Download as PDF, PPTX
More Related Content
Similar to Brain-Computer Interface and States of Vigilance
![ANIMAL_CELL_,_TISSUE_AND_ORGAN_CULTURE[1].pptx](https://cdn.slidesharecdn.com/ss_thumbnails/animalcelltissueandorganculture1-260204172026-4462b440-thumbnail.jpg?width=640&height=640&fit=bounds)
![Polymer [ बहुलक ] Chemistry Notes PDF - Irfanullah Mehar - JJ Sir Chemistry.pdf](https://cdn.slidesharecdn.com/ss_thumbnails/polymerchemistrynotespdf-irfanullahmehar-jjsirchemistry-260210172118-3f9b37f7-thumbnail.jpg?width=640&height=640&fit=bounds)
Brain-Computer Interface and States of Vigilance
- 1. Coma Science Group- GIGA University of Liège, Belgium Brain-Computer Interface and States of Vigilance 03/01/2017 Stephen Larroque stephen.larroque@ulg.ac.be
- 2. Crucial question How tocommunicate with someone who cannot communicate?
- 3. Crucial question How tocommunicate with someone who cannot communicate? ... communicate with their brain!
- 4. BCI Outline 1. Identifyingthe (sub)population 2. Identifying biomarkers 3. Modelling (= decoder) 4. Applications: detection, communication & rehabilitation
- 5. Overview 1. Identifying thepopulation • Disorders of consciousness • Clinical diagnoses 2. Identifying biomarkers • Functional neuroimagery and EEG 3. Modelling • Brain-computer interfaces (BCI) 4. Applications: communication & rehabilitation • Conclusions
- 6. Coma Science Group- GIGA University of Liège, Belgium Disorders of consciousness
- 7. Disorders of consciousness(DOC) Bruno, Vanhaudenhuyse et al J Neurology, 2011
- 8. A new namefor «vegetative»: unresponsive “There’s nothing we can do… he’ll always be a vegetable.” Laureys et al, BMC Medicine 2011
- 9. Disorders of consciousness HimanshuBansal Foundation, pre and post stem cell administration, january 2016
- 10. Coma: diagnostic criteria(signs) • No eye opening even with stimulation • No sign of awareness of self or environment • Transitory state: from 1 hour to ~1 month • Often due to brainstem damages • Respiratory assistance necessary AWARENESS WAKEFULNESS Posner et al, 2007
- 11. Vegetative state/Unresponsive wakefulness syndrome •No sign of awareness of self or environment • No coherent and voluntary responses to visual, auditive, tactile or nociceptive stimulations • Wake-sleep cycle • Hypothalamic and brainstem functions (ie, can breathe without help) • Rarely: vocalizations, emotions • Can be stable or evolve (or regress…) AWARENESS WAKEFULNESS Posner et al, 2007
- 12. Minimally conscious state •Awareness of self and environment: • Visual pursuit and/or fixation (mirror, person or object) • Appropriate smiling or crying • Object localization • Object manipulation • Command following • Non-functional communication (out-of-context) • Intelligible verbalizations Emergence from MCS: functional communication or functional object use Beware of aphasia! AWARENESS WAKEFULNESS Posner et al, 2007
- 13. Do they feelpain? Low level primary sensory area Activation But no secondary sensory area Functionally disconnected cortical activation Noxious electrical stimulation Laureys et al, Neuroimage, 2002 Laureys, Nature Reviews Neuroscience, 2005
- 14. Pain in minimallyconscious state Laureys et al, Neuroimage, 2002 Boly et al Lancet Neurology, 2008
- 15. Auditory perception Laureys etal., Brain, 2000 Boly et al, Archives of Neurology, 2004 DISCONNECTED CONNECTED
- 16. Locked-in syndrome American Congressof Rehabilitation Medicine, 1995 WAKEFULNESS AWARENESS • Spontaneous eye opening • Aphonia or hypophonia • Quadriplegia or quadriparesia • Communication with horizontal or vertical eye movements or blinking • Preservation of consciousness and cognitive functions
- 17. DOC summary Posner etal, 2007
- 18. Locked-in syndrome Demertzi etal, Informa Brain Injury, 2014 (psychologists, nurses) (students, engineers) (n=3332)
- 19. Coma Science Group- GIGA University of Liège, Belgium Clinical Diagnosis 05/01/2015
- 20.
- 21.
- 22. Diagnostic error n=103 post-comatosepatients – 45 clinical consensus diagnosis ‘vegetative state’ – 18 signs of awareness (Coma Recovery Scale-Revise) 40% potential misdiagnosis Schnakers et al, BMC Neurology 2009
- 23. GCS/GLS/FOUR Wijdicks et al.,Ann Neurol (2005) Teasdale G, Jennett B, Lancet (1974) Born, Acta Neurochir (1988)
- 24. GCS, GLS orFOUR? Bruno et al, Neurocritical Care, 2011 n=146 131 intubated (74%) Inter-rater reliability - GCS (к 0.65), GLS (к 0.66), FOUR (к 0.75) Outcome prediction - GCS ≃ GLS ≃ FOUR Different sensitivity to subpopulations: - GCS: 71 diagnosed “vegetative”/unresponsive - FOUR: identified 8 MCS- (eye tracking - 11%) n=60 GCS : 29 diagnosed “vegetative”/unresponsive FOUR : 24 PVS/UWS Schnakers et al, Annals of Neurology, 2006
- 25. Coma recovery scalerevised
- 26. Visual fixation =reflex Bruno et al, BMC Neurology 2010
- 27. Visual pursuit: usea mirror! Vanhaudenhuyse et al., JNNP 2008 Assessed without a mirror: 29% VS
- 28. New knowledge, newscales • FOUR Visual subscale LIS detection • CRS-R MCS criteria Better diagnosis sensitivity • Visual pursuit use a mirror! • Ambiguous signs of consciousness : e.g. visual fixation
- 29. OUTCOME (Projet fédéralBelge) Unresponsive state (n=296) TraumaticNon-TraumaticAnoxic % % % Minimally conscious state (n=271) N=107 N=94 N=95 N=100 N=128 N=43 EMERGENCE MCS Dead UWS
- 30. Coma Science Group- GIGA University of Liège, Belgium Neuroimagery and biomarkers 05/01/2015
- 31. Consciousness ≠ globalbrain function Laureys et al., Lancet Neurology, 2004 (=MCS)
- 32. Consciousness ≈ frontoparietal Laureyset al, Neuroimage 1999 areas that are systematically dysfunctional in the unresponsive state areas that recover metabolism after recovery from the unresponsive state Laureys et al, J Neurol Neurosurg Psychiatry, 1999
- 33. “Resting state” fMRIdefault mode connectivity Vanhaudenhuyse and Noirhomme et al., Brain 2010 Soddu et al., Human Brain Mapping 2011
- 34. Consciousness ≈ connectivity EEG-TMS Rosanovaand Gosseries et al, Brain 2012
- 35. Consciousness ≈ integration? Casaliand Gosseries et al, 2013
- 36.
- 37.
- 38. Event-related potentials (ERP) Gueritet al., Progress in Brain Research 2005 • Activity onset following sensory stimulation • Evoked potentials vs Induced potentials • P = positive • N = negative • Number = time after stimulus onset (msec)
- 39. Event-related potentials -Averaging Adapted from Tallon-Baudry and Bertrand, 1999 • Average potential across trials / subjects • Need to synchronize potentials: locking time • 2 types of locking: - evoked: lock at stimulus onset time (left squares) - induced: lock at potential’s peak (right squares)
- 40. P300 = biomarkerof surprise • P300 = ~300 msec positive wave after stimulus onset • General biomarker of surprise • Happens in all sensory modalities, often studies in visual or auditory • Two subtypes: - P3a = new stimulus, amplitude ∝ surprise (beware of habituation!) - P3b = memorize and wait until item appears. Amplitude ∝ task complexity, vigilance, motivation, stimulus rarity, etc… • Oddball paradigm (see also von Restoff or Sternberg): 1. Ask subject to remember an item 2. Present several stimuli sequentially 3. Present the item to be remembered 4. Capture P3b • Beware: surprise => P3, but P3 => surprise!
- 41. Late ERP /Late Positive Component Minimally conscious state µV 0.0 -2.5 -5.0 -7.5 -10.0 2.5 5.0 7.5 10.0 Other Name N1 ms100 300 500 800 1000 1300200 400 600 700 900 1100 1200 15001400 P3 Own Name Laureys, Perrin et al., Neurology, 2004 Perrin et al, Archives in Neurology, 2006
- 42. Predicting outcome inchronic DOC Unresponsive Wakefulness syndrome? ATYPICAL ‘HIGH LEVEL’ CORTICAL ACTIVATION Di et al, Neurology, 2007 ACTIVATION TO THE OWN NAME Di et al, Clinical Medicine, 2008
- 43. Automated consciousness classifier Phillipset al, NeuroImage, 2011 “Relevance Vector Machine” on FDG-PET data in DOC
- 44. Overview 1. Identifying thepopulation • Disorders of consciousness • Clinical diagnoses 2. Identifying biomarkers • Functional neuroimagery and EEG 3. Modelling • Brain-computer interfaces (BCI) 4. Applications: communication & rehabilitation • Conclusions
- 45. Coma Science Group- GIGA University of Liège, Belgium Brain-Computer Interface 05/01/2015
- 46. Disorders of consciousness Bruno,Vanhaudenhuyse et al J Neurology, 2011
- 47. fMRI proof ofconcept • 30-seconds period of mental imagery/resting • Each imagery task was repeated 10 times. Owen, Coleman, Boly, Davis, Laureys and Pickard, Science, 2006 Boly et al, NeuroImage 2007
- 48. fMRI proof ofconcept • 30-seconds period of mental imagery/resting • Each imagery task was repeated 10 times. Owen, Coleman, Boly, Davis, Laureys and Pickard, Science, 2006 Boly et al, NeuroImage 2007
- 49. Statistical parametric mapping Changesin the neural signal associated with the presentation of a stimulus General Linear Model Design matrix Y = X x β + ε Adapted from “Methods for dummies 2011-12” Hikaru Tsujimura and Hsuan-Chen Wu Model and predictions!
- 50. Result of SPM •30-seconds period of mental imagery/resting • Each imagery task was repeated 10 times. Owen, Coleman, Boly, Davis, Laureys and Pickard, Science, 2006 Boly et al, NeuroImage 2007
- 51. Brain – ComputerInterface: from neural signal to model to BCI Wolpaw et al, Clin Neur 2002 ?
- 52. Communication using TennisTest Monti & Vanhaudenhuyse, Coleman, Boly, Pickard, Tshibanda, Owen, Laureys New England J Med 2010
- 53. Coma Science Group- GIGA University of Liège, Belgium Communication 05/01/2015
- 54. Sensorimotor rythms Wolpaw etal., 2002
- 55. EEG-based Brain ComputerInterfaces SCIENCE GROUP COMA Cruse et al, Lancet 2012 3/16 VS/UWS (19%) - 2/5 traumatic (40%) - 1/11 non-traumatic (9%) Cruse et al, Neurology 2012 7/23 MCS (30%) - 7/15 traumatic (49%) - 0/8 non-traumatic (0%)
- 56. ERP proof ofconcept ms-200 50 300 550 800 1050 1300 -5 -10 -15 -20 -25 5 10 15 20 25 Pz (µV) Count own name Coma or total Locked-in syndrome? Listen to other name Count other name Listen to own name Schnakers et al, Neurology, 2008 Schnakers et al, Neurocase, 2009
- 57. Auditory P300 Lule &Noirhomme et al, Clinical Neurophysioloy 2013
- 58. Evoked potentials Lule &Noirhomme et al, Clinical Neurophysioloy 2013
- 59. Signal processing • Trainingset • Classifier • Linear discriminant analysis • Step-wise LDA
- 60. Auditory P3 results Lule& Noirhomme et al, Clinical Neurophysioloy 2013
- 61. Vibrotactile stimulation Lehembre etal, Clin EEG and Neuroscience 2012 • 6 LIS patients • 5/6 with visual deficit • All able to elicit a P3
- 62. Steady state evokedpotential Lesenfants et al., 2011 Guger et al., 2012
- 63. Covert SSVEP Use covertattention Lesenfants et al., 2011
- 64. Covert SSVEP • 7/12above chance level • 5/12 communication • 1/4 LIS communication Lesenfants et al., 2014
- 65. EMG of commandfollowing « Move your right hand » Bekinschtein et al, JNNP 2008
- 66. Coma Science Group- GIGA University of Liège, Belgium Rehabilitation 05/01/2015
- 67. Neuroprosthetics Rupp and Gerner,2007 Pfurtscheller, et al., 2008 • High cervical spinal cord injuries - below C4 (C1-3 are too vital) • Speed performance of these systems is low • Training, training, training
- 68. Linear model prostheticarm.. in 85 days! Collinger et al, Lancet, 2013
- 69. Feedback • Ecological feedback •Recruit and/or reinforce patient’s sensorimotor experience Pfurtscheller, et al., 2008
- 70. Direct brain stimulationto harness brain’s plasticity for rehabilitation Losey et al, Frontiers in Robotics and AI, 2016 Hidden maze game 92% success with TMS inputs, 15% without!
- 71. Information on theprocess • Recruited process (brain/motoneurons) • Recruited muscles • Hybrid Wagner et al., 2012
- 72. Do we reallyneed to access the brain? Intention decoding Repin, 1888 Yarbus, 1967 Faisal et al, 2015
- 73. Do we reallyneed the brain? Intention decoding Repin, 1888 Yarbus, 1967 Faisal et al, 2015
- 74. Do we reallyneed the brain? Intention decoding Repin, 1888 Yarbus, 1967 Faisal et al, 2015
- 75. Do we reallyneed the brain? Intention decoding Repin, 1888 Yarbus, 1967 Faisal et al, 2015 Pupil dilation for locked-in patients (Stoll et al, 2013)
- 76. Coma Science Group- GIGA University of Liège, Belgium Conclusion 05/01/2015
- 77. BCI for DOCand other pathologies • BCIs may enable the detection of response to command and communication • Results need to be interpreted with great caution • Statistical validation • Be aware of false negative • Numerous challenges: subpopulation identification, feature extraction, hardware restrictions, physical restrictions (lots of noise!), expensive, etc. • Results could have a high impact on rehabilitation strategies, quality of life and prognosis • Exciting era for BCI, lots of AI models and wonderful technological advances!
- 78. Quality of life Brunoet al., 2011 Anamnestic Comparative Self Assessment (Bernheim et al) X X Locked-in patients average n=70 X X Matched healthy controls average n=70
- 79. Further information • Seeprevious slides for references • Books: • The neurology of consciousness: cognitive neuroscience and neuropathology , Laureys, Steven, Olivia Gosseries, and Giulio Tononi, Academic Press, 2015. • Videos, documentaries: • Charlie Ford Coma 1 to 7 • Holly’s Road To Recovery - A TBI Survivor • Human Neuroscience in the wild – Aldo Faisal – FSC 2016
- 80. Coma Science Group- GIGA University of Liège, Belgium Thank you! 05/01/2015
- 81. MRI: DTI &spectroscopy Tshibanda et al, Prog Brain Res, 2009 Tshibanda et al, Neuroradiology, 2010
- 82. Combining neuroimaging approach Brunoet al, Prog Brain Res, 2011 Erik Ziegler, Cyclotron Art Committee
- 83. Emotions in MCSpatients Laureys et al., Neurology, 2004
- 84. Statistical analysis •Statistical analysisof latency, amplitude •For all derivations and potentials •ANOVA, t-test •Group level
- 85. BCI features • Synchronous/asynchronous •Dependent/independent • Robust to artifacts
- 86. Coma Science Group- GIGA University of Liège, Belgium Challenges 05/01/2015
- 87. Should we trustthe machine? A Soddu
- 88. Statistical test: auditoryP3 Permutation test: 8/14 Binomial test: 7/14 8% 4% accuracy
- 89. False Negative References Techniqueused- Brain response False negative /Number of MCS+ Monti, et al. 2010 fMRI- imagery 17/18 Bardin et al., 2011 fMRI - imagery 2/5 Schnakers et al. EEG- P3 2/9 Lulé et al. , 2012 EEG-P3 5/6 Goldfine et al, 2011 EEG-motor imagery 1,5/3
- 90. Feedback and motivation PassiveP3 Active P3 Lugo et al., 2012
- 91. Feedback and motivation •Always give the feedback in a positive way • Goal oriented task • Lots of tricky issues, see Sellers et al., 2010
- 92. BCI Limits forhome use • Not usable by non-technical personal • Do not provide basic communication capacities (just binary: yes/no) • Not easily configured for the needs of each user • Not suited to periodic long-distance technical oversights