Cybathlon Experiences Of The Graz BCI Racing Team Mirage91 .
Statthaler et al. Journal of NeuroEngineering and Rehabilitation (2017) 14:129DOI 10.1186/s12984-017-0344-9RESEARCHOpen AccessCybathlon experiences of the GrazBCI racing team Mirage91 in thebrain-computer interface disciplineKarina Statthaler1,2, Andreas Schwarz1,2, David Steyrl1,2, Reinmar Kobler1,2, Maria Katharina Höller1,2,Julia Brandstetter2, Lea Hehenberger1,2, Marvin Bigga2 and Gernot Müller-Putz1,2*AbstractBackground: In this work, we share our experiences made at the world-wide first CYBATHLON, an event organizedby the Eidgenössische Technische Hochschule Zürich (ETH Zürich), which took place in Zurich in October 2016. It isa championship for severely motor impaired people using assistive prototype devices to compete against eachother. Our team, the Graz BCI Racing Team MIRAGE91 from Graz University of Technology, participated in thediscipline “Brain-Computer Interface Race”. A brain-computer interface (BCI) is a device facilitating control ofapplications via the user’s thoughts. Prominent applications include assistive technology such as wheelchairs,neuroprostheses or communication devices. In the CYBATHLON BCI Race, pilots compete in a BCI-controlledcomputer game.Methods: We report on setting up our team, the BCI customization to our pilot including long term training andthe final BCI system. Furthermore, we describe CYBATHLON participation and analyze our CYBATHLON result.Results: We found that our pilot was compliant over the whole time and that we could significantly reduce theaverage runtime between start and finish from initially 178 s to 143 s. After the release of the final championshipspecifications with shorter track length, the average runtime converged to 120 s. We successfully participated in thequalification race at CYBATHLON 2016, but performed notably worse than during training, with a runtime of 196 s.Discussion: We speculate that shifts in the features, due to the nonstationarities in the electroencephalogram (EEG),but also arousal are possible reasons for the unexpected result. Potential counteracting measures are discussed.Conclusions: The CYBATHLON 2016 was a great opportunity for our student team. We consolidated our theoreticalknowledge and turned it into practice, allowing our pilot to play a computer game. However, further research isrequired to make BCI technology invariant to non-task related changes of the EEG.Keywords: Brain-computer Interface (BCI), Electroencephalogram (EEG), Stroke, CYBATHLON, Mental imageryBackgroundIn October 2016, a novel event called CYBATHLON, organized by the Eidgenössische Technische HochschuleZürich (ETH Zürich), took place in Zurich, Switzerland,for the first time [1]. The vision of this event is toprovide a platform for pilots with severe motor impairments to compete against each other with the support of* Correspondence: gernot.mueller@tugraz.at1Institute of Neural Engineering, Laboratory of Brain-Computer Interfaces,Graz University of Technology, 8010 Graz, Austria2Graz BCI Racing Team MIRAGE91, Graz University of Technology, Graz,Austriatechnical assistive systems and to drive forward theirdevelopment [2].The competition is composed of six different disciplines,according to the respective type of assistive system thepilots are using. The disciplines are: Functional ElectricalStimulation Bike Race, Powered Arm Prosthesis Race,Powered Leg Prosthesis Race, Powered Exoskeleton Race,Powered Wheelchair Race, and Brain-Computer Interface(BCI) Race. The races are designed to test the ability ofpilots to navigate through a series of everyday tasks withinminimal time. Details can be found on the CYBATHLONhomepage [1]. The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication o/1.0/) applies to the data made available in this article, unless otherwise stated.
Statthaler et al. Journal of NeuroEngineering and Rehabilitation (2017) 14:129Besides the pilot, the supporting team of caregiversand engineers is a key factor in a successful participationin any of the disciplines. The competition between pilotsis thus, by extension, a competition between teams. TheGraz BCI Lab formed a team named “MIRAGE91” tocompete in the BCI Race discipline [3, 4].A BCI is a device that enables users to interact withtheir environment by intentionally modulating theirbrain activity [5]. The non-invasive Graz-BCI focuses onthe changes of oscillatory components in electroencephalography (EEG) signals due to different mental tasks,like motor imagery or mental arithmetic [6, 7]. It translates the changes into computer commands to controlan application. Potential BCI-related applications includespelling devices [8] painting [9] or even music composing [10]. Furthermore, control scenarios like upper armmotor neuroprosthesis [11–14] or wheelchair control[15, 16] are investigated. In the case of the BCI Race, theapplication is a computer game. The game “BrainRunners” was specifically developed for the CYBATHLONcompetition and provided to the teams in advance toenable them to efficiently prepare for the race. The pilotcontrols an avatar in a race against up to three competitors. The avatar continuously moves forward along astraight race track. The race track itself consists of apseudorandom sequence of pads, i.e. three different action pads and one rest pad. The avatar receives a speedboost on action pads if the pilot sends the rightcommand with regard to the field, but is slowed downwhenever a wrong command is triggered. On rest pads,there is no correct command, but the avatar is sloweddown with any command. Therefore, in the optimalcase, the pilot is able to control four different commandsreliably (no command and 3 action commands) [1].This paper aims at sharing the experiences of the GrazBCI Racing Team MIRAGE91 gathered at the CYBATHLON 2016. We describe the preparations, starting fromhow we formed the team and found our pilot, to ourmulti-stage training procedure to individualize and adaptthe BCI technology to our pilot, and the final BCI technology setup in chapter 2. We report on the practicalknowledge we have gained at the event itself in chapter3, and finally, we discuss organizational challenges, thepositive public awareness, future plans and close withlessons learned in chapter 4.PreparationsMIRAGE91 - the Graz BCI racing team - familiarizingstudents with BCI researchSince the BCI field [17, 18], is very interdisciplinary, itrequires knowledge and expertise from many areas suchas neurophysiology, anatomy, psychology, neuroscience,computer science, biomedical engineering, electronics,software engineering, machine learning, statistics et cetera.Page 2 of 16Bringing students into the field usually involves disproportional effort, not only for the educator but also forstudents themselves. One of our strategies to introducestudents into BCI early on is to offer classes at masterlevel in several study programs. Additionally, the BCI Labof the Graz University of Technology has founded theGraz BCI Racing Team.During courses in our study programs Informationand Computer Engineering and Biomedical Engineering,we announced the idea of establishing a team to participate in the BCI Race and asked for interested students.In October 2014, we started with first informative meetings; we developed the idea, explained the CYBATHLON and highlighted several tasks in such a team: BCIdevelopment, creation of paradigms for training, analysisof the BCI Race game, search for potential pilots,organization of pilot training, maintenance of a website,public relations, sponsoring, and team outfit. In thisway, we were able to shape a loose group of studentsinto the Graz BCI Racing Team, named MIRAGE91(Motor Imagery Racing Graz established 1991, the yearwhen BCI research started in Graz). Our BCI RacingTeam consists of PhD, Master, and Bachelor level students of the study programs Information and ComputerEngineering, Biomedical Engineering, Computer Scienceand Mathematics. The team was announced officially bythe university and has its own website [4].As one of the first activities, we participated in theCYBATHLON rehearsal in July 2015, where we wereable to familiarize ourselves with the competition handling, our BCI, and available infrastructure. This was ofspecial importance, since we needed to know how toorganize our participation in the actual championship inOctober 2016 with a severely handicapped pilot.With this project, we were able to attract students tomake their first experiences with BCI research, to workwith pilots, and to meet other young scientists in aninternational setting. Fig. 1 shows a picture of the team,taken in Zurich at the CYBATHLON 2016.Pilot recruiting and statusAfter the rehearsal, our main goal was to identify a suitable pilot for our team. We were contacted by VAMED,an Austrian global provider in the healthcare sector.They were looking for an Austrian team participating inthe CYBATHLON 2016 and they brought us in contactwith the Neurological Center in Kapfenberg (NTK),where we established first contact with our pilot oneyear before the CYBATHLON 2016.The pilot of the Graz BCI Racing Team MIRAGE91was a 37 year old male. Before he received a stroke, hehad been an active athlete. His discipline was luge racingon natural tracks. In 01/2014, he was diagnosed with anextended stroke of the brainstem and cerebellum (right
Statthaler et al. Journal of NeuroEngineering and Rehabilitation (2017) 14:129Page 3 of 16Fig. 1 The MIRAGE91 team at the CYBATHLON 2016.side) resulting from a thrombosis of the basilar veinwhich lead to an incomplete locked-in syndrome. Athospital admission, the patient was almost completelyparalyzed with little residual ability in the upper extremity. During treatment, the motor abilities have sinceincreased to a point where he is able to operate an electric wheelchair using a joystick as an assistive device.Currently, though severely speech impaired, he is vigilant and fully aware of his environment.TrainingReliable BCI control is a complex mission, not only forpilots, but also from a technical point of view. Althoughthere have been first attempts towards plug and playBCIs, we decided to closely tailor a BCI to our pilotmanually [19]. Tailoring a BCI includes the technicalperspective, but also other aspects, like customizing theset of mental tasks, and is referred to as user centereddesign [20–22].Based on findings in previous studies [23–26] as wellas our own experiences, we came up with a four stepplan [27] to guide our pilot towards achieving reliablemulti-class BCI control (see Fig. 2).In the first step, we started with a pre-screeningsession to evaluate whether the pilot candidate is able totrigger discriminable sensorimotor rhythm (SMR) basedbrain patterns on demand. We were also interested inthe pilot’s ability to concentrate and to understand ourinstructions. This step was a milestone for both the pilotand the MIRAGE91 Racing Team, to decide whethercontinued effort and training was reasonable.Studies from Friedrich et al. [25] and Müller-Putz etal. [23] indicate that there is a large number of mentaltasks which induce changes in oscillatory EEG components. These changes can be utilized to discriminatedifferent mental tasks. However, their findings suggestthat discrimination performance varies between taskcombinations and individual users. As a second step inour tailoring process, we conducted a screening of eightdifferent mental tasks for our pilot to find sets of fourtasks with distinct patterns. Ultimately, the pilot chosethe most comfortable 4-task combination out of the bestperforming sets.In step three, we put our findings to the test in anonline BCI system. For the first time, the pilot receivedfeedback according to his mental actions. We wereprimarily curious about the performance of the chosen4-task combination, but also about the pilot’s compliance to feedback.In the fourth step, we used the information gatheredin the previous steps to optimize the BCI system forour pilot, including modern machine-learning methods[23–26, 28, 29], transfer of calibration trials from onesession to the next to reduce setup time, and a customized 4-task combination. This tailored setup was eventually used to perform training sessions over a periodof six months.Step 1: Pre-screeningIt was necessary to carry out a pre-screening of the pilotcandidate in order to assess his suitability for the discipline. Three points had to be clarified: (1) The pilot’sability to understand and perform the requested tasks,(2) his capability to elicit distinguishable brain patternsand (3) the effects of the performed tasks on the pilot. Itwas necessary to assure that executing the tasks did notcause harmful side effects such as spasms or discomfortfor the pilot. We performed two pre-screening sessionson two separate days.
Statthaler et al. Journal of NeuroEngineering and Rehabilitation (2017) 14:129Page 4 of 16Fig. 2 4 Stage training procedure: In pre-screening (step 1), the BCI aptitude of the pilot was evaluated. In step 2, screening, the best 4-classcombination out of a pool of mental strategies was identified. Stage 3 tested the pilot’s compliance with receiving feedback. Based on allcollected data, a closely tailored BCI was implemented. In stage 4 the pilot started training with the competition gameWe recorded EEG using a biosignal amplifier with 16active electrodes (g.tec, Austria) at a sample rate of512 Hz. A notch filter (50 Hz) was used in the recordingprocess along with a bandpass filter with cutoff frequencies of 0.1 and 100 Hz (8th order butterworth filter).EEG was recorded at the positions C3, Cz and C4. Weplaced four additional electrodes in an equidistant setup(2.5 cm) orthogonally around each position to allow forLaplacian derivations. The one remaining electrode waslocated at position AFz. Reference and ground electrodeswere placed on the right earlobe and frontally, respectively. The whole electrode setup is shown in Fig. 3.In both sessions, the standard Graz-BCI paradigm withthree classes was used [6] (see Fig. 4). At second 3, aFig. 3 Electrode setup: The 16 black-outlined electrodes were usedfor the pre-screening stage. The consecutive stages used allplotted electrodescross was displayed on the screen followed by an auditory cue at second 1 to get the pilot candidate’s attention. At second 0, a visual cue was presented for 1.25 sinstructing the candidate on the designated task. In thepre-screening, we chose abstract arrows as cues. Thepilot candidate performed the task for the next 5 s, untilthe cross vanished at second 5. Thereafter, an inter-trialbreak of 2–3 s followed to permit the pilot candidate tomove his eyes freely.In the first session, four consecutive runs wererecorded. Each run comprised 10 trials per class (TPC) inpseudo randomized order, i.e. in total, 40 TPC were performed. We focused on three different motor imagerytasks: repeated opening and closing of the (1) right and (2)left hand and (3) plantar flexion/extension of both feet.For the second session, we changed the tasks to twomotor imagery classes (right hand and both feet) and onerest class. During the rest trials, the designated pilot wasinstructed to relax and perform no mental imagery. Thistime, 50 trials per class (five runs) were recorded.We rejected artifact contaminated trials usingstatistical parameters: (1) amplitude threshold (amplitude exceeds / 100 μV), (2) abnormal jointprobability and (3) abnormal kurtosis. As thresholdfor the latter two, we used four times the standarddeviation (STD) [19, 28].We calculated time-frequency maps using 5 pointLaplacian derivations [30] for positions C3, Cz and C4.A bandpass filter between 2 and 40 Hz (Butterworth,causal, 6th order) was applied and data were cut intosegments lasting from 3 s before until 5 s after the cue.Event-related desynchronization and synchronization(ERD/S) of the designated pilot were analyzed [31] usinga reference interval of second 2 to second 1 beforethe cue. The results were tested for statistical significance with t-percentile bootstrapping at a significancelevel of alpha 0.05. Significant differences are shown incolor in Fig. 5a.
Statthaler et al. Journal of NeuroEngineering and Rehabilitation (2017) 14:129Page 5 of 16Fig. 4 Graz-BCI Paradigm: At second 3, a cross appeared on the screen, followed by an auditory cue at second 1 to get the attention of thepilot candidate. At second 0, the cue is presented, followed by a five second imagery period. Depending on the cue, the pilot performed thedesignated task for the whole imagery periodWe were also interested in how well the recordedmental tasks were discriminable against each other.Therefore, the data were bandpass-filtered between 6and 35 Hz using a 4th order zero-phase butterworth filter. To avoid overfitting, we separated trials into trainingand test data using 10 times 5 fold cross-validation. Ineach fold, we trained regularized common spatial patternsfilters (CSP) [32–34] for each possible class combinationusing data from second 1 to 4 with respect to thevisual cue. From each CSP class combination we tookthe first and last two projections (which hold themost discriminative information for the class combination) and calculated 12 logarithmic bandpower projections using a moving average filter over the last second(step size: 1 sample). In a second step, training of a shrinkage Linear Discriminant Analysis (sLDA) classifier [35]abFig. 5 Pre-screening results for session 1 (left) and 2 (right): a ERD/ERS maps calculated for right hand and both feet MI (left side). b Cross-validationaccuracy curves summarize the course of classification accuracy over the average trial (chance level calculated using an adjusted wald-interval,alpha 0.05). The confusion matrix summarizes the performance of the classifier across a session’s trials
Statthaler et al. Journal of NeuroEngineering and Rehabilitation (2017) 14:129was performed using bandpower features 2.5 s after thevisual cue. These calculated models were then applied tothe (fold-specific) test data to assess fold performance. Toevaluate the overall class performance, we also calculatedthe confusion matrix over the feedback period from second 1 to 4. A trial was marked as correct if the majority ofpredictions within the trial were correct. All trials werethus evaluated. We performed row-wise normalizationand calculated the percentage for each matrix value.Analysis of the recorded data showed that the pilotcandidate was able to generate distinguishable brainpatterns in both sessions (see Fig. 5). We had the impression that the pilot was excited and nervous duringthe first session, which we attributed to the novelty ofthe situation and his first contact with BCI technology.This perceived excitement and nervosity may be areason for the low-frequency EOG artifacts in the timefrequency maps right after presentation of the cue(second 0). Classification accuracies exceeded chancelevel in both pre-screening sessions. Chance levels werecalculated using an adjusted Wald interval with an alphaof 0.05 [36].For the first session, the maximum accuracy was52.7% approximately two seconds after cue presentation.Analysis of the confusion matrix showed that left handmotor imagery classification performance was lowest ofthe tested mental tasks. Since results from the firstsession already indicated that the pilot candidate wasable to produce distinguishable patterns, we exchangedleft hand motor imagery with a rest class. In the secondsession, the pilot candidate was mo
discipline “Brain-Computer Interface Race”. A brain-computer interface (BCI) is a device facilitating control of applications via the user’s thoughts. Prominent applications include assistive technology such as wheelchairs, neuroprostheses or communication devices. In the CYBATHLON BCI Race, pilots compete in a BCI-controlled computer game.
May 02, 2018 · D. Program Evaluation ͟The organization has provided a description of the framework for how each program will be evaluated. The framework should include all the elements below: ͟The evaluation methods are cost-effective for the organization ͟Quantitative and qualitative data is being collected (at Basics tier, data collection must have begun)
Silat is a combative art of self-defense and survival rooted from Matay archipelago. It was traced at thé early of Langkasuka Kingdom (2nd century CE) till thé reign of Melaka (Malaysia) Sultanate era (13th century). Silat has now evolved to become part of social culture and tradition with thé appearance of a fine physical and spiritual .
On an exceptional basis, Member States may request UNESCO to provide thé candidates with access to thé platform so they can complète thé form by themselves. Thèse requests must be addressed to esd rize unesco. or by 15 A ril 2021 UNESCO will provide thé nomineewith accessto thé platform via their émail address.
̶The leading indicator of employee engagement is based on the quality of the relationship between employee and supervisor Empower your managers! ̶Help them understand the impact on the organization ̶Share important changes, plan options, tasks, and deadlines ̶Provide key messages and talking points ̶Prepare them to answer employee questions
Dr. Sunita Bharatwal** Dr. Pawan Garga*** Abstract Customer satisfaction is derived from thè functionalities and values, a product or Service can provide. The current study aims to segregate thè dimensions of ordine Service quality and gather insights on its impact on web shopping. The trends of purchases have
Chính Văn.- Còn đức Thế tôn thì tuệ giác cực kỳ trong sạch 8: hiện hành bất nhị 9, đạt đến vô tướng 10, đứng vào chỗ đứng của các đức Thế tôn 11, thể hiện tính bình đẳng của các Ngài, đến chỗ không còn chướng ngại 12, giáo pháp không thể khuynh đảo, tâm thức không bị cản trở, cái được
Technical Program TU Graz and TU Wien I ESCAPE28 5 escape28.tugraz.at, escape28.conforganizer.net Welcome! The 28 th European Symposium on Computer-Aided Process Engineering (ESCAPE) takes place in Graz, Austria from Sunday, June 10 th to Wednesday, June 13 th, 2018. ESCAPE28 is jointly organized by TU Graz and TU Wien (Austria), and the Scientific
ANATOMI Adalah ilmu yang . “osteon”: tulang; “logos”: ilmu skeleton: kerangka Fungsi tulang/kerangka: - melindungi organ vital - penghasil sel darah - menyimpan/mengganti kalsium dan pospat - alat gerak pasif - perlekatan otot - memberi bentuk tubuh - menjaga atau menegakkan tubuh. Skeleton/kerangka dibagi menjadi: 1. S. axiale sesuai aksis korporis (sumbu badan): a. columna .
