数の独立成分に分解する信号分解手段と、この信号分解手段により分解された独立成分の中のノイズ成分を独立成分自体の状態のみに基づいて判定して取り除くノイズ成分除去手段と、各非ノイズ独立成分に基づき検出信号を復元する信号復元手段と、復元された各検出信号の各々の独立成分に対応する脳内の活動(位置、方向、強さ)を得る信号解析手段とを備えている。
【0012】
〔作用〕
次に、この発明の生体信号計測装置により生体信号の計測を行う時のノイズ成分の除去作用について説明する。
この発明の装置により生体信号計測を実行する際は、先ず複数個のセンサが被検体の診断対象領域の直ぐ傍にセットされて、各センサによって例えば生体活動電流源により発生する微小な生体磁気が複数個のセンサにより各々測定されて原検出信号として出力される。
【0013】
そして、原検出信号における独立成分の個数が独立成分個数求出手段により求出されるとともに、成分間相互依存解消手段により、原検出信号において生体活動電流源による生体信号成分とノイズ成分との相互依存性が解消される(つまり生体信号成分とノイズ成分との間が無相関化される)。ついで、信号分解手段により、生体信号成分とノイズ成分との相互依存性が解消された検出信号は、独立成分個数求出手段により予め求出した個数の数の独立成分に分解される。その後、ノイズ成分除去手段により、各独立成分の中のノイズ成分が各独立成分自体の状態のみに基づいて判定されて取り除かれてから、信号復元手段によってノイズ成分が除去された残りの非ノイズ独立成分に基づき検出信号(復元検出信号)が復元されて信号解析手段へ送出される。そして、信号解析手段によって、復元検出信号に基づいて生体解析が行われ、個々の生体信号の場所と活動波形が的確に把握される。
【0014】
この発明の装置では、出願人の先願に係る発明の装置と同様、統計的に独立性の高い複数の信号に分解する独立成分分析(ICA:Independence Component Analysis)手法に従って、センサにより検出された原検出信号(観測信号)が生体活動の各電流源とその他の各発生源ごとの独立成分に分解された上で、各独立成分について各独立成分自体の状態のみに基づきノイズ成分であるか否かが判定される。ノイズ成分と判定された独立成分が除かれた後、残りの各非ノイズ独立成分によって検出信号が復元される。その結果、ノイズ検出専用のセンサを別途に設けずとも、生体信号測定用のセンサで得られた信号だけで、ノイズ成分が十分に除去された生体信号解析用の復元検出信号が各成分に分解した形で得られる。
【0015】
そして、この発明の装置の場合、加えて、独立成分に分解される前に、原検出信号における生体活動電流源による生体信号成分とノイズ成分との間が無相関化されるので、ハイレベルの(大きい)ノイズ成分であっても、独立成分の分解の際にノイズ成分が生体信号成分の側に紛れ込み難くなっていて、分解後は単独の成分のみからなるはずの独立成分が複数の独立成分を含む未独立状態となる心配がない。また、独立成分の分解は予め求出しておいた原検出信号の独立成分の個数の独立成分に検出信号が分解されるので、センサの個数が発生源全個数より大幅に多い場合であっても、検出信号の分解個数が事前に十分に絞られていて、分解後は一つの独立成分だけに纏まっているはずの成分が複数の独立成分に分割されてしまう心配もない。
すなわち、この発明では、センサの個数が発生源全個数より大幅に多い場合や、ノイズレベルが大きい場合であっても、検出信号の独立成分への分解は適正に行われるのである。
【0016】
【発明の実施の形態】
続いて、この発明の一実施例を図面を参照しながら説明する。図1は本発明に係る生体信号計測装置の一例である生体磁気計測装置の全体構成を示すブロック図である。
【0017】
実施例の生体磁気計測装置は、図1に示すように、被検体(患者)Mの診断対象領域において生体活動電流源により生じる微小な生体磁気を測定するマルチチャンネルSQUIDセンサ1と、マルチチャンネルSQUIDセンサ1で得られる出力データ(出力信号)を適宜に変換・収集して原磁気検出信号として出力するデータ変換収集部2と、データ変換収集部2から送られてくる原磁気検出信号のノイズ成分除去処理や生体磁気解析処理などを行う信号処理部3とを備えている。
【0018】
マルチチャンネルSQUIDセンサ1は、デュアーと呼ばれる容器内に微小な生体磁気の測定に適した多数のSQUIDセンサ1aが縦横にアレイ状に配列された形で液体窒素などの冷媒に浸漬・収納されている構成になっている。実施例装置のマルチチャンネルSQUIDセンサ1には、128チャンネル分のSQUIDセンサ1aが配備されており、各チャンネルの1回当たりの生体磁気の測定時間は例えば512msec(ミリ秒)である。なお、多数のSQUIDセンサ1aがこの発明における複数個のセンサに相当する。
【0019】
また、マルチチャンネルSQUIDセンサ1の後段のデータ変換収集部2は、各SQUIDセンサ1aの出力信号をディジタル信号に変換して収集し、これを原磁気検出信号として信号処理部3へ送り込む。実施例装置の場合、原磁気検出信号は128行512列の行列形態の信号として扱われる構成となっている。つまり、各SQUIDセンサ1aの出力信号を1kHのサンプリング周期(1ミリ秒間隔)で512回続けて取り込み、データ数512個の原磁気検出信号を、128チャンネル分得ており、原磁気検出信号は128行512列の行列の形態で扱われることになるのである。したがって、原磁気検出信号の行列における行数や列数は、SQUIDセンサ1aのチャンネル数や出力信号のサンプリング回数に応じて変わることになる。
【0020】
信号処理部3は、この発明の生体磁気計測装置における特徴的な構成部分であって、原磁気検出信号に基づいて原磁気検出信号における独立成分の個数m0 、および後で信号処理に用いる後述の二つの変換行列Wp ,Wq を予め先に求出する予備処理部4と、原磁気検出信号において生体活動電流源による生体信号成分とノイズ成分との相互依存性を解消する成分間相互依存解消部5と、生体信号成分とノイズ成分との相互依存性が解消された磁気検出信号を複数の独立成分に分解する磁気信号分解部6と、磁気信号分解部6により分解された独立成分の中のノイズ成分を独立成分自体の状態のみに基づいて判定して取り除くノイズ成分除去部7と、ノイズ成分が除去された残りの非ノイズ独立成分に基づき元の磁気検出信号の形へ戻して各成分毎の復元磁気検出信号として個別に出力する磁気信号復元部8とを備えているとともに、磁気信号復元部8からの復元磁気検出信号に基づいて生体磁気解析を行う磁気解析部9を備えている他、生体内の分極(ダイポール)を破壊して生体活動電流を流すための刺激を被検体M(生体)に与えるための刺激付与部DTなどを備えている。以下、この信号処理部3の各部構成を、より詳しく説明する。
【0021】
すなわち、予備処理部4は、計測実行に伴って得られる128行512列の行列形態の原磁気検出信号Bにおける独立成分の個数m0 を求出する演算を行うのに加え、成分間相互依存解消部5で用いるFactor Loading Matrix の疑似逆行列としてのm0 行128列の第1変換行列Wp 、および、磁気信号分解部6で用いるRotation Matrix(回転行列)としてのm0 行128列の第2変換行列Wq とを求める演算を行うよう構成されている。独立成分の個数m0 や各変換行列Wp ,Wq を求める演算プロセスは後で具体的に説明する。
【0022】
一方、成分間相互依存解消部5は、生体信号成分とノイズ成分との相互依存性を解消する処理として、第1変換行列Wp と原磁気検出信号Bとの積(Wp B)を求める行列演算を行うよう構成されている。この演算は原磁気検出信号Bに対してFA手法によるSphering処理を施すことに相当する。(Wp B)はm0 行512列の行列となる。
他方、磁気信号分解部6は、成分間相互依存解消後の磁気検出信号である(Wp B)をm0 個の独立成分に分解する処理として、第2変換行列Wq と(Wp B)との積〔Wq (Wp B)〕を求める行列演算を行うよう構成されている。この演算はICA手法によるRotation処理を施すことに相当している。磁気信号分解部6による行列演算の結果、〔Wq (Wp B)〕はm0 行512列のICA(独立成分分析)信号Xとなる。なお、原磁気検出信号Bの独立成分の個数m0 は、発生源全個数であって、また変換行列Wp ,Wq の行数ともなっている。
【0023】
続いて、第1変換行列Wp および独立成分の個数m0 を求める演算プロセスを説明する。実施例装置の予備処理部4の場合、第1変換行列Wp と個数m0 は、FA(Factor Analysis =因子分析)手法に従って求められる。FA手法の場合、先ず次の(1)式のように、原磁気検出信号Bと、Factor Loading Matrix (因子負荷量行列)A(=Wp -1)とが、原磁気検出信号Bの行列と同一行列構成の因子ベクトルfおよび外乱ε(独自因子分散行列)を介して関係づけられるよう設定される。因子ベクトルfは生体信号成分に相当し、εはノイズ成分に相当する。
B=Af+ε ・・・(1)
因子負荷量行列Aは、センサの数と同じ行数(128行)と因子の数と同じ列数(m列)の行列である。そして、FA手法では統計処理で因子負荷量行列Aを求出することになるが、例えば次の(2)式で示すMLE(Maximum Likelihood Estimate)評価関数L(A,σ)を用いる方法が考えられる。
【0024】
L(A,σ)=−{tr〔C(σ+AAT -1〕+Log det(σ+AAT )+128Log2 π}・・・(2)
但し、σはεに対応する対角行列,trはトレース(対角要素の和),Cは原磁気検出信号Bの相関行列の平均値(Σi=1 N BBT /N:(但し、Nはサンプリング数,BT はBの転置行列),AT はAの転置行列,det は行列式の値である。
そして、MLE評価関数L(A,σ)が最大となる時のAM ,σM が求める行列として決定される。なお、MLE評価関数L(A,σ)の値を求出するのには、例えばニューラルネットワーク法に属するEMアルゴリズムを用いる方法が考えられる。
【0025】
一方、第1変換行列Wp の行数でもある独立成分の個数m0 は、次の(3)式で示すMDL評価関数が用いられる。
MDL=−L(AM ,σM )+( LogN)÷N×{128(m+1)-m(m-1)/2 }・・・(3)
この(3)式において、m=1,m=2,・・・, m= (1/2){2 ×128 −√(8×128 +1)}の時の各MDL評価関数の値を個々に求めて、MDL値が最小値となる時のmを、独立成分の個数m0 と決定して因子負荷量行列Aを完成させた後、完成した128行m0 列の因子負荷量行列Aを逆行列へ変換することにより、m0 行128列の第1変換行列Wp を求出する。
【0026】
つまり、上の(1)式に示すように、第1変換行列Wp の元である因子負荷量行列Aは、原磁気検出信号Bを生体信号成分に相当する因子ベクトルfと、ノイズ成分に相当する外乱εとに別かれていることを前提として設定されているので、逆に第1変換行列Wp で原磁気検出信号Bを乗算処理(Wp B)することにより、原磁気検出信号Bにおいて生体信号成分とノイズ成分の間の相互依存性を解消(無相関化)することができるのである。
このように、原磁気検出信号が独立成分に分解される前に、原磁気検出信号における生体活動電流源による生体信号成分とノイズ成分との間が無相関化されていれば、大きなノイズ成分であっても独立成分の分解の際に生体信号成分の側にノイズ成分が紛れ込み難くなり、また、分解後は単独の成分のみからなるはずの独立成分が複数の独立成分を含む未独立状態となる心配がなく、S/N比が低い場合でも、磁気検出信号の独立成分への分解は適正になされる。
【0027】
次に、第2変換行列Wq を求める演算プロセスを説明する。実施例装置の予備処理部4の場合、第2変換行列Wq はICA手法に従って求められる。第2変換行列Wq は、無相関化された磁気検出信号をm0 個の独立成分に分解するものであって、先に求出した第1変換行列Wp と原磁気検出信号B並びに次の(4)式に従って求出される。
Σk=1 r Σi NO j|(Wq Mk Wq T ij2 ・・(4)
即ち、この(4)式の値(対角要素の和)を最小にする時のWq が求めるm0 行128列の第2変換行列Wq として求出される。
但し、Wq T はWq の転置行列,(Wq Mk Wq T ijは行列(Wq Mk Wq T )のij成分,「i NO j」はi≠j,Mk は次の(5)式で示す通りのものである。
k =〈Wp B(t)Wp B(t+τk )〉 ・・(5)
但し、τk はデータサンプリングの時間差であって、k=1,・・・,rであり、具体的には,1msec,2msec,・・・,rmsecである。また、〈 〉はアンサンブル平均であることを示す。
【0028】
実施例装置の磁気信号分解部6は、独立成分の分解の際、予め求出しておいた個数m0 の独立成分に原磁気検出信号を分解するので、センサの個数が発生源全個数より大幅に多い場合であっても、検出信号の分解個数が事前に十分に絞られていて、分解後は一つの独立成分だけに纏まって含まれているはずの成分が複数の独立成分に分割されてしまう心配がなく、磁気検出信号の独立成分への分解は適正になされる。
【0029】
また、実施例装置の場合、磁気信号分解部6は、同一事象についてICA信号Xを複数回繰り返し求めて加算平均する構成にもなっている。なお、実施例の装置の場合、ICA信号Xを加算平均する代わりに、予備処理部4で原磁気検出信号Bを加算平均することもできるよう構成されている。
ここでの繰り返し回数は、数回〜数百回までの間の適宜の回数が選ばれる。この構成により、例えば、音を聞いた時の脳の反応を検査する場合、ICA信号Xまたは原磁気検出信号Bの加算平均処理によって、目の筋肉から発生する(スパイク波的な)磁気や脳から定常的に発生するα波による磁気の他、量子ノイズなどの不要成分を除去できる。
なお、逆に、目の筋肉から発生する磁気や脳から定常的に発生するα波による磁気を残したい場合には、加算平均の繰り返し回数を少なくするか、或いは加算平均しないようにすればよい。
【0030】
続いて、ノイズ成分除去部7より後段の構成を具体的に説明する。ノイズ成分除去部7は、磁気信号分解部6によって求められたICA信号Xにおけるm0 個の各独立成分についてノイズ成分であるか否かを先ず判別する。実施例装置の場合、ICA信号Xの各行ベクトルについて、全測定時間512msecのうち測定開始から被検体に刺激付与部DTにより刺激が与えられる時点までの非検査対象区間(例えば0〜100msec)の標準偏差値Maと、刺激付与部DTにより刺激が与えられた時点以降の検査対象区間(例えば100msec〜512msec)の標準偏差値Mbの比Ma/Mbを求め、これが一定値以上の場合、その行ベクトルに対応する独立成分はノイズ成分であると判別し、一定値未満の場合、その行ベクトルに対応する独立成分は真の信号成分(生体信号成分)であると判別するように構成されている。
【0031】
非ノイズ独立成分(生体信号成分)を決定づける生体磁気は、刺激付与部DTにより刺激が与えられた時点以降に発生するので、行ベクトルの要素は刺激が与えられた時点以降に大きくなる。逆に、ノイズ成分に対応するノイズ磁気は、刺激付与部DTにより刺激が与えられた時点以降とは直接関係がなく、行ベクトルの要素は刺激が与えられた時点の前後で変化が少ない。したがって、非検査対象区間の標準偏差値Maと検査対象区間の標準偏差値Mbの比Ma/Mbについては、真の信号成分である独立成分の方の比Ma/Mbは小さく、ノイズ独立成分の方の比Ma/Mbは大きくなり、比Ma/Mbの大小を監視することで独立成分がノイズ成分であるか否かの判別が可能となる。
【0032】
すなわち、ノイズ成分除去部7は、ノイズ成分(の中でも特に環境ノイズと称するようなノイズ成分)であると判別されたものを除去する。ノイズ成分の除去は、ノイズ成分と判別された独立成分に対応する行ベクトルを0に置換することにより行われる。ノイズ成分除去部7によるノイズ成分除去に伴ってICA信号XはICA信号Xaとなる。
【0033】
磁気信号復元部8は、ノイズ成分除去部7で求められたICA信号Xaと、ICA(独立成分分析)行列としての第2変換行列Wq の逆行例である128行m0 列の逆ICA行列Wq -1とを用いて、128行512列の復元磁気検出信号Baを求める演算を行う。つまり、磁気信号復元部8において、Wq -1Xaなる行列演算が行われて、原磁気検出信号Bからノイズ成分が十分に除去された復元磁気検出信号Ba(=Wq -1Xa)が求められるのである。
【0034】
なお、磁気信号復元部8においては、独立信号源ごとに磁気検出信号を復元することも可能である。つまり、この場合には、ノイズ成分除去部7で求められたICA信号Xaの中で、復元独立成分に対応する行ベクトルのみを残し、その他の要素を全て『0』に置換したICA信号Xa’を用いてWq -1Xa’なる演算を各独立成分毎に行うことにより、復元磁気検出信号Baを求めることもできるのである。
【0035】
そして、磁気解析部9は、復元磁気検出信号Baに基づいて生体磁気解析を行う。具体的には、復元磁気検出信号Baの1ダイポール解析の結果、生体活動電流源の重心位置分布が求められたり、復元磁気検出信号BaのSpatial Filterの結果、生体活動電流源の空間分布が求められたりして、生体活動電流源の状態が把握できる。
【0036】
実施例装置によれば、ICA(独立成分分析)手法に従って、原磁気検出信号Bが、各磁気発生源ごとの独立成分に分解された上で、各独立成分について各独立成分自体の状態のみに基づきノイズ成分であるか否かが判定され、ノイズ成分と判定された独立成分が除かれた後、非ノイズ独立成分に従って磁気検出信号が復元されており、ノイズ磁気検出専用の磁気センサを別途に設けずとも、生体磁気測定用の磁気センサで得られた信号だけで、ノイズ成分が十分に除去された生体磁気解析用の復元磁気検出信号が容易に得られている。
【0037】
なお、実施例装置の信号処理部3は、コンピュータおよびその制御プログラム等を中心に構成されているものである。
【0038】
さらに、実施例装置は、解析結果を画面に映し出す表示モニタ10および解析結果をシートに印刷して出力するプリンター11といった出力機器類を備えており、必要に応じて磁気解析部9で得られた生体活動電流源の重心位置や空間分布を表示モニタ10に表示させたり、プリンター11で印刷させたりできる構成にもなっている。
【0039】
続いて、以上に詳述した構成を有する実施例の生体磁気計測装置により、生体磁気の計測を行う時の装置動作を、図面を参照しながら具体的に説明する。図2は実施例装置による生体磁気の計測実行の様子を経時的に示すフローチャートであり、図3はICA信号Xの一部を示すグラフである。
なお、以下の測定の場合、図3に示すように、測定開始から100msec経過した時点TMで被検体Mに「絵」を見せて被検体Mの後頭部側大脳内下方寄りに生体活動電流源が集中して生じるような刺激を付与した。また、計測現場は、ノイズレベルが相当に高くてS/N比は低い状況にある。
【0040】
〔ステップS1〕被検体Mの傍らに128チャンネルSQUIDセンサ1をセットして、測定を所定回数繰り返し実行することにより、128行512列の原磁気検出信号Bが信号処理部3へ所定回数繰り返し入力される。
【0041】
〔ステップS2〕予備処理部4により、原磁気検出信号Bが加算平均されるとともに、加算平均された原磁気検出信号Bに基づき原磁気検出信号Bの独立成分の個数m0 と第1,第2の両変換行列Wp ,Wq が求出される。ここでは独立成分の個数m0 は19であり、変換行列Wp ,Wq は19行128列となった。
原磁気検出信号Bの加算平均処理によって、不規則的に発生するノイズ成分や量子ノイズなどの不要成分除去が可能となる。
【0042】
〔ステップS3〕成分間相互依存解消部5により、Wp Bなる行列演算が行われることで、原磁気検出信号Bにおいて生体信号成分とノイズ成分との間が無相関化される。
【0043】
〔ステップS4〕磁気信号分解部6により、Wq (Wp B)なる行列演算が行われることにより、磁気検出信号が19個の独立成分に分解されてICA信号Xが求められる。図3には19個の独立成分のうち9個分の独立成分ICA1〜ICA9を図示する。
【0044】
〔ステップS5〕ノイズ成分除去部7によって、ICA信号Xにおける各独立成分がノイズ成分であるか否かが判定されて、ノイズ成分と判定された独立成分は除かれ、ICA信号Xaが求められる。例えば、図3の場合、独立成分ICA8,ICA9は電源などによるノイズ成分として除かれる。
【0045】
〔ステップS6〕磁気信号復元部8によってWq -1Xaなる行列演算が行われて復元磁気検出信号Baが求められる。
【0046】
〔ステップS7〕得られた復元磁気検出信号Baに従って磁気解析部9によりSpatial Filter法に準拠した生体磁気解析処理が行われ、被検体Mの大脳内における生体活動電流源の空間分布が求められるとともに、表示モニタ10やプリンター11により解析結果が出力されて、計測は終了となる。
【0047】
図4は解析処理で得られた結果を示す図であって、被検体Mの後頭部側大脳内における生体活動電流源の発生状況を示す模式図である。図4の場合、各矢印がが各生体活動電流源に相当し、矢印の長さは電流源の強度に比例し、矢印の方向が電流源の向きを示す。長い矢印が集中している箇所に真の電流源が存在していると推定される。
また、図5は、図4の中に一点鎖線で示す9個の各ポイント(10,10)〜(12,12)に対応する解析電流源の時間変化を示すグラフである。例えば,図4と図5は、図4のポイント(12,12)の解析電流源は図5の(12,12)の直ぐ右のグラフが示すという対応付けになっている。図4の各グラフは横軸が時間,縦軸が電流源強度を示し、略中央の縦直線が刺激付与後100msec経過した時点を示す。
【0048】
なお、図6は従来装置により得られた解析結果を示す図であって、図4と同様、被検体Mの後頭部側大脳内における生体活動電流源の発生状況を示す模式図である。また、図7は、図5と同様の9個のポイントに対応する従来装置の各解析電流源の時間変化を示すグラフである。
【0049】
実施例装置による解析結果は、図4に示すように、生体活動電流源が後頭部側大脳内下方寄りによく集中して発生しており、被検体Mに加えられた刺激によく対応した正確な解析結果が得られている。また、復元磁気検出信号も、正確な結果に呼応して、図5に示すように、刺激付与から検出信号強度が最大となる予定時点である100msec経過後の位置にピークが単一で出現するという適正な波形となっている。
【0050】
これに対して、従来装置の場合、図6に示すように、生体活動電流源が後頭部側大脳内略中央近傍に二つに別れる感じで非集中的に発生しており、被検体Mに加えられた刺激と対応しておらず、解析結果は不正確である。
また、従来装置の場合、解析電流源の時間変化も、図7に示すように、刺激付与から検出信号強度が最大となる予定時点である100msec経過後の位置以外のところにも、顕著なピークが出現するという不適正な波形となっている。従来装置の場合、センサの個数が128個と電流源19個(独立成分の個数に相当)より大幅に多く、またS/N比が低いために、独立成分への分解が適正に行われず、正確な解析結果が得られないと推察される。
【0051】
しかし、図5および図6の結果から分かるように、実施例装置によれば、センサの個数が128個と電流源19個より大幅に多くても、またS/N比が低くても、磁気検出信号において生体信号成分とノイズ成分の間を無相関化するとともに、予め求めた独立成分の個数m0 の独立成分に磁気検出信号を分解することにより、磁気検出信号の独立成分への分解が適正に行われるので、正確な解析結果を得ることができるのである。
【0052】
この発明は、上記実施の形態に限られることはなく、下記のように変形実施することができる。
【0053】
(1)実施例装置では、ICA信号Xの独立成分における非検査対象区間の標準偏差値Maと検査対象区間の標準偏差値Mbの比Ma/Mbに基づき独立成分がノイズ成分であるか否かを判別する構成であったが、ICA信号Xの独立成分の信号強度に基づき独立成分がノイズ成分であるか否かを判別する構成の装置が、変形例として挙げられる。例えば1pT(ピコテスラ)以上のものはノイズと判別する。但し、この変形例の場合、判定対象の独立成分の信号強度は正規化された値ではなく、生の磁場強度に対応する値に変換される必要がある。
【0054】
(2)また、ICA信号Xの独立成分の周波数に基づき独立成分がノイズ成分であるか否かを判別する構成の装置も、変形例として挙げられる。例えば、ICA信号XにFFTをかけて一番支配的な周波数を求め、例えば100Hz以上であればノイズと判別する。
【0055】
(3)さらに、実施例を含めて前述した三つのノイズ成分判別方式の二つあるいは三つの方式を併用して独立成分がノイズ成分であるか否かを判別する構成の装置も、変形例として挙げられる。
【0056】
また、上述した実施例では、脳磁計による生体磁気計測装置を例に採って説明したが、脳波計などを用いた生体信号計測装置であっても同様の効果を得ることができる。
【0057】
【発明の効果】
以上に詳述したように、この発明の生体信号計測装置によれば、所謂ICA(独立成分分析)手法に従って、複数個のセンサにより検出された原検出信号(観測信号)は、生体活動の各電流源とその他の各信号源ごとの独立成分に分解された上で、各独立成分について各独立成分自体の状態のみに基づきノイズ成分であるか否かが判定され、ノイズ成分と判定された独立成分が除かれた後、残りの非ノイズ独立成分によって検出信号が各々の情報源毎に分離して復元される。したがって、ノイズ検出専用のセンサを別途に設けずとも、生体信号測定用のセンサで得られた信号だけで、ノイズ成分が十分に除去された生体信号解析用の復元検出信号が容易に得られる。その結果、復元生体検出信号に基づいて行われる信号解析も正確なものとなる。
【0058】
加えて、この発明の生体信号計測装置によれば、独立成分に分解される前に、原検出信号における生体活動電流源による生体信号成分とノイズ成分との間が無相関化される構成を備えているので、大きなノイズ成分であっても、独立成分の分解の際にノイズ成分が生体信号成分の側に紛れ込み難くなる。また、分解後は単独の成分のみからなるはずの独立成分が複数の独立成分を含む未独立状態となる心配がなくなる。さらに、予め求出しておいた独立成分の個数に等しい個数の独立成分に原検出信号が分解されるので、センサの個数が発生源全個数より大幅に多い場合であっても、検出信号の分解個数が事前に十分に絞られていて、分解後は一つの独立成分だけに纏まって含まれているはずの成分が複数の独立成分に分割されてしまう心配がなくなる。したがって、センサの個数が発生源全個数より大幅に多い場合や、ノイズレベルが大きい場合でも、検出信号の独立成分への分解は適正になされることから、正確な信号解析が行われることとなる。
【図面の簡単な説明】
【図1】実施例に係る生体磁気計測装置の全体構成を示すブロック図である。
【図2】実施例装置による生体磁気の計測実行の様子を経時的に示すフローチャートである。
【図3】実施例装置により得られたICA信号Xの一部を示すグラフである。
【図4】実施例装置により解析された後頭部側大脳内の生体活動電流源の発生状況を示す模式図である。
【図5】実施例装置により計測された後頭部側大脳内の特定の9ポイントに対応する解析電流源の時間変化を示すグラフである。
【図6】従来装置により解析された後頭部側大脳内の生体活動電流源の発生状況を示す模式図である。
【図7】従来装置により計測された後頭部側大脳内の特定の9ポイントに対応する解析電流源の時間変化を示すグラフである。
【符号の説明】
1 …マルチチャンネルSQUIDセンサ
1a …SQUIDセンサ
2 …データ変換収集部
3 …信号処理部
4 …予備処理部
5 …成分間相互依存解消部
6 …磁気信号分解部
7 …ノイズ成分除去部
8 …磁気信号復元部
9 …磁気解析部
M …被検体
[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a biological signal measuring device such as a magnetoencephalograph or electroencephalograph, and more particularly to a technique for estimating a brain activity by removing a noise component from a detection signal obtained by a sensor for measuring a biological signal.
[0002]
[Prior art]
A minute biomagnetism (biomagnetic field) is generated from the living body due to the bioactive current flowing in the living body. For example, the biomagnetism generated from the brain is called a magnetoencephalogram, and includes an induced magnetoencephalogram that is generated by applying a stimulus to the living body, and a spontaneous magnetoencephalogram that is naturally generated such as an α wave or a spike wave of epilepsy.
[0003]
In recent years, a multichannel SQUID sensor using a SQUID (Superconducting Quantum Interference Device) has been developed as a magnetometer capable of measuring minute biomagnetism emitted from a living body. The multi-channel SQUID sensor has a configuration in which a large number of SQUID sensors are immersed and accommodated in a refrigerant such as liquid nitrogen in a container called a dewar.
[0004]
In the case of a biological signal measurement device equipped with this multi-channel SQUID sensor (hereinafter abbreviated as “flux meter” where appropriate), that is, a biomagnetism measurement device, if the magnetometer is placed beside the subject's region of interest, for example, the head. In addition, the minute biomagnetism generated from the bioactive current source generated in the head is measured non-invasively by each SQUID sensor in the magnetometer and output as a magnetic detection signal, and the magnetic detection signal from the SQUID sensor Based on the biomagnetic analysis, the state (for example, position, orientation, size, etc.) of the bioactive current source can be grasped (see, for example, JP-A-7-327943). .
[0005]
[Problems to be solved by the invention]
However, the conventional biomagnetic measurement device has a problem that it is difficult to sufficiently remove the noise component included in the magnetic detection signal detected by each SQUID sensor. The biomagnetism that is the object of measurement is very weak, so it is inevitable that noise magnetism (such as environmental noise) coming from another magnetic source other than the bioactive current source is mixed. Therefore, the magnetic detection signal from each SQUID sensor includes a noise component due to mixed noise magnetism, and unless the noise component is sufficiently removed from the magnetic detection signal, an accurate analysis of biomagnetism will not be realized. .
[0006]
Therefore, the magnetic field of the SQUID sensor for biomagnetism measurement is obtained by using the noise magnetism detection signal obtained by simultaneously measuring only the noise magnetism with a magnetic sensor dedicated to noise magnetism detection different from the SQUID sensor for biomagnetism measurement. It has also been proposed to perform a correction process that removes a noise component contained in the detection signal.
[0007]
In this case, the magnetic sensor dedicated to noise magnetism detection and the SQUID sensor for biomagnetism measurement have different installation positions, and the noise magnetism detection signal obtained by the magnetic sensor dedicated to noise magnetism detection is the same as that of the SQUID sensor for biomagnetism measurement. Since it does not exactly correspond to the noise component contained in the magnetic detection signal, the magnetic detection signal of the SQUID sensor for biomagnetic measurement is based on the noise magnetic detection signal obtained by the magnetic sensor dedicated to noise magnetic detection. The noise component contained in is estimated. However, since noise magnetism presents a complex aspect, it is very difficult to accurately obtain noise components at spatially different positions. As a result, from the magnetic detection signal of the SQUID sensor for biomagnetic measurement It is still impossible to sufficiently remove the noise component.
[0008]
Therefore, as a result of repeated investigations to deal with the noise component of the magnetic detection signal, the present applicant has obtained original detection signals (observation signals) detected by a plurality of sensors according to a so-called ICA (independent component analysis) technique. Decomposes each current source of life activity and independent component for each other signal source, and determines whether each independent component is a noise component based on only the state of each independent component itself, and determines that it is a noise component A biological signal measuring device characterized by comprising a configuration in which a detection signal is separated and restored for each information source by the remaining non-noise independent components after removing the independent components. It is proposed as 11-173839. The biological signal measuring apparatus according to the applicant's prior application can perform accurate signal analysis based on the restored detection signal for analyzing biological signals from which noise components are sufficiently removed, without providing a separate sensor dedicated to noise detection. Is possible.
[0009]
However, even in the biological signal measuring device of the prior application, the number of original detection signals (that is, the number of sensors) is different from each generation source (signal source) including each biological activity current source (signal source) and other noise sources. If the total number of generation sources is significantly larger than the total number of sources, or if the noise level is large (S / N ratio is low), there is a problem that decomposition into independent components is not performed properly. If the number of sensors is significantly larger than the total number of sources, components that should have been combined into one independent component after decomposition are divided into multiple independent components, or if the noise level is high, Has an inconvenience that an independent component that should be composed of only a single component includes a plurality of independent components, and is not properly decomposed into independent components, resulting in poor signal analysis reliability.
[0010]
In view of the above circumstances, the present invention is capable of properly decomposing into independent components and obtaining an accurate signal analysis result even when the number of sensors is larger than the total number of sources or when the noise level is large. It is an object of the present invention to provide a biosignal measurement device that can be obtained.
[0011]
[Means for Solving the Problems]
  In order to solve the above problems, a biological signal measuring apparatus according to the present invention is a biological signal measuring apparatus including a plurality of sensors that measure minute biological signals generated by a biological active current source in a diagnosis target region of a subject. Independent component number obtaining means for obtaining the number of independent components in the detection signals detected by the plurality of sensors;By multiplying the detection signal, the detection signal is (a) the detection signal detected by the plurality of sensors, and the sum of the biological signal component by the biological activity current source and (b) the noise component.The interdependence cancellation means between components that eliminates the interdependence between the biological signal component and the noise component (which decorrelates between the biological signal component and the noise component), and the interdependency between the biological signal component and the noise component A signal decomposing means for decomposing the canceled detection signal into the number of independent components obtained by the independent component number obtaining means, and a noise component in the independent components decomposed by the signal decomposing means as independent components themselves A noise component removing unit that determines and removes only based on the state of the signal, a signal restoring unit that restores the detection signal based on each non-noise independent component, and a brain that corresponds to each independent component of each restored detection signal Signal analysis means for obtaining activity (position, direction, strength).
[0012]
[Action]
Next, the noise component removing action when the biological signal is measured by the biological signal measuring apparatus of the present invention will be described.
When performing biological signal measurement using the apparatus of the present invention, a plurality of sensors are first set immediately next to the diagnosis target region of the subject, and the minute biomagnetism generated by, for example, a bioactive current source is generated by each sensor. Each is measured by a plurality of sensors and output as an original detection signal.
[0013]
Then, the number of independent components in the original detection signal is obtained by the independent component number obtaining means, and the mutual detection of the inter-component interdependence is performed by the biological detection current signal source and the noise component in the original detection signal. The dependency is eliminated (that is, the biosignal component and the noise component are decorrelated). Next, the detection signal from which the interdependence between the biological signal component and the noise component is eliminated by the signal decomposing means is decomposed into the number of independent components obtained in advance by the independent component number obtaining means. Thereafter, the noise component removal means determines and removes the noise component in each independent component based only on the state of each independent component itself, and then the remaining non-noise independent signal from which the noise component is removed by the signal restoration means. A detection signal (restoration detection signal) is restored based on the component and sent to the signal analysis means. Then, the signal analysis means performs biological analysis based on the restoration detection signal, and accurately grasps the location and activity waveform of each biological signal.
[0014]
In the apparatus of the present invention, as in the apparatus of the invention of the applicant's prior application, it was detected by a sensor according to an independent component analysis (ICA) method that decomposes into a plurality of statistically highly independent signals. Whether or not the original detection signal (observation signal) is a noise component based on only the state of each independent component itself after being decomposed into independent components for each current source of life activity and each other generation source Is determined. After the independent component determined as the noise component is removed, the detection signal is restored by each remaining non-noise independent component. As a result, the recovery detection signal for biosignal analysis from which the noise component has been sufficiently removed can be decomposed into each component by using only the signal obtained by the sensor for biosignal measurement, without providing a dedicated sensor for noise detection. Obtained in the form.
[0015]
In the case of the device of the present invention, in addition, before being decomposed into independent components, the biosignal component and the noise component by the bioactive current source in the original detection signal are decorrelated, so that the high level Even if it is a (large) noise component, it is difficult for the noise component to be mixed into the biological signal component side when the independent component is decomposed, and after the decomposition, there are multiple independent components that should consist of only a single component. There is no worry of becoming an independent state containing ingredients. In addition, the decomposition of the independent component is performed by dividing the detection signal into independent components equal to the number of independent components of the original detection signal obtained in advance, so even if the number of sensors is significantly larger than the total number of sources. The number of detection signals to be decomposed is sufficiently narrowed in advance, and there is no fear that components that should have been combined into only one independent component after the decomposition will be divided into a plurality of independent components.
That is, according to the present invention, even when the number of sensors is significantly larger than the total number of generation sources or when the noise level is large, the detection signal is properly decomposed into independent components.
[0016]
DETAILED DESCRIPTION OF THE INVENTION
Next, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing an overall configuration of a biomagnetic measurement apparatus which is an example of a biosignal measurement apparatus according to the present invention.
[0017]
As shown in FIG. 1, the biomagnetic measurement apparatus of the embodiment includes a multichannel SQUID sensor 1 that measures minute biomagnetism generated by a bioactive current source in a diagnosis target region of a subject (patient) M, and a multichannel SQUID. A data conversion collecting unit 2 that appropriately converts and collects output data (output signal) obtained by the sensor 1 and outputs it as an original magnetic detection signal, and a noise component of the original magnetic detection signal sent from the data conversion collecting unit 2 And a signal processing unit 3 that performs a removal process, a biomagnetic analysis process, and the like.
[0018]
The multi-channel SQUID sensor 1 is immersed and accommodated in a refrigerant such as liquid nitrogen in the form of a large number of SQUID sensors 1a that are suitable for measuring minute biomagnetism in a container called a dewar arranged vertically and horizontally. It is configured. The multi-channel SQUID sensor 1 of the embodiment apparatus is provided with SQUID sensors 1a for 128 channels, and the measurement time of biomagnetism per time of each channel is, for example, 512 msec (milliseconds). A number of SQUID sensors 1a correspond to a plurality of sensors in the present invention.
[0019]
Further, the data conversion / collection unit 2 subsequent to the multi-channel SQUID sensor 1 converts and collects output signals from the respective SQUID sensors 1a into digital signals, and sends them to the signal processing unit 3 as original magnetic detection signals. In the case of the embodiment apparatus, the primary magnetic detection signal is handled as a signal in a matrix form of 128 rows and 512 columns. In other words, the output signal of each SQUID sensor 1a is continuously fetched 512 times at a sampling period (1 millisecond interval) of 1 kHz, and the primary magnetic detection signal of 512 data is obtained for 128 channels. It is handled in the form of a matrix of 128 rows and 512 columns. Accordingly, the number of rows and the number of columns in the matrix of the original magnetism detection signal varies depending on the number of channels of the SQUID sensor 1a and the number of sampling of the output signal.
[0020]
The signal processing unit 3 is a characteristic component in the biomagnetic measurement device of the present invention, and the number m of independent components in the original magnetic detection signal based on the original magnetic detection signal.0, And a preliminary processing unit 4 that obtains two transformation matrices Wp and Wq, which will be described later, used in signal processing in advance, and interdependence between a biological signal component and a noise component by a biological activity current source in the original magnetic detection signal Inter-component interdependence canceling unit 5 for canceling the characteristics, a magnetic signal decomposing unit 6 for decomposing the magnetic detection signal from which the interdependence between the biological signal component and the noise component is resolved, into a plurality of independent components, and a magnetic signal decomposing unit The noise component removing unit 7 that determines and removes the noise component in the independent component decomposed by 6 based only on the state of the independent component itself, and the original magnetism based on the remaining non-noise independent component from which the noise component has been removed And a magnetic signal restoration unit 8 that returns to the form of the detection signal and outputs it individually as a restoration magnetism detection signal for each component, and a biomagnetic solution based on the restoration magnetism detection signal from the magnetic signal restoration unit 8 In addition to a magnetic analysis unit 9 for performing analysis, a stimulus applying unit DT for applying a stimulus for destroying the polarization (dipole) in the living body to flow a biological activity current to the subject M (living body) is provided. ing. Hereinafter, the configuration of each part of the signal processing unit 3 will be described in more detail.
[0021]
In other words, the preliminary processing unit 4 determines the number m of independent components in the primary magnetic detection signal B in the form of a matrix of 128 rows and 512 columns obtained along with the execution of measurement.0M as a pseudo inverse matrix of the Factor Loading Matrix used in the inter-component interdependence elimination unit 50A first transformation matrix Wp of 128 rows and m as a rotation matrix used in the magnetic signal decomposition unit 60An operation for obtaining the second transformation matrix Wq of 128 rows and columns is performed. Number of independent components m0The calculation process for obtaining the transformation matrices Wp and Wq will be described in detail later.
[0022]
On the other hand, the inter-component interdependency canceling unit 5 performs a matrix operation for obtaining a product (Wp B) of the first transformation matrix Wp and the original magnetic detection signal B as processing for canceling the interdependence between the biological signal component and the noise component. Is configured to do. This calculation corresponds to performing Sphering processing by the FA method on the original magnetic detection signal B. (Wp B) is m0It becomes a matrix of rows 512 columns.
On the other hand, the magnetic signal decomposing unit 6 calculates (Wp B), which is a magnetic detection signal after eliminating the interdependence between components, as m.0As a process of decomposing into independent components, a matrix operation for obtaining a product [Wq (Wp B)] of the second transformation matrix Wq and (Wp B) is performed. This calculation corresponds to performing rotation processing by the ICA method. As a result of matrix calculation by the magnetic signal decomposing unit 6, [Wq (Wp B)] is m0The ICA (Independent Component Analysis) signal X in the row 512 column. Note that the number m of independent components of the original magnetic detection signal B0Is the total number of sources and is also the number of rows of the transformation matrices Wp and Wq.
[0023]
Subsequently, the first transformation matrix Wp and the number m of independent components0A calculation process for obtaining the above will be described. In the case of the preliminary processing unit 4 of the embodiment apparatus, the first transformation matrix Wp and the number m0Is determined according to the FA (Factor Analysis) method. In the case of the FA method, first, as shown in the following equation (1), the original magnetic detection signal B and the Factor Loading Matrix (factor loading matrix) A (= Wp-1) Are related to each other through a factor vector f and a disturbance ε (unique factor dispersion matrix) having the same matrix configuration as the matrix of the original magnetic detection signal B. The factor vector f corresponds to a biological signal component, and ε corresponds to a noise component.
B = Af + ε (1)
The factor loading matrix A is a matrix having the same number of rows as the number of sensors (128 rows) and the same number of columns as the number of factors (m columns). In the FA method, the factor load matrix A is obtained by statistical processing. For example, a method using an MLE (Maximum Likelihood Estimate) evaluation function L (A, σ) expressed by the following equation (2) is considered. It is done.
[0024]
L (A, σ) = − {tr [C (σ + AA)T)-1] + Log det (σ + AAT) + 128Log2 π} (2)
Where σ is a diagonal matrix corresponding to ε, tr is a trace (sum of diagonal elements), and C is an average value of the correlation matrix of the original magnetic detection signal B (Σi = 1 NBBT/ N: (where N is the number of samplings, BTIs the transpose of B), ATIs the transposed matrix of A, and det is the value of the determinant.
A when the MLE evaluation function L (A, σ) is maximumM, ΣMIs determined as a desired matrix. In order to obtain the value of the MLE evaluation function L (A, σ), for example, a method using an EM algorithm belonging to the neural network method can be considered.
[0025]
On the other hand, the number m of independent components that is also the number of rows of the first transformation matrix Wp.0Is an MDL evaluation function expressed by the following equation (3).
MDL = -L (AM, ΣM) + (LogN) ÷ N × {128 (m + 1) -m (m-1) / 2} (3)
In this equation (3), m = 1, m = 2,..., M = (1/2) {2 × 128 −√ (8 × 128 + 1)} Where m when the MDL value is the minimum value is the number m of independent components.0And the factor loading matrix A is completed, and the completed 128 rows m0By transforming the column factor loading matrix A into an inverse matrix, m0A first transformation matrix Wp with 128 rows and columns is obtained.
[0026]
That is, as shown in the above equation (1), the factor load matrix A which is the element of the first transformation matrix Wp is equivalent to the factor vector f corresponding to the biomagnetic signal component and the noise component of the original magnetic detection signal B. Since it is set on the premise that it is separated from the disturbance ε to be transmitted, the biomagnetism detection signal B is multiplied by the first transformation matrix Wp (Wp B), so that the biomagnetism detection signal B The interdependence between the signal component and the noise component can be eliminated (decorrelated).
Thus, if the correlation between the biological signal component by the biological activity current source and the noise component in the original magnetic detection signal is uncorrelated before the primary magnetic detection signal is decomposed into independent components, a large noise component is generated. Even when there is an independent component decomposition, noise components are less likely to be mixed into the biological signal component side, and after decomposition, the independent component, which should consist of only a single component, is a non-independent state including a plurality of independent components. Even when the S / N ratio is low, the magnetic detection signal is properly decomposed into independent components.
[0027]
Next, the calculation process for obtaining the second transformation matrix Wq will be described. In the case of the preliminary processing unit 4 of the embodiment apparatus, the second transformation matrix Wq is obtained according to the ICA method. The second transformation matrix Wq converts the decorrelated magnetic detection signal into m0It is decomposed into independent components, and is obtained according to the first transformation matrix Wp, the original magnetic detection signal B, and the following equation (4).
Σk = 1 rΣi NO j| (Wq MkWqT)ij2  (4)
In other words, Wq when the value of equation (4) (sum of diagonal elements) is minimized is obtained m0It is obtained as a second transformation matrix Wq having 128 rows and columns.
However, WqTIs the transposed matrix of Wq, (Wq MkWqT)ijIs a matrix (Wq MkWqTIj component, “i NO j” is i ≠ j, MkIs as shown by the following equation (5).
Mk= <Wp B (t) Wp B (t + τk)> (5)
Where τkIs a time difference of data sampling, and k = 1,..., R, specifically, 1 msec, 2 msec,. <> Indicates an ensemble average.
[0028]
The magnetic signal decomposing unit 6 of the embodiment apparatus determines the number m obtained in advance when decomposing the independent component.0Therefore, even if the number of sensors is significantly larger than the total number of sources, the number of detection signals is sufficiently reduced in advance and There is no fear that components that should be included in only one independent component are divided into a plurality of independent components, and the magnetic detection signal is properly decomposed into independent components.
[0029]
Further, in the case of the embodiment apparatus, the magnetic signal decomposing unit 6 is configured to repeatedly calculate the ICA signal X for the same event a plurality of times and perform the averaging. In the case of the apparatus of the embodiment, instead of averaging the ICA signal X, the preliminary processing unit 4 can also add and average the original magnetic detection signal B.
As the number of repetitions, an appropriate number of times from several times to several hundred times is selected. With this configuration, for example, when examining the response of the brain when listening to sound, the magnetic field or brain (spike wave) generated from the muscle of the eye by the addition averaging process of the ICA signal X or the original magnetic detection signal B In addition to the magnetism caused by the α waves that are generated constantly, unnecessary components such as quantum noise can be removed.
Conversely, when it is desired to retain the magnetism generated from the muscles of the eyes or the α-waves generated constantly from the brain, the number of repetitions of addition averaging may be reduced or the addition averaging may not be performed. .
[0030]
Next, the configuration subsequent to the noise component removal unit 7 will be described in detail. The noise component removing unit 7 calculates m in the ICA signal X obtained by the magnetic signal decomposing unit 6.0First, it is determined whether or not each individual component is a noise component. In the case of the embodiment apparatus, with respect to each row vector of the ICA signal X, the standard of the non-examination target section (for example, 0 to 100 msec) from the start of measurement to the time point when the stimulus is applied to the subject by the stimulus applying unit DT in the total measurement time 512 msec. A ratio Ma / Mb of a deviation value Ma and a standard deviation value Mb of a section to be inspected (for example, 100 msec to 512 msec) after the time point when the stimulus is applied by the stimulus applying unit DT is obtained. It is determined that the independent component corresponding to is a noise component, and if it is less than a certain value, the independent component corresponding to the row vector is determined to be a true signal component (biological signal component).
[0031]
The biomagnetism that determines the non-noise independent component (biological signal component) is generated after the point of time when the stimulus is applied by the stimulus applying unit DT, so that the row vector element becomes larger after the point of time when the stimulus is applied. Conversely, the noise magnetism corresponding to the noise component is not directly related to the time after the stimulus is applied by the stimulus applying unit DT, and the elements of the row vector change little before and after the time when the stimulus is applied. Therefore, regarding the ratio Ma / Mb between the standard deviation value Ma of the non-inspection target section and the standard deviation value Mb of the inspection target section, the ratio Ma / Mb of the independent component that is the true signal component is small, and the noise independent component The ratio Ma / Mb becomes larger, and it is possible to determine whether the independent component is a noise component by monitoring the magnitude of the ratio Ma / Mb.
[0032]
In other words, the noise component removal unit 7 removes noise components that have been determined to be noise components (in particular, noise components called environmental noise). The removal of the noise component is performed by replacing the row vector corresponding to the independent component determined as the noise component with 0. As the noise component is removed by the noise component removing unit 7, the ICA signal X becomes the ICA signal Xa.
[0033]
The magnetic signal restoration unit 8 is an inverse example of the ICA signal Xa obtained by the noise component removal unit 7 and the second transformation matrix Wq as an ICA (independent component analysis) matrix.0Inverse ICA matrix of columns Wq-1Are used to calculate the restored magnetic detection signal Ba of 128 rows and 512 columns. That is, in the magnetic signal restoration unit 8, Wq-1The restored magnetic detection signal Ba (= Wq) in which the matrix operation Xa is performed and the noise component is sufficiently removed from the original magnetic detection signal B.-1Xa) is required.
[0034]
The magnetic signal restoration unit 8 can restore the magnetic detection signal for each independent signal source. That is, in this case, in the ICA signal Xa obtained by the noise component removing unit 7, only the row vector corresponding to the restored independent component is left and all other elements are replaced with “0”. Using Wq-1The restored magnetic detection signal Ba can also be obtained by performing the calculation of Xa 'for each independent component.
[0035]
The magnetic analysis unit 9 performs biomagnetic analysis based on the restored magnetic detection signal Ba. Specifically, as a result of one dipole analysis of the restored magnetic detection signal Ba, the center of gravity position distribution of the bioactive current source is obtained, or as a result of the Spatial Filter of the restored magnetic detection signal Ba, the spatial distribution of the bioactive current source is obtained. The state of the life activity current source can be grasped.
[0036]
According to the embodiment apparatus, according to the ICA (Independent Component Analysis) method, the original magnetic detection signal B is decomposed into independent components for each magnetic generation source, and each independent component is only in the state of each independent component itself. It is determined whether or not it is a noise component, and after the independent component determined to be a noise component is removed, the magnetic detection signal is restored according to the non-noise independent component. Even if it is not provided, a restored magnetic detection signal for biomagnetic analysis from which noise components are sufficiently removed can be easily obtained only by a signal obtained by a magnetic sensor for biomagnetism measurement.
[0037]
In addition, the signal processing unit 3 of the embodiment apparatus is mainly configured by a computer and its control program.
[0038]
Further, the apparatus according to the embodiment includes an output device such as a display monitor 10 that displays the analysis result on a screen and a printer 11 that prints the analysis result on a sheet and outputs the sheet, and the magnetic analysis unit 9 obtains the output as necessary. The center of gravity position and the spatial distribution of the life activity current source can be displayed on the display monitor 10 or printed by the printer 11.
[0039]
Next, the operation of the apparatus when measuring biomagnetism by the biomagnetism measuring apparatus of the embodiment having the configuration detailed above will be specifically described with reference to the drawings. FIG. 2 is a flowchart showing how biomagnetism measurement is executed by the embodiment apparatus over time, and FIG. 3 is a graph showing a part of the ICA signal X.
In the following measurement, as shown in FIG. 3, a bioactive current source is shown near the lower back of the occipital side of the subject M at a time point TM when 100 msec has elapsed from the start of measurement. Stimulation that occurs in a concentrated manner was applied. In addition, the measurement site is in a situation where the noise level is considerably high and the S / N ratio is low.
[0040]
[Step S1] The 128-channel SQUID sensor 1 is set beside the subject M and the measurement is repeatedly performed a predetermined number of times, whereby the 128-row 512-column primary magnetic detection signal B is repeatedly input to the signal processing unit 3 a predetermined number of times. Is done.
[0041]
[Step S2] The preliminary processing unit 4 adds and averages the original magnetism detection signal B, and the number m of independent components of the original magnetism detection signal B based on the added and averaged original magnetism detection signal B.0The first and second transformation matrices Wp and Wq are obtained. Here, the number m of independent components0Is 19, and the transformation matrices Wp and Wq are 19 rows and 128 columns.
By adding and averaging the original magnetism detection signals B, it is possible to remove unnecessary components such as irregularly generated noise components and quantum noise.
[0042]
[Step S3] The inter-component interdependence elimination unit 5 performs a matrix operation of Wp B, whereby the biomagnetic signal component and the noise component are uncorrelated in the original magnetic detection signal B.
[0043]
[Step S4] The magnetic signal decomposing unit 6 performs a matrix operation of Wq (Wp B), whereby the magnetic detection signal is decomposed into 19 independent components to obtain the ICA signal X. FIG. 3 illustrates nine independent components ICA1 to ICA9 among the 19 independent components.
[0044]
[Step S5] The noise component removing unit 7 determines whether or not each independent component in the ICA signal X is a noise component, and the independent component determined to be a noise component is removed to obtain an ICA signal Xa. For example, in the case of FIG. 3, the independent components ICA8 and ICA9 are removed as noise components due to a power source or the like.
[0045]
[Step S6] The magnetic signal restoration unit 8 performs Wq-1A matrix operation Xa is performed to obtain a restored magnetic detection signal Ba.
[0046]
[Step S7] The magnetic analysis unit 9 performs biomagnetic analysis processing based on the Spatial Filter method according to the obtained restored magnetic detection signal Ba to obtain the spatial distribution of the bioactive current source in the cerebrum of the subject M. Then, the analysis result is output by the display monitor 10 or the printer 11, and the measurement ends.
[0047]
FIG. 4 is a diagram showing a result obtained by the analysis process, and is a schematic diagram showing a generation state of a bioactive current source in the occipital cerebrum of the subject M. In the case of FIG. 4, each arrow corresponds to each life activity current source, the length of the arrow is proportional to the intensity of the current source, and the direction of the arrow indicates the direction of the current source. It is presumed that a true current source exists where the long arrows are concentrated.
FIG. 5 is a graph showing a time change of the analysis current source corresponding to each of the nine points (10, 10) to (12, 12) indicated by a one-dot chain line in FIG. For example, in FIGS. 4 and 5, the analysis current source at the point (12, 12) in FIG. 4 is associated with the right graph of (12, 12) in FIG. In each graph of FIG. 4, the horizontal axis indicates time, the vertical axis indicates the current source intensity, and the substantially central vertical line indicates a point in time when 100 msec has elapsed after the stimulus is applied.
[0048]
FIG. 6 is a diagram showing an analysis result obtained by a conventional apparatus, and is a schematic diagram showing a generation state of a bioactive current source in the occipital cerebrum of the subject M, as in FIG. FIG. 7 is a graph showing the time change of each analysis current source of the conventional apparatus corresponding to nine points similar to FIG.
[0049]
As shown in FIG. 4, the analysis result by the example apparatus shows that the biologically active current source is generated in a concentrated manner near the lower part of the occipital side of the cerebrum, and is accurate corresponding to the stimulus applied to the subject M. Analysis results are obtained. In addition, as shown in FIG. 5, the restored magnetic detection signal also has a single peak at a position after the elapse of 100 msec, which is the scheduled time when the detection signal intensity is maximized after the stimulation, as shown in FIG. This is a proper waveform.
[0050]
On the other hand, in the case of the conventional apparatus, as shown in FIG. 6, the bioactive current source is generated in a non-concentrated manner with a feeling that it is divided into two near the center of the occipital region. Analysis result is inaccurate.
Further, in the case of the conventional apparatus, as shown in FIG. 7, the time change of the analysis current source also has a prominent peak at locations other than the position after 100 msec, which is the scheduled time when the detection signal intensity is maximized after the stimulus is applied. Is an inappropriate waveform that appears. In the case of the conventional device, the number of sensors is significantly larger than 128 and 19 current sources (corresponding to the number of independent components), and since the S / N ratio is low, decomposition into independent components is not performed properly. It is assumed that an accurate analysis result cannot be obtained.
[0051]
However, as can be seen from the results of FIG. 5 and FIG. 6, according to the embodiment apparatus, even if the number of sensors is significantly larger than 128 and 19 current sources, and the S / N ratio is low, the magnetic The correlation between the biological signal component and the noise component in the detection signal is made uncorrelated and the number m of independent components obtained in advance is m.0By decomposing the magnetic detection signal into independent components, the magnetic detection signal is properly decomposed into independent components, so that an accurate analysis result can be obtained.
[0052]
The present invention is not limited to the above-described embodiment, and can be modified as follows.
[0053]
(1) In the embodiment apparatus, whether or not the independent component is a noise component based on the ratio Ma / Mb between the standard deviation value Ma of the non-inspection target section and the standard deviation value Mb of the inspection target section in the independent component of the ICA signal X. However, an apparatus having a configuration for determining whether or not an independent component is a noise component based on the signal strength of the independent component of the ICA signal X can be given as a modified example. For example, a thing of 1 pT (picotesla) or more is determined as noise. However, in this modification, the signal strength of the independent component to be determined needs to be converted to a value corresponding to the raw magnetic field strength, not a normalized value.
[0054]
(2) An apparatus having a configuration for determining whether or not an independent component is a noise component based on the frequency of the independent component of the ICA signal X is also given as a modified example. For example, the most dominant frequency is obtained by applying FFT to the ICA signal X. For example, if it is 100 Hz or more, it is determined as noise.
[0055]
(3) Furthermore, an apparatus having a configuration for discriminating whether or not an independent component is a noise component by using two or three of the three noise component discrimination methods described above including the embodiment is also a modified example. Can be mentioned.
[0056]
In the above-described embodiments, the biomagnetic measurement apparatus using a magnetoencephalograph has been described as an example. However, the same effect can be obtained even with a biosignal measurement apparatus using an electroencephalograph or the like.
[0057]
【The invention's effect】
As described in detail above, according to the biological signal measuring apparatus of the present invention, the original detection signals (observation signals) detected by the plurality of sensors according to the so-called ICA (independent component analysis) technique are used for each of the biological activities. After being decomposed into independent components for each current source and each other signal source, it is determined whether each independent component is a noise component based only on the state of each independent component itself. After the components are removed, the detection signal is separated and restored for each information source by the remaining non-noise independent components. Therefore, without providing a separate sensor dedicated to noise detection, a restoration detection signal for analyzing a biological signal from which noise components have been sufficiently removed can be easily obtained using only the signal obtained by the sensor for measuring the biological signal. As a result, the signal analysis performed based on the restored living body detection signal is also accurate.
[0058]
In addition, according to the biological signal measurement device of the present invention, the biological signal component and the noise component by the biological activity current source in the original detection signal are decorrelated before being decomposed into independent components. Therefore, even if it is a large noise component, it is difficult for the noise component to be mixed into the biological signal component side when the independent component is decomposed. In addition, after decomposition, there is no fear that an independent component, which should be composed of only a single component, becomes an independent state including a plurality of independent components. Furthermore, since the original detection signal is decomposed into the number of independent components equal to the number of independent components obtained in advance, the detection signal is decomposed even when the number of sensors is significantly larger than the total number of sources. The number is sufficiently narrowed down in advance, and there is no fear that the components that should be included in only one independent component after decomposition are divided into a plurality of independent components. Therefore, even when the number of sensors is significantly larger than the total number of generation sources or when the noise level is large, the detection signal is properly decomposed into independent components, so that accurate signal analysis is performed. .
[Brief description of the drawings]
FIG. 1 is a block diagram illustrating an overall configuration of a biomagnetic measurement apparatus according to an embodiment.
FIG. 2 is a flowchart showing a state in which biomagnetism measurement is executed by the embodiment apparatus over time;
FIG. 3 is a graph showing a part of the ICA signal X obtained by the example device.
FIG. 4 is a schematic diagram showing a generation state of a bioactive current source in the occipital cerebrum analyzed by the example device.
FIG. 5 is a graph showing a time change of an analysis current source corresponding to specific nine points in the occipital cerebrum measured by the example device.
FIG. 6 is a schematic diagram showing a generation state of a bioactive current source in the occipital cerebrum analyzed by a conventional device.
FIG. 7 is a graph showing a time change of an analysis current source corresponding to specific nine points in the occipital cerebrum measured by a conventional device.
[Explanation of symbols]
1 ... Multi-channel SQUID sensor
1a SQUID sensor
2 ... Data conversion and collection unit
3 Signal processing unit
4 ... Preliminary processing section
5 ... Intercomponent elimination part
6 ... Magnetic signal decomposition part
7: Noise component removal unit
8 ... Magnetic signal restoration part
9 ... Magnetic analysis part
M: Subject

Claims (1)

被検体の診断対象領域において生体活動電流源により生じる微小な生体信号を測定する複数個のセンサを備えた生体信号計測装置において、前記複数個のセンサにより検出された検出信号における独立成分の個数を求出する独立成分個数求出手段と、前記検出信号に乗算処理を行って、前記検出信号を
(a)前記複数個のセンサにより検出された検出信号において生体活動電流源による生体信号成分と
(b)ノイズ成分
との和とすることにより、前記生体信号成分と前記ノイズ成分との相互依存性を解消する(前記生体信号成分と前記ノイズ成分との間を無相関化する)成分間相互依存解消手段と、前記生体信号成分と前記ノイズ成分との相互依存性が解消された検出信号を前記独立成分個数求出手段により求出した個数の数の独立成分に分解する信号分解手段と、この信号分解手段により分解された独立成分の中のノイズ成分を独立成分自体の状態のみに基づいて判定して取り除くノイズ成分除去手段と、各非ノイズ独立成分に基づき検出信号を復元する信号復元手段と、復元された各検出信号の各々の独立成分に対応する脳内の活動(位置、方向、強さ)を得る信号解析手段とを備えていることを特徴とする生体信号計測装置。
In a biological signal measuring apparatus having a plurality of sensors for measuring minute biological signals generated by a biological activity current source in a diagnosis target region of a subject, the number of independent components in detection signals detected by the plurality of sensors is determined. Independent component number obtaining means for obtaining , multiplying the detection signal, the detection signal
(A) in the detection signals detected by the plurality of sensors, the biological signal component by the biological activity current source;
(B) Noise component
And by the sum, the biological signal component and eliminating interdependency between said noise component (decorrelating between the biological signal component and the noise component) component between interdependence eliminating means with, the number of independent components to degrade the signal decomposition unit number that issued determined by the biological signal component and the noise component and the independent component number Motomede means a detection signal interdependence is eliminated, and this signal decomposition unit A noise component removing unit that determines and removes a noise component in the decomposed independent component based on only the state of the independent component itself, a signal restoring unit that restores a detection signal based on each non-noise independent component, and a restored A biological signal measuring apparatus comprising: signal analysis means for obtaining activity (position, direction, intensity) in the brain corresponding to each independent component of each detection signal.
JP30413299A 1999-10-26 1999-10-26 Biological signal measuring device Expired - Fee Related JP4236352B2 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP30413299A JP4236352B2 (en) 1999-10-26 1999-10-26 Biological signal measuring device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP30413299A JP4236352B2 (en) 1999-10-26 1999-10-26 Biological signal measuring device

Publications (2)

Publication Number Publication Date
JP2001120511A JP2001120511A (en) 2001-05-08
JP4236352B2 true JP4236352B2 (en) 2009-03-11

Family

ID=17929435

Family Applications (1)

Application Number Title Priority Date Filing Date
JP30413299A Expired - Fee Related JP4236352B2 (en) 1999-10-26 1999-10-26 Biological signal measuring device

Country Status (1)

Country Link
JP (1) JP4236352B2 (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102540130A (en) * 2012-01-10 2012-07-04 中国科学院半导体研究所 Signal separating method applied to on-line angular difference monitoring system of current sensor

Families Citing this family (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP4807598B2 (en) * 2002-10-31 2011-11-02 日本光電工業株式会社 Pulse photometer
JP4352315B2 (en) * 2002-10-31 2009-10-28 日本光電工業株式会社 Signal processing method / apparatus and pulse photometer using the same
DE10312519A1 (en) * 2003-03-20 2004-10-28 Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. Method for triggering occupant-assisted measures in a vehicle
US7025728B2 (en) * 2003-06-30 2006-04-11 Nihon Kohden Corporation Method for reducing noise, and pulse photometer using the method
JP4876988B2 (en) * 2007-03-13 2012-02-15 日本電気株式会社 Brain computer interface device
JP2012146116A (en) * 2011-01-12 2012-08-02 Kyushu Institute Of Technology Device and method for identifying speech content
US20210204883A1 (en) * 2016-02-19 2021-07-08 The Regents Of The University Of California Systems and Methods for Artifact Reduction in Recordings of Neural Activity
JP6785446B2 (en) 2016-04-14 2020-11-18 パナソニックIpマネジメント株式会社 Biological signal measurement system
CA3055631A1 (en) * 2017-03-15 2018-09-20 The Regents Of The University Of California Removal of stimulation artifact in multi-channel neural recordings
JP6996095B2 (en) * 2017-03-17 2022-01-17 株式会社リコー Information display devices, biological signal measurement systems and programs

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102540130A (en) * 2012-01-10 2012-07-04 中国科学院半导体研究所 Signal separating method applied to on-line angular difference monitoring system of current sensor

Also Published As

Publication number Publication date
JP2001120511A (en) 2001-05-08

Similar Documents

Publication Publication Date Title
JP4068763B2 (en) Biological signal measuring device
US7729740B2 (en) Noise cancellation in magnetoencephalography and electroencephalography with isolated reference sensors
Van Dijk et al. Prestimulus oscillatory activity in the alpha band predicts visual discrimination ability
Ghuman et al. A wavelet-based method for measuring the oscillatory dynamics of resting-state functional connectivity in MEG
JP4236352B2 (en) Biological signal measuring device
US6195576B1 (en) Quantitative magnetoencephalogram system and method
EP1049402B1 (en) Method for measuring, estimating and displaying rms current density maps
EP2624748B1 (en) Device for use in electro-biological signal measurement in the presence of a magnetic field
US5269325A (en) Analysis of biological signals using data from arrays of sensors
Yamasaki et al. Effects of distraction on pain-related somatosensory evoked magnetic fields and potentials following painful electrical stimulation
Jaušovec et al. Differences in EEG current density related to intelligence
Tang et al. Independent components of magnetoencephalography: single-trial response onset times
Parkkonen Instrumentation and data preprocessing
CN105520732A (en) Method for assessing brain function and portable automatic brain function assessment apparatus
Wei et al. Altered functional connectivity among default, attention, and control networks in idiopathic generalized epilepsy
Ichikawa et al. Subjective and model-estimated reward prediction: Association with the feedback-related negativity (FRN) and reward prediction error in a reinforcement learning task
US20120043969A1 (en) Impedance Tomography Apparatus
Wang et al. Detection of P300 brain waves using a Magneto-Impedance sensor
EP0477434B2 (en) Analysis of biological signals using data from arrays of sensors
Westin et al. Detection of interictal epileptiform discharges: A comparison of on-scalp MEG and conventional MEG measurements
Takemi et al. Is event-related desynchronization a biomarker representing corticospinal excitability?
Tuunanen et al. Comparison of BOLD fMRI and MEG characteristics to vibrotactile stimulation
Okada et al. An automatic identification and removal method for eye-blink artifacts in event-related magnetoencephalographic measurements
US20210186400A1 (en) Apparatuses, systems, and methods for suppression of artifacts in non-invasive electromagnetic recordings
Hamada et al. Effect of interstimulus interval on attentional modulation of cortical activities in human somatosensory areas

Legal Events

Date Code Title Description
RD03 Notification of appointment of power of attorney

Free format text: JAPANESE INTERMEDIATE CODE: A7423

Effective date: 20040303

A621 Written request for application examination

Free format text: JAPANESE INTERMEDIATE CODE: A621

Effective date: 20060621

A977 Report on retrieval

Free format text: JAPANESE INTERMEDIATE CODE: A971007

Effective date: 20080623

A131 Notification of reasons for refusal

Free format text: JAPANESE INTERMEDIATE CODE: A131

Effective date: 20080729

A521 Request for written amendment filed

Free format text: JAPANESE INTERMEDIATE CODE: A523

Effective date: 20080926

TRDD Decision of grant or rejection written
A01 Written decision to grant a patent or to grant a registration (utility model)

Free format text: JAPANESE INTERMEDIATE CODE: A01

Effective date: 20081209

A01 Written decision to grant a patent or to grant a registration (utility model)

Free format text: JAPANESE INTERMEDIATE CODE: A01

A61 First payment of annual fees (during grant procedure)

Free format text: JAPANESE INTERMEDIATE CODE: A61

Effective date: 20081216

R150 Certificate of patent or registration of utility model

Free format text: JAPANESE INTERMEDIATE CODE: R150

FPAY Renewal fee payment (event date is renewal date of database)

Free format text: PAYMENT UNTIL: 20111226

Year of fee payment: 3

LAPS Cancellation because of no payment of annual fees