2. ICA ICA () (Blind Source Separation BBS) 2) Fig. 1 Model of Optical Topography. ( ) ICA 2.2 ICA ICA 3) n 1 1 x 1 (t) 2 x 2 (t) n x(t) 1 x(t

Size: px

Start display at page:

Download "2. ICA ICA () (Blind Source Separation BBS) 2) Fig. 1 Model of Optical Topography. ( ) ICA 2.2 ICA ICA 3) n 1 1 x 1 (t) 2 x 2 (t) n x(t) 1 x(t"

あいりつまがみ
7 years ago
Views:

1 ICA (Independent Component Analysis) Denoising Method using ICA for Optical Topography Yamato Yokota, 1 Tomoyuki Hiroyasu 2 and Hisatake Yokouchi 2 Optical topography is one of the promising ways to analyze the activity of the brain. One of the problems in analyzing is the noise.optical topography irradiates the near-infrared light from the scalp, and measures the cerebral blood flow from the reflection strength.therefore, it data include blood flow except cerebral blood flow and various sensor noises. Usually Optical topography is recorded using a digital band passed filter and averaging are applied to the data.but these filters have the possibility of loss the useful information.therefore we used ICA (Independent Component Analysis) and biosignal sensors for denoising. In this paper, we reported the denoising experimental result with the use of optical topography data. 1. NIRS Near Infra-Red Spectoroscopy EEG (Electroencephalography, ) MEG (Magnetoencephalography, ) 20 30mm oxy-hb doxy-hb oxy-hb deoxy-hb 1) (Independent Componet Analysis, ICA) ICA ICA 1 Graduate School of Life and Medical Sciences, Doshisha University 2 Faculty of Life and Medical Sciences, Doshisha University 1 c 2011 Information Processing Society of Japan

2. ICA ICA () (Blind Source Separation BBS) 2) 2.1 1 1 Fig. 1 Model of Optical Topography. ( ) ICA 2.2 ICA ICA 3) n 1 1 x 1 (t) 2 x 2 (t) n x(t) 1 x(t) = (x 1(t), x 2(t),..., x n(t)) T, (t = 1, 2,.

2 2. ICA ICA () (Blind Source Separation BBS) 2) Fig. 1 Model of Optical Topography. ( ) ICA 2.2 ICA ICA 3) n 1 1 x 1 (t) 2 x 2 (t) n x(t) 1 x(t) = (x 1(t), x 2(t),..., x n(t)) T, (t = 1, 2,..., l) (1) t l m 2 2 s(t) = (s 1(t), s 2(t),..., s m(t)) T, (t = 1, 2,..., l) (2) x(t) s(t) n m A 3 x(t) = As(t) (3) x(t) s(t) A A W w ij 4 s i (t) = w ij x j (4) j s(t) A c 2011 Information Processing Society of Japan

3 rank(a) ( 1 ) ( 2 ) ( 3 ) Kullback-Leibler ICA ( ) s A s i A a i 1 FastICA 1 4) 2.3 ICA ICA ICA 2 ( 1 ) 2 (x, y) = (x i, y i )(i = 1, 2, 3,..., n) 5 n i=1 (x i x)(y i ȳ) n i=1 (x i x) 2 n i=1 (y i ȳ) 2 (5) ( 2 ) ( 3 ) ( 4 ) ( 5 ) ICA FastICA 2 ICA ( 6 ) 3. ETG ICA ICA ICA 2 3 c 2011 Information Processing Society of Japan

Fig. 3 3 Blood stream change according to pulse. 2 Fig. 2 Measurement position. 1.5Hz ICA 1 2 Table 1 1 Experiment environment ETG-7100 (Probe5 Oxy-Hb) 22 0.

4 Fig. 3 3 Blood stream change according to pulse. 2 Fig. 2 Measurement position. 1.5Hz ICA 1 2 Table 1 1 Experiment environment ETG-7100 (Probe5 Oxy-Hb) sec BM innovation BM-CS5-US Fig. 4 Blood stream of CH ICA 5 ICA c 2011 Information Processing Society of Japan

5 情報処理学会研究報告図 6 ICA7 の周波数スペクトル Fig. 6 Result of FFT to ICA7. 被験者の平均心拍数は 85.6 回 (約 1.42Hz) であったため各独立成分の周波数スペクトルを調べたところ ICA7 が 1.46Hz にピークを持つことが分かった ICA7 を FFT(Fast Fourier Transform) により周波数成分に分解した結果を図 6 に示すセンサで計測した心拍数と近い周期の成分を多く持つ ICA7 を脈拍に起因する成分とみなし ICA を除去した状態で元の光トポグラフィデータへ ICA の逆変換を行ったその際のチャンネル 1 のデータを図 7 に示す考察 ICA により脈拍に起因する血流変化を除去したことでバンドパスフィルタや移動平均処理を行わずに脈拍による振動を取り除いた光トポグラフィデータを得ることが出来た本実験で除去した ICA7 には 1.46Hz 付近の周波数成分の他にも 0.1Hz 2.9Hz 付近の成分も含んでいた ICA の特徴からこれらの成分は脈拍と考えられる信号に対して独立した成分でなかったと仮定した場合脈拍と脈拍により誘発された信号 (ノイズ) を除去できたと考えられる 3.2 頭部動作ノイズの除去本節では ICA を使用した光トポグラフィデータからの頭部動作に関するノイズ除去実験の結果を示す実験には 3 章で示した機器被験者を使用し被験者には頭部を前方へ 45 図 5 ICA による出力 Fig. 5 Result of ICA 5 c 2011 Information Processing Society of Japan

6 Fig. 7 7 ICA7 1 Result of removing ICA7 from Optical Topography Data. 8 1 Fig. 8 Blood stream of CH1. 5) ICA ICA 9 ICA 2 10 ICA ICA1 ICA1 1 Texas Instruments EZ430-Chronos ICA ICA ICA 8 4. ICA 2 ICA 6 c 2011 Information Processing Society of Japan

7 Fig ICA1 1 Result of removing ICA7 from Optical Topography Data. 9 ICA Fig. 9 Result of ICA ICA ICA 10 Acceleration sensor data. 1),,.. 12, No.47, pp , ).., ),.., ) Aapo Hyvrinen and Erkki Oja. Independent component analysis: A tutorial. 5),,.., No.10, pp , c 2011 Information Processing Society of Japan

DPA,, ShareLog 3) 4) 2.2 Strino Strino STRain-based user Interface with tacticle of elastic Natural ObjectsStrino 1 Strino ) PC Log-Log (2007 6)

DPA,, ShareLog 3) 4) 2.2 Strino Strino STRain-based user Interface with tacticle of elastic Natural ObjectsStrino 1 Strino ) PC Log-Log (2007 6) 1 2 1 3 Experimental Evaluation of Convenient Strain Measurement Using a Magnet for Digital Public Art Junghyun Kim, 1 Makoto Iida, 2 Takeshi Naemura 1 and Hiroyuki Ota 3 We present a basic technology