1 / 78

Using Graphs to Mine Brain Dynamics

Using Graphs to Mine Brain Dynamics. Introduction. Metho do logical framework (short description). Application to Auditory M100-responses to detect the influence of attention. Future trends.

debra
Download Presentation

Using Graphs to Mine Brain Dynamics

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Using Graphs to Mine Brain Dynamics

  2. Introduction Methodological framework (short description) Application to Auditory M100-responses to detect the influence of attention Future trends

  3. A "graph" is collection of vertices (or 'nodes‘)and a collection of edges that connect pairs of vertices. ‘‘Seven Bridges of Königsberg’’ Initiated with Euler’s paper in 1736 Introductory comments Graph theory is the study of graphs: mathematical structures used to model pairwise relations between objects from a certain collection.

  4. created by O. Sporns Graphs in Brain Research : network analysis (small-world network) systems-approach (MI-maps) 

  5. Graphs for Single-TrialAnalysis • Past Liu, Ioannides, Gartner Elect. and clinical Neurophysiology 106 (1998) 64–78 • Current-practise • Future-trends (mining informationfrom multisite recordings)

  6. I. Outline of our methodological approach A synopsis of response dynamics and its variability by means of Semantic Geodesic Maps Graphs play an instrumental role in : computing neighborhood relations, deriving faithfull visualizations of reduced dimensionality describing and contrasting the essence of response variability in an objective way MST, WW-test , ISOMAP, Laplaceans , commute times, etc Brain dynamics can be compared at a glance

  7. IEEE SP Magazine, 2004 [ Laskaris & Ioannides, 2001 & 2002 ]

  8. Step_the spatiotemporal dynamics are decomposed Design of thespatial filterused to extractthetemporal patternsconveyingtheregional response dynamics

  9. Clustering & Vector Quantization Feature extraction Embedding in Feature Space Minimal Spanning Treeof the codebook Interactive Study of pattern variability Orderly presentation of response variability MST-ordering of the code vectors Step_ Pattern & Graph-theoretic Analysisof the extracted ST-patterns

  10. - Step_Within-group Analysisof regional response dynamics

  11. Step_Within-group Analysis of multichannel single-trial signals

  12. Step_Within-group Analysis of single-trial MFT-solutions

  13. Wald-Wolfowitztest (WW-test) {Xi}i=1...N vs {Yj} j=1...M    R=12   Dissimilar Similar  P-value 0 A non-parametric test for comparing distributions

  14. II. Application to MEG Auditory (M100) responses The Scope : to understand the emergence of M100-response (in averaged data) characterize its variablility (at Single-Trial level) and describe the influence (if any) of attention

  15. The Motive :‘‘ New BCI approaches based on selective Attention to Auditory Stimulus Streams ’’N. Hill, C. Raths (mda_07) Exogenous (i.e. stimulus-driven) BCI's rely on the conscious direction of theuser's attention. For paralysed users, this means covert attention Covert attention do affect auditory ERPs.

  16. MEG-data were recorded at RIKEN (BSI, Japan) CTF-OMEGA (151-channels)

  17. Repeated stimulation ( ISI: 2sec ) Count # 2 sec binaural-stimuli [ 1kHz tones, 0.2s, 45 dB ], passive listening task ( 120 trials ) & attenting task ( 120 ± 5 trials ) right-hemisphere response left-hemisphere response

  18. IIa. Ensemble characterization of (M100-related) brain waves [ 3-20 ] Hz IIb. Unsupervised classification of (M100-related) brain waves and Prototyping IIc. Empirical Mode Decomposition for enhancing(M100-related) brain waves mutiscale analysis single-channel ICA for oscillatory components

  19. IIa. Ensemble characterization of (M100-related) brain waves

  20. time-aligned ST-segments are extracted brought to common feature space and subsequenlty transfromed ( via MDS ) to point-diagrams representing brain-waves relative scattering

  21. Attentive-responses show significantly reduced scattering : smaller MST-length

  22. normalize Clustering tendencies are apparent in phase-representation space and higher for theattentive responses

  23. Similar observations can be made for the other hemisphere of the same subject

  24. And for other subjects as well

  25. The variability of (snaphosts of) Brain-waves is smaller for the attentive task

  26. The clustering of (snaphosts of) Brain-waves is higher within the phase-representation domain

  27. IIb. Unsupervised classification of (M100-related) brain waves and Prototyping

  28. Discriminative Prototyping Discriminative Prototyping Kokiopoulou & Saad, Pattern Recognition : ‘‘Enhanced graph-based dimensionality reduction with repulsion Laplaceans’’ By contrasting brain-waves from control condition against the M100-related brain waves we deduce an abstract space wherein Neural-Gas based Prototyping is first carried out and then prototypes are ranked based on an ‘‘SNR-classification index’’

  29. D-r2 Laplacean & eigenanalysis D-r1 Repulsion-graph

  30. By embedding brain-waves from control condition, we can the define high-SNR regions in the reduced-space Exploiting the abundance of spontaneous brain activity snapshots, we can accurately/precisely rank the differnt voronoi regions.

  31. The derived ranks are utilized to order the prototypical responses accordingly

  32. Taking a closer look at the high-SNR group of STs, a ‘highlighting’ of response dynamics is achieved

  33. that can be enhanced via a ‘‘Trial-Temporal’’ format There is an apparent organization in the brain waves @ 100 ms

  34. And be enriched via contrast with the ‘void-of-M100-response’ ST-group

  35. There is lack of any kind of organization in the brain waves

  36. Discriminative Prototypingfor attentive(M100) responses

  37. Even for theattentive-taskthere are ST-groups‘void-of-M100’

  38. The high-SNR group of STs, clearly shows a phase-reorganization of prestimulus activity accompanied with an enhancement of oscillations

  39. portrayed even better in the ‘Trial-Temporal’ format

  40. On the contrary, the low-SNR group of STs

  41. shows no prominent stimulus-induced changes

  42. Similar observations can be made for the other hemisphere of the same subject

More Related