EEG / ECoG Ontology Droplet
Signal source
Signal source !"#$%&'()*%+,-,*%./0012,% !"#$"%&'()*%+'"#%'#%,-#./-"%01-23%-(#4'%-"%+32#5-*,"+6)-7 34$5"6789*% :;<%:=>%?"9@@
Signal source
Source localization !!"#$%&#'()*+,-#.'%/+%0#+&1)(2%/+)& 3$%0'#4)')5(%'67 3)8($-#0)$%0+9%/+)& :)&/)&;#<=#>=;#?-(&./-+&;#@=#!=;#A#B8-(;#!=#4=#CDEEFG=#H+.2%/$6#&-5%/+*+/7#I+/6# *+.8%0J)&07#%&,#%8,+)*+.8%0#.'--$6=# !"#$%&'()(*"#)+,-&./01234-&.562 ./78&
Constrained localization = v Es where v is th = m Bs m v E = = = 2 + " W ! 1 ! A x As x , A , where # W = Wx ! s m B Err (1) = + x As n W 2 = " ! # Err W Wx s 2 = " ( + ) ! # Err W W As n s 2 = " ( ! ) s + # WA I Wn 2 = " + # Ms Wn , where = ! M WA I Ms 2 Wn 2 = " # + " # Tr MRM T Tr WCW T = ( ) + ( ) ! Dale & Sereno (1993) 1 RA T ARA T = ( + ) W C
!"#$%&'#()%#*+,-%#$%.)( • #',-(-/01+%2#+3%4*0+-5+2-6+#7,#'.8#$%)(+ 8)$.,4()%.-$3+92)$/#+!:;+9-8,-$#$%+ ()%#$90<)8,(.%4*# – =)>.$/+%2#+9-$$#9%.-$+?<6+)$+!:;+#55#9%+)$*+)+?').$+#55#9%+ 9)$+?#+%'.9>0 • :#9-88#$*#*+'#)*.$/1 – @49>A+BC+DEFFGHC+ !"#$"%&'()*%+'"#%'#%,-#./-"%01-23%-(# 4'%-"%+32#5-*,"+6)-7#5,-#8$5#4&-99:#;3<=&+(>-#8!? • B-8#+IJ-%92)3K+62.(#+'#)*.$/+!:;+,),#'31 – L-%+#"#'0-$#+43#3+%2#+3)8#+'#5#'#$9#+#(#9%'-*# – B-8#%.8#3+$#/)%."#+.3+4, – M#6)'#+-5+3,)%.)(+9().83 – N2#''0&,.9>.$/+.3+3%)$*)'*+,')9%.9# – M#6)'#+-5+?.)3#*+8#)34'#3
The good, the bad ... • Cheap, easy to use • High temporal resolution • Clinical use for anesthesia, epilepsy • Research use for sleep, attention, cognition, perception • Very poor spatial resolution • Many artifacts: eye movement, blinking, facial gestures, heart activity
!"#$%#&'()*+#',"%- • **./0 • 1"234,45&26) 7"#$%#&'()82&39: – ;#6,2)<=>?)@AB – 1C#,2)<?>D)@AB E&3#"9#&L)*M)NML)@466(2"3L)*M)EML)K)OP66#"L)OM)OM)<FJJDBM)E,,#&,45&)72'464,2,#9) – E6+C2)<D>=F)@AB -%6,4+6#)9,4-%6%9)7#2,%"#9)4&)+2"266#6)4&)C%-2&)Q49%26)'5",#RM) !"##$%&'()*+*,-.'/01/23.'/4456/4478' – G#,2)<=F>F?)@AB !"#$%&'$()$*+, – H2--2)<)IJ)K)%+B
What’s it good for a F 3 -F z Disk Screw R L Solenoid C 3 -F z Grating Finger P 3 -F z b c O 1 -F z 1 µ V 100 ms a b 100 100 80 80 10 ms 60 60 180 ms 400 ms 40 40 20 20 0 0 Air M4 L3 R3 M2 Air M4 L3 R3 c d 100 100 80 80 60 60 40 40 20 20 0 0 Air M4 L6 R6 SC Air M4 L6 R6 SC R L Zangaladze (1999)
What’s it good for Varela (1999)
Correlation is not causation, right?
Big data • Cheap sets ✓ Neurosky, eMotiv ✓ Toys, games (Mindflex) ✓ Kickstarter project ✓ Carnegie Mellon (Bryan Murphy) • Massive repositories ✓ G. Church - Harvard
... and the ugly Electrocorticography (ECoG) • Electrodes under the dura • Many fewer artifacts (eyes, facial, scalp diffusion) • Limited used in humans: epileptic ablation pre- operative guidance
� � � � � � Flexible, foldable, actively multiplexed, high-density - Nature Neuroscience , Dec. 2011 electrode array for mapping brain activity in vivo Jonathan Viventi 1,2,13 , Dae-Hyeong Kim 3,13 , Leif Vigeland 4 , Eric S Frechette 5 , Justin A Blanco 6 , Yun-Soung Kim 7 , Andrew E Avrin 8 , Vineet R Tiruvadi 9 , Suk-Won Hwang 7 , Ann C Vanleer 9 , Drausin F Wulsin 9 , Kathryn Davis 5 , Casey E Gelber 9 , Larry Palmer 4 , Jan Van der Spiegel 8 , Jian Wu 10 , Jianliang Xiao 11 , Yonggang Huang 12 , Diego Contreras 4 , John A Rogers 7 & Brian Litt 5,9 a b 1.6 + V Output 10 –4 60 1 mm 5V Row Select 1.2 10 –5 40 I d (mA) I d ( µ A) I d (A) Elect 0.8 10 –6 3V -rode 20 0.4 10 –7 1V 0 0 Buffer –2 0 2 4 6 0 1 2 3 4 Multiplexer V d (V) V g (V) c Pt contact Pt electrodes VIA Multilayer offset VIA structure 1 st ML Horizontal / 2 nd 300 µ m vertical ML interconnect 200 µ m Doped Si ribbons on Figure 1 Flexible, high-resolution multiplexed electrode array. polyimide Si ( a ) Photograph of a 360-channel high-density active electrode d array. The electrode size and spacing was 300 × 300 � m and 2 mm 500 � m, respectively. Inset, a closer view showing a few unit cells. ( b ) Schematic circuit diagram of single unit cell containing two matched transistors (left), transfer characteristics of drain-to- source current ( I d ) from a representative flexible transistor on linear (blue) and logarithmic (red) scales as gate to source voltage ( V g ) was swept from − 2 to +5 V, demonstrating the threshold voltage ( V t ) of the transistor (center). Right, current-voltage characteristics of a representative flexible silicon transistor. I d was plotted as a function of drain-to-source voltage ( V d ). V g was varied from 0 to 5 V Side Right Left in 1-V steps. ( c ) Schematic exploded view (left) and corresponding e f microscope image of each layer: doped silicon nanoribbons (right 0.8 0.3 Si PI 12.5 µ m Stiffness (10 –6 Nm 2 ) frame, bottom), after vertical and horizontal interconnection with SiO 2 PI 25 µ m 0.6 Strain (%) 0.2 arrows indicating the first and second metal layers (ML, right Metal frame, second from bottom), after water-proof encapsulation (right 0.4 Previous 0.1 frame, third from bottom) and after platinum electrode deposition 0.2 Current (right frame, top). Green dashed lines illustrated the offset via 0 0 structure, critical for preventing leakage current while submerged 400 600 800 1,000 0 10 20 30 in conductive fluid. ( d ) Images of folded electrode array around low Thickness of epoxy ( µ m) Bending radius ( µ m) modulus polydimethylsiloxane (PDMS) insert. ( e ) Bending stiffness of electrode array for varying epoxy thicknesses and two different polyimide (PI) substrate thicknesses. A nearly tenfold increase in flexibility between the current device and our prior work was shown. ( f ) Induced strain in different layers depending on the change in bending radius.
� � � � � � � � � � a a d , d f n g o . b ss n h ) 5 mm ch 10 µ V 100 µ V b h n • d Multiplexing along column, speed <5µsec e. ls , • y Sampling rate > 10kS/sec - i- • Low cross-talk f e s- • se Sampling area 10 x 9 mm r 65 ms 145 ms e - c • Lateral d 21 Claim: sample 80 x 80 mm, 25,600 channels 7 e Frontal Occipital at > 1.2 kS/sec 19 k d Medial re 18 17 e Prediction incorrect of Prediction off by one square t Prediction correct o n selecting 10 of the 15 trials, averaging the evoked responses and repeat-
Finger Movement Classification for an Electrocorticographic BCI - Neural Engineering , 2007 Pradeep Shenoy Kai J. Miller Jeffrey G. Ojemann Rajesh P.N. Rao Dept. of Computer Science and Engineering University of Washington { pshenoy,kai,rao } @cs.washington.edu, jeff.ojemann@seattlechildrens.org Classifying Individual Fingers als, 0.6 LPM SVM 0.5 0.4 Error 0.3 0.2 0.1 channel. 0 s1 s2 s3 s4 s5 s6 Subject Fig. 1. Classifying finger movement activity: The figure shows the 5-class cross-validation error for the LPM and SVM classifiers, across 6 features, subjects. The results show that a high degree of accuracy is possible in distinguishing individual finger movements using ECOG. Also, the LPM consistently outperforms the SVM. (Chance level for a 5-class problem is 80% error.)
Reconstructing Speech from Human Auditory Cortex Brian N. Pasley 1 * , Stephen V. David 2 , Nima Mesgarani 2,3 , Adeen Flinker 1 , Shihab A. Shamma 2 , Nathan E. Crone 4 , Robert T. Knight 1,3,5 , Edward F. Chang 3 1 Helen Wills Neuroscience Institute, University of California Berkeley, Berkeley, California, United States of America, 2 Institute for Systems Research and Department of Electrical and Computer Engineering, University of Maryland, College Park, Maryland, United States of America, 3 Department of Neurological Surgery, University of California–San Francisco, San Francisco, California, United States of America, 4 Department of Neurology, The Johns Hopkins University, Baltimore, Maryland, United States of America, 5 Department of Psychology, University of California Berkeley, Berkeley, California, United States of America Improved reconstruction adding wavelet analysis, to account for frequency sweeps
Recommend
More recommend