Travel grant reports 2011
Dirk Schubert, Donders Inst. for Brain, Cognition & Behaviour, UMC Nijmegen, The Netherlands
Gaute Einevoll, Department of Mathematical Sciences and Technology, Group of Dr., Norwegian University of Life Sciences, Aas, Norway
Aim of the lab visit:
Establishing joint projects for testing of the biophysical forward-modeling scheme for extracellular potentials on data from in vitro recordings with multielectrode arrays (MEA). Eventually the insight gained from this work will help to more correctly interpret and predict EEG signals. Such collaboration will fit into the INCF program “Multi-Scale Modeling” since calculation of mesoscopic measures of neural activity is an important feature of large-scale/multi-scale, modeling.
During this stay we had the chance to discuss in detail the different computational approaches that are currently being used at the Gaute lab and their potential applicability for modeling and interpreting our experimental data. By giving a seminar talk about “Functional connectivity in the rat somatosensory cortex - From basic physiological concepts to pathophysiology” I could introduce and discuss the biological background of the experimental data and animal model system that will provide the data for the joined projects. The bilateral exchange of biophysical and biological knowledge turned out to be very efficient and during my stay we decided on the details for two different joined projects:
1) Main project: In Aas the Torbjørn Bækø Ness will have his main PhD project on modeling and interpreting the experimental data on combined single cell stimulation and MEA-LFP recordings in acute rat brain slice preparations. In particular we will focus on the large layer Vb pyramidal neurons and, as a counterpart in terms of morphological organization, on layer IV spiny stellate cells. The respective experimental data will be produced by my group in Nijmegen. End of this year we will host Torbjørn Bækø Ness in Nijmegen for several days. That will give him the chance to participate in the experiments that proved the data he will perform his modeling on;
2) Secondary project: In order to correctly interpret the layer specific synaptic origins of stimulus evoked LFP signals, the complexness of their functional origins demands modeling the temporal and spatial interactions between all the possible sources for synaptic activity. In a side project Espen Hagen in Gaute Einevoll’s lab will model these interactions based on already existing experimental data sets from Nijmegen.
During the visit we already started to work on both projects and we could test the applicability of the different model strategies with partially very promising preliminary results.
Finally, we became aware of the complementary nature of other joined projects we have with Daniel Wójcik in Warsaw, therefore during my stay we decided to plan regular (yearly) joined meetings of our three groups, during which we can discuss our approaches more extensively.
Oriol Castejon - Dep. Matemàtica Aplicada I, Universitat Politècnica de Catalunya
To attend the training Workshop of the Mathematical Neuroscience 2011 meeting: "An Introduction to Mathematical Neuroscience" held in Edinburgh, April 2011.
Activities and Accomplishments:
This workshop was aimed at PhD students and post-docs which where attending to the Mathematical Neuroscience 2011 meeting, to provide an introduction of some of the concepts that were covered in the main conference. It consisted in four lessons, which are the following:
- From networks to normal forms: using reduced models to understand network dynamics, by Alex Roxin (Institut d'Investigacions Biomèdiques August Pi i Sunyer). Its aim was to explain how and why we can use simple models to understand the collective behaviour of large populations of neurons.
- Introduction to correlations in neural coding and dynamics, by Eric Shea-Brown (University of Washington), who gave an overview of the literature on correlated neural activity, covering both dynamical aspects of this problem and the impact on levels of encoded information.
- Techniques in neurophysiology, by John White (University of Utah). In this lecture, some of the techniques used in experimental cellular neurophysiology were introduced, emphasizing some opportunities and challenges for meaningful interaction between theorists and experimentalists.
- Introduction to neural oscillators, by Bard Ermentrout (University of Pittsburgh). Here, the notion of limit cycle was discussed, as the relevance of bifurcation theory in theoretical neuroscience (mainly the SNIC bifurcation, but also the Hopf bifurcation). At the end of the talk, a brief introduction to Phase Response Curves (PRCs) was given.
In conclusion, this Training Workshop gave me the chance to get to know some topics which are of high interest, from people that are experts on the field. I think this was very useful to extend my knowdlege in neuroscience, and that I would not have had this opportunity in my university. Therefore I am very thankful to the INCF for their support, which made it possible for me to attend to this training workshop.
Hans Ekkehard Plesser, Norwegian Univ of Life Sciences, Aas, Norway
Purpose and Accomplishments:
The goal was to meet during March 2011 to discuss with Mikael Djurfeldt and Hans Ekkehard Plesser and decide on a possible implementation of a library for the connection set algebra (libcsa) by Mikael Djurfeldt and the interaction between libcsa and NEST).
Also a workshop around these issues was arranged. The workshop programme is shown below.
Thursday 17 March:
09.00 -- Introduction to CSA (Mikael)
The purpose of this session was to ensure that everyone had a basic understanding of the logic behind the CSA from a user perspective.
10.30 -- Current CSA implementation (Mikael)
A first introduction to how CSA works behind the scenes, i.e., how connections are created based on CSA specifications.
11.30 -- Lunch
12.30 -- Introduction to MUSIC (Research School Tutorial; Jochen & Mikael)
13.30 -- Short break
14.00 -- Connecting Python-CSA and NEST: Design
We designed the interface connecting the Python-based CSA with NEST
15.00 -- Connecting Python-CSA and NEST: Prototyping
We hacked away to create a prototype
16.30 -- Connecting Python-CSA and NEST: First summary
We reviewed the day's achievements
17.30 -- Dinner (Drøbak)
Friday 18 March:
09.00 -- Connections in NEST: Datatypes and -structures (Hans Ekkehard/Jochen)
A brief presentation on the data types and structures representing connections in NEST
09.45 -- Connections in NEST: Connection routines (Hans Ekkehard)
A brief presentation of the basic connection routines in NEST
10.30 -- Coffee break
11.00 -- Connections in NEST: Topology module routines (Hans Ekkehard)
A brief presentation of the logic and implementation of the Topology module
11.45 -- Lunch
12.30 -- Connecting Python-CSA and NEST: Further hacking
We improved on our solutions from Thursday
14.30 -- Break
15.00 -- Connecting Python-CSA and NEST: Status summary
We summarized how far we have gotten
15.30 -- Planning future work
Jessica Ann Turner, PhD – Mind Research Network, Albuquerque, NM, USA
The aims of this 2 days visit were two-fold: 1- presenting the progresses made in the INCF Program on Neuroimaging Datasharing during the first Belgian Neuroinformatics meeting, 2- discuss and share insights regarding ontology development with Yann Le Franc who is currently developing an ontology in the context of the INCF MultiScale Modeling program.
Yann Le Franc, Belgian Neuroinformatics node, Theoretical Neurobiology lab, University of Antwerp, Wilrijk, Belgium
Activities and accomplishments:
1- I gave a 30 minutes lecture summarizing the different datasharing efforts for neuroimaging data.
2- I interacted with members of the Belgian community attending the meeting to promote INCF activities and encourage them to participate in the effort.
3- I discussed with Yann Le Franc regarding general aspects of ontology design and gave my insights regarding the development of the Computational Neuroscience Ontology
4- I discussed with Yann Le Franc about the existing technical solution for datasharing that could potentially be deployed in Belgium at the initiative of the Belgian node (Database infrastructure, Ontologies, Electronic Data Capture software and scientific workflows, …).
Added value for the INCF:
This visit was used to promote INCF activities in Belgium, supporting the current effort of the Belgian node to create a dynamic Belgian community involved in Neuroinformatics developments. In addition, it created an opportunity for knowledge transfer between two INCF programs.
Tahereh Toosi, Institute for Research in Fundamental Sciences (IPM), School of Cognitive Sciences, Niavaran,Tehran, IRAN
To attend the advanced course in computational neuroscience held in Beldewo, Poland from 1st to 27th August 2011.
This course was organized in four weeks with different themes in computational neuroscience.
First week: During this week, the basic yet important concepts about neural dynamics were taught by leading scientists in this field. These concepts were mainly about Hodgkin-Huxley model, phase plane analysis, cable theory, and synaptic plasticity. Also, in this week we had some tutorials about main simulation softwares in computational neuroscience, like Python, Neuron, Matlab, XPP and NEST. I was assigned a tutor to do a project during the course under his supervision.
Second week: This week was mainly about dynamics of neural networks and concepts like balanced networks, bistability in balanced networks, synchrony in spiking neural networks, models of primary visual cortex. We had also very useful tutorials on stochastic point processes and also Fokker-Planck equation.
Third week: In this week, spike train models and population coding were described in details. Also, reinforcement learning and a model of bottom-up attention were presented as some example of computational modeling in higher cognitive processes.
Forth week: During the final week, the main of courses was about computational modeling from receptor level to behavior, e.g. from retina to behavior. Also, there were course on variability in cortical activity. This course was ended with final presentation of student’s projects.
I got an overview of the computational neuroscience methods in this very tense exciting course. Also, I had the chance to meet some prominent members of computational neuroscience community. Thus, I could evaluate the ideas in my mind about my PhD thesis by talking about them to various scientists. I am sure that the simulation softwares that I learned to use during this course will help me to express my ideas in modeling during my research. Furthermore; I have done a project entitled “Effect of cortical states on responsiveness in cortical networks” with the help of my tutor.
Carlos Bernardo Perez Etchegoyen. Instituto de Fisiología, Biología molecular y Neurociencias (IFiByNe), Facultad de Ciencias Exactas y Naturales (FCEN), Universidad de Buenos Aires (UBA). Argentina.
To go to the Advanced Course in Computational Neuroscience in Bedlewo, Poland. July 30th to August 27th. I attended to the course to get an insight in the field of Theoretical neurosciences. Specifically, I was interested in the models of single neurons.
The objective of the project was to create a model of a leech motoneuron in order to reproduce its firing properties. When this motoneuron was stimulated with square current pulses, it fired with high frequency at the beginning of the pulses; then the firing frequency adapted fast and after the first second approximately it rebounded to high frequency again.
I worked on the model during the four weeks of the course. The first one was used mainly on acquiring the basic knowledge on modeling with Neuron and organizing the project. The second and the third week were used to create the model neuron, and the last week was used to create the final presentation.
The course consisted of talks on different levels of computational neurosciences: from a single neuron in the first week to complex circuit’s dynamics at the end of the course. I enjoyed mostly the first two weeks because they were very useful for the project. The overall quality of the course was very good.
Besides the talks in the morning, we had the rest of the day to work on our projects or to interact with speakers and students. This interaction was very informative, we talked to people from many parts of the world, sharing our experiences on life and sciences.
During the course I learned how to model a single neuron using the Neuron software. The help of my tutor (Margarita Zachariou) and the talks were very important to create the model. I also get a good introduction on Python language, which will be of great help in the future. The objective of the project was fulfilled and I could model the firing properties of the motoneuron.
The course was a very nice experience; I met people from many places, with whom I will stay in touch.
To attend the advanced course in computational neuroscience held in Beldewo, Poland from 1st to 27th August 2011.
ACCN is a course in computational neuroscience taught by well-known computational neuroscientists from all around the world. The aim of this course is to teach techniques and methods in computational neuroscience and familiarize the students to researches in this field. Tutorials on different software and packages such as MATLAB, Neuron, Python and XPP were another part of this course. This course also improves the collaboration among the research groups.
This course consisted of 4 weeks with a specific theme for each week. Assigning a director for each week gave a great variety and flavor to the course. A series of talks held every morning for 4 hours. First week was about Hodgkin-Huxley model, phase plane, cable theory, firing rate models and modeling approach. Second week was about cortical network activity, the large compartment modeling, models of primary visual cortex and motor cortex and stochastic point process. Third week, we learned about computational neuroscience methods, spike train models, reinforcement learning and population coding. Finally last week was about computational approaches for motor control and auditory systems.
Other than talks, each student has to work on a specific project defined previously in the application. So, afternoon and evening were dedicated to research. A series of tutorial on different software held on the afternoon of the first week.
The prominent part of this course was that there was a great opportunity to discuss the project or any other question in this field with well-known scientists in this area. We got to know recent research projects and be able to discuss and talk about them with peers or speakers.
I have learned about many existing and new methods for modeling and computational studies during the course. I was acquainted with many techniques such as cable theory, stochastic point process and modeling different part of the brain with variety of techniques and methods.
The outstanding part of the course for me was that I managed to discuss my research with leaders of several leading research groups in Neuroscience and computational modeling such as Peter Latham, Carl van Vreesijk, John Rinzel, Yifat Prut and many more. All the discussions gave me some important insights and new ideas for my PhD research project and the project I was doing there. Another good opportunity was the interaction with other students and peers. I talked about my projects and get the new and fresh ideas and references and I learned from different experiences and views. In general, this course was a good opportunity to expand the network and collaboration with other laboratories.
To participate in Advanced Courses of Computational Neuroscience (ACCN)-IBRO school, August 1 – 28, 2011, Bedlewo, Poland
Activities and Accomplishments:
This course offered a great opportunity for me to get familiar with advanced concepts and methods in different levels of computational neuroscience. I strongly impressed by the course’s atmosphere with lots of interactions between students, lecturers, directors and organizers. Each week had special flavor, different director, lectures and topic. I am sure that my own research will have a big benefit from this course. This course represented good starting point for improving my motivation toward my research of interest, computational neuroscience. This course gathered computational neuroscientists as well as professors, lectures and tutors from different countries to participate, which have proved to be highly beneficial, not only for the scientific ideas exchanges, but also in terms of learning opportunities for students and young researchers like me. It was highly support current efforts to reach international standards in basic, theoretical and applied neuroscience research.
This course provided for me the gateway toward computational neuroscience researches with their multidisplinary axes. So many questions, ideas and aspects were addressed through great symphony of lectures, tutorial sessions as well as social events which were provided. Beside that this course was a great chance for me to express myself as young neuroscientist. In addition, it provided multiple opportunities to build personal and scientific bonds with other students from all over the world. The most important feedbacks I received from the course were; 1. Gaining an opportunity to participate as a young neuroscientist and to make my own contribution to the discipline. 2. Learning what others have done before me to develop my own ideas – going beyond what are already known, exploring new topics of research. 3. Challenging the conventional wisdom as well as organizing and developing most of scientific skills which will be a key for improving my motivation toward my research of interest.
I would like to express my gratitude to the INCF for supporting my participation with partial travel grant (300 Euros).
The purpose of this travel grant was to attend the g-node „Advanced Scientific Programming in Python“ Summer School in St. Andrews, Scotland in September 2011.
The Summer School lasted 6 days. After each lecture exercises were performed, so that the new knowledge could be applied. The lectures covered different topics ranging from basic programming styles and data visualization to parallel computing and advanced methods in Numpy.
The exercises were done in groups of two under the pair-programming paradigm. This was not only an opportunity to learn from your partner, it also enhanced the social interaction. Several tutors were available for each exercise so that each problem/ question could be solved quickly.
Two afternoons and the complete last day were reserved for a team-programming project. Each team consisted of 6 people and the task was to implement a bot for a Pacman game. This was not only a good opportunity to apply the newly learned techniques, it also included some theoretical work in how to design software, how to split the work up in work packages, define milestones and most important: how to work in a group. Besides the fun part, I think this was really good experience, since especially in science software is often developed by more than one person.
During this Summer school I learned a lot about how I can improve my software by using Cython and parallel programming or fancy indexing in Numpy. I also learned how important unit-testing is, how this can be implemented and applied. Furthermore I learnt how to use “git” - a powerful tool for version control of small and middle-sized software-projects.
The pair-programming and the team-programming parts were not only lots of fun; they also showed me how the development of software can be improved if four instead of two eyes are looking at the code.
The social events in the evening helped to create a warm and nice atmosphere, also between the students and the tutors.
All together the g-node advanced scientific programming in Python summer school was a perfect
Carlos Andrés Mugruza Vassallo, The University of Dundee, UK
To participate in the FENS-IBRO Imaging Training Center in Switzerland, Aug/Sept 2011: IMAGING NEURAL FUNCTION IN ANIMALS AND HUMANS and to explore computational tools and applications.
Activities & Results:
I'm a PhD student (finishing 2nd year) at the Neuroscience and Development Group, University of Dundee, UK. To study some theoretical aspects of orienting of attention, I do EEG and fMRI recordings and, I use and make some Matlab scripts under toolboxes such as EEGLAB and SPM for my PhD work (including some Ansi C extensions). These scripts are to analyse EEG and/or fMRI data in attention and memory studies.
The primary goal of my trip was to participate in the summer course: FENS-IBRO Imaging Training Center. This summer course was composed in morning lectures and afternoon projects. One can see the morning lectures (http://www.unil.ch/ln/page81919.html). Every week was a different project done by 2 or 3 students. I am more than happy to report that this has happened better than I expected, even without strong hands-on computational features.
I am going to report 3 projects in which I was involved and I participated actively:
Week 1: From functional connectivity to effective connectivity: New methods for analysis of integration mechanisms in human brain.
Days 1-4: TRAINING. DCM in fMRI (Dr. Romina Rytzar), Synchronization in EEG (Dr. Christian Carmeli) in Knyazeva group. The project was devoted to learn methods of imaging large-scale functional/effective connectivity and their application to real human EEG data. We have explored separate EEG and fMRI in controls and Alzheimer's patients. Basically, in this training, we did DCM in fMRI (Dr. Romina Rytzar), analyse Synchronization in EEG (Dr. Christian Carmeli) and record (just a sample participant with ANT 128-channel EEG/ERP machine) and discuss recent research of the Connectivity Neuroimaging Group in Knyazeva's group.
Days 3-5: OUR COMPLEMENTARY APPROACH. We proposed to work in the a further validation of EEG synchronization analysis, since EEG synchronization maps can have different components in those different groups. Therefore we propose to use Independent Component Analysis (ICA) to show pattern difference in synchronization maps and do a validation based on t-test across control and Alzheimer groups. As results, we found a statistical difference between. This framework is going to be explored in depth by Knyazeva’s group and possibly in collaboration with myself.
Week 2: Optical imaging and fMRI of tactile sensory processing in mice
Optical imaging in mice (Dr Carl Petersen) & fMRI in rats (Dr. Nathalie Just)
Days 1-4: TRAINING. The mouse provides a useful model system for studying mechanistic aspects of brain function. Mice receive important sensory information from mystacial vibrissae. These whiskers surrounding the snout allow mice to discriminate textures and to build up spatial representations useful for object recognition and navigation. Dr. Petersen and Dr Just show us the current state of their research; they aim to measure sensory responses evoked by whisker stimulation in anesthetized mice through intrinsic optical imaging and fMRI. They are in the initial steps and the striking first goal is develop fMRI in mice
Days 2-5: OUR COMPLEMENTARY APPROACH. We received optical imaging data recorded in day 1 and we have worked in the imaging processing. We use a 2-D Gaussian filter and we were able to distinguish C1 & C2 with just 7 trials. In conclusion, there should be a further study to validate this statistically. Next step should be to find the optimal minimum number of trials to use. Thus, a good thing to do would be to run maybe 200 trials and look with how many trials you can find a reliable centroid between C1, C2 & C3, hopefully less than 10 trials and that number can be used in an simultaneous optical imaging & fMRI to propose a block design and maybe after an event related design.
Week 3: Simultaneous high resolution EEG and fMRI
Days 1-4: TRAINING. Simultaneous EEG & fMRI: epileptic focus.
Days 1-4: fMRI pre-processing (Dr. Anna Custo), EEG informed fMRI (Dr. Serge Vuillemoz, group leader) and structural informed EEG (Dr. Laurent Spinelly). In this project the recording of high-density EEG (256 channels) in the MRI was demonstrated and the correction and spatial-temporal analysis of the EEG and the convolution of EEG signals with the BOLD response were learned. We have done this analysis taking spikes of an epileptic patient. These spike are around 100 uV, which are clearly around 5 to 20 times bigger than the P3a wave of orienting response in our auditory paradigm.
Days 5-onwards: WORKING IN MY SPECIFIC PhD PROJECT. With this training, now is clear that the typical reconstruction image would not be enough to analyse simultaneous EEG&fMRI in the oddball number parity decision task we are studying in Dundee, I tested this in a current data we have in our lab and I can say, more regressors can help to predict sources of activation. Specific activities that I am following to reach this analysis are:
1) the use of general linear modelling developed by Pernet et al (2011) in EEG recordings in schizophrenic & controls in an attention paradigm, and
2) make a semiautomatic model selection of regressors to be used as predictors of EEG, the idea is to test stimuli properties to predict ERP waveforms.
An extra feature that really surprises me was in Basel, in FMI Dr. Otto Fajardo is doing analysis of tail movement in zebra fish using erosions, dilations and skeletonization in video recordings under Python.
Outcome for INCF:
Our integrated analysis can let us to analyse EEG and fMRI datasets. The idea is to improve the semi-automatic analysis of EEG data made by myself inside the team of Dr Potter. This semi-automatic toolbox can manage multiple EEG datasets and analyse with several regressors in the context of general linear modelling (GLM). We have developed this using current EEGLAB and LIMO toolboxes, and we can evaluate several regressors in specific electrodes chosen by the user with pre-configured multiple comparisons. Adding this to GLM in fMRI we can make a robust way to analyse EEG and fMRI. We hope that those above specific activities will allow us to use this for study many datasets even in different groups of participants.
I would like to thank INCF for their recognition through this travel grant. I also want to thank: Federation on European Neuroscience (FENS) for cover expenses and accommodation, and Lemanic Neuroscience in Switzerland for their kind scientific support and hard work towards make better scientists for the future.
Zbigniew Jędrzejewski-Szmek, Institute of Experimental Physics, University of Warsaw, Poland
To participate in the EuroSciPy 2011 conference and to deliver the "Advanced Python Constructs" tutorial during the Advanced Tutorial Tracks
I'm a PhD student at the Institute of Experimental Physics, University of Warsaw. I use the Python scientific stack for my PhD work,
including Python, NumPy, Scipy, and plotting tools, with some C and Fortran extensions.
The primary goal of my trip was to participate in the EuroSciPy 2011 conference (http://www.euroscipy.org/conference/euroscipy2011)
and tutorial tracks and to deliver the "Advanced Python Constructs" tutorial during the Advanced Tutorial Tracks. I'm happy to
report that this has happened as planned.
The tutorials are are divided into two parallel "tracks", introductory and advanced. The first one is intended for people who are beginners
or intermediate with Python and the scientific stack for Python. The second one is intended for experienced users of the libraries, who
mostly already know how to do things, but want to improve their efficiency, learn new tricks, etc. Because the lectures are often
delivered by the developers of the software, who share their insight, show their own configuration and habits, the advanced tutorials are
actually useful even for other developers of the same software. "Advanced Python Constructs" tutorial is slightly different, because
it doesn't concern a specific library or tool, but is intended to encourage scientists writing more complicated software for their own
needs, or even working on Python libraries for other people, to exploit new features which appear in the Python language and can be
used to simplify, shorten, and clarify Python code.
Two sets of overlapping materials were created for my tutorial:
- beamer slides
- a chapter in the Scipy Lecture Notes
Since they are both freely licensed, I think they can be useful to other people learning Python.
Oliver Weihberger, Bernstein Center Freiburg, Germany
To participate in the Berkeley Summer Course in Mining and Modeling of Neuroscience Data, July 11 – 22, 2011, Berkeley, CA, USA
This Course was the first of its kind and put together by Fritz Sommer and Jeff Teeters from the Redwood Center for Theoretical Neuroscience at UC Berkeley. Its goal was to introduce students with a mostly theoretical background to the major open questions in neuroscience and teach state-of-the-art techniques for the analysis and modeling of neural datasets. The course was structured in a series of lectures in the morning and Matlab exercises on the corresponding topic in the afternoon. Additional evening lectures by neuroscientists from different fields completed the course program.
There were three course instructors in the first week. Rob Kass introduced different models for spike train simulation. Regression models for spike train statistics were presented. The recurring theme of Bayesian inference, i.e. Bayes’ theorem, was introduced. Frederic Theunissen talked about the occurrence of spectral-temporal fields in auditory cortex. Quantitative methods for their analytical extraction and validation were presented. Sonja Grün talked about unitary event analysis and parallel spike train analysis. Different techniques for the extraction of higher order correlations in spike train data were presented.
The weekend between the two course weeks provided ample time to enjoy various activities in the Bay Area. An organized trip to different places on Sunday (Muir Woods, the Pacific, Golden Gate Bridge, San Francisco) offered a great possibility to explore them in a big group.
The second course week was held by three more course instructors. Odelia Schwartz introduced natural scene statistics. She presented her bottom-up approach to extract receptive fields from theoretical principles. Jonathan Pillow lectured about the problem of neural coding. He talked about maximum likelihood estimation, Volterra/Wiener Kernels, spike-triggered average and covariance and general linear models. Furthermore, quantitative measures from information theory were presented. Maneesh Sahani talked about dynamical systems theory. He introduced novel techniques from machine learning to extract motion patterns from neurophysiological recordings.
This course offered a great opportunity to get familiar with advanced concepts and methods for neurophysiological data analysis. I particularly liked the course’s atmosphere with lots of interactions between students and lecturers. I am sure that my own research will have a big benefit from this course. I would like to thank the organizers Fritz Sommer and Jeff Teeters as well as the tutors from the Redwood Center for making this course such a success. I’d like to thank the lecturers for very good course material and their didactic skills. Finally, I’d like to express my gratitude to the INCF for supporting my participation with a generous travel grant.
Tomislav Milekovic, Bernstein Center Freiburg, Germany
To attend the CRCNS summer course in Berkeley, California, USA, July 11th - 22th 2011. The Berkeley summer course was in mining and modelling of neuroscience data.
Aim of the CRCNS summer course in mining and modelling of neuroscience data was to teach young researchers to apply novel techniques in data modelling and data analysis to datasets recorded within their projects. My PhD project is oriented towards research and development of brain machine interface (BMI) in humans using epicortical field potential recordings. Within the project, neurophysiological recordings from human patients are collected and analyzed. Therefore I was interested in learning about the presented data analysis methods so that I could apply them in the future. In addition, these novel methods are used more and more frequently in the contemporary scientific publications. Learning the underlying principles on which these methods are based will help me understand the work of other research groups.
The summer course ran for two weeks and consisted of series of workshops, each workshop presenting a group of computational methods for analysis of neural recordings. Every workshop consisted of several lectures, given by a senior researcher that has a number of publications using the presented method or has even invented the presented method, and several examples written in MATLAB code that had to be solved by students themselves. These workshops ran every day from 9am till 5pm with several breaks.
In addition to the workshops, there was an evening session where a faculty member from the University of California would present his work. On some occasions, within the evening session, there was an interactive session with the workshop lecturers used to answer questions from the students.
During the course I managed to learn how to apply several complex data modelling methods in an efficient and tractable way. In addition, I got a good overview of the modelling MATLAB tools used to efficiently solve the problems of encoding and decoding of neural data. Most of the workshops used these tools within the practical sessions which made the sessions a good tutorial on the use of these tools. Therefore, using these methods in the future will be quite straightforward.
I also got a chance to hear some of the most prominent research in the field of neuroscience present their work. I would single out prof. Jose Carmena who presented his work on neuronal plasticity using brain machine interfaces.
I would like to express my utmost gratitude to Redwood Center for Theoretical Neuroscience for organising this symposium, especially the organizing committee consisting of Fritz Sommer and Jeff Teeters. Finally I would like to thank International Neuroinformatics Coordinating Facility (INCF) for providing me with a travel grant to attend the CRCNS summer course. Without their help I would not have been able to attend this wonderful event.
Yohei Tomita, ESPCI, ParisTech SIGMA lab, France
The goal of my visit, July 2011, is brain computer interface (BCI) experiments with near infrared spectroscopy (NIRS).
Our research collaboration between France/ESPCI and Japan/Keio University aims to develop joint EEG-NIRS-BCI which improves existing BCI performance such as BCIs with only EEG or NIRS. We have been developing EEG-BCI in Paris. In Keio University, NIRS data will be measured in the same BCI paradigm used in Paris. Measured data will be used to find optimal electrode placement of EEG and NIRS on the head, and features assembly method will be developed.
Prof. Yasue Mitsukura, Faculty of Science and Technology, Keio University, Kanagawa, Japan
Activities and Accomplishments:
1) Discussion with Dr. Hovagim BAKARDJIAN, a researcher in LABSP, RIKEN BSI. We contacted with Dr. Hovagim BAKARDJIAN during my stay. He is a specialist of BCI researches. He advised us about experimental conditions and we discussed about our research collaboration with brainstorming. His plenty of experiences will be helpful to success our research.
2) SSVEP experiment with NRIS For measurement of hemoglobin in the brain, NIRS system of FOIRE 3000 which is Shimadzu product was used thanks to help of Prof. Mitsukura. OxyHb, DeoxyHb, and total amount of hemoglobin can be calculated using three infrared rays whose wavelengths are 780, 805 and 830 nm. Sampling rate is 25 ms at the minimum. With this system, 3×3 probes are set to the frontal and the occipital area. Subjects sit on a chair with relaxing and are asked to restrict head movement. In that condition, subjects see the blank area for 1.0 min and then see each stimulus for 10.0 s in succession. The SSVEP interface has 13 commands which have different flashing frequency.
3) Toward the further development for developing simultaneous measurement, the measured data with NIRS will be analyzed to detect probe placement which can be used with EEG electrodes at the same time. In a next visit, simultaneous measurement with EEG and NIRS will be achieved.
Finally I appreciate International Neuroinformatics Coordinating Facility (INCF) to give me this opportunity. That stay is very fruitful and it will lead further development of our research.
Hovagim Bakardjian, RIKEN, Brain Science Institute, Japan
To initiate a long-term collaboration effort on creating new approaches for robust Brain-Computer Interfaces (BCI) using multi-modal recordings.
Francois Benoit Vialatte, Laboratory Sigma, ESPCI ParisTech, Paris, France
Scientific activities and Accomplishments:
This short collaboration visit, June 2011, involved:
1) A lecture on SSVEP-based Brain-Computer Interfaces using EEG at ESPCI ParisTech;
2) Daily joint programming sessions at the host laboratory in order to create a new improved SSVEP-BCI with 14 independent commands and very high reliability using EEG and NIRS simultaneously. The work and discussions put the foundations for (A) the data recording interfaces (EEG and NIRS), (B) the SSVEP stimulation module, (C) and the online analysis module.
3) In addition, preliminary discussions were held on another important new long-term project at a large hospital in Paris. Protocols were discussed for recording high-density EEG data for Alzheimer’s disease patients at various stages of dementia and creating a large-scale database.
Immediately after the visit, the BCI joint collaboration effort has proceeded quickly with pilot EEG and NIRS brain recordings both in Japan and in France, as well as with preparations for the assembly of the full online BCI system within 2 months, followed by improvement and experimental evaluation within 4-5 months.
I would like to thank INCF for their recognition through this travel grant. I also want to thank the heads of the hosting and sending laboratories and all involved colleagues for their scientific support.
Tomislav Milekovic, Bernstein Center Freiburg, Germany
To attend the Brain Computer Interfacing in 2011: From EEG to cortical implants, Utrecht, Netherlands, May 20th - 21th 2011
BCI in 2011 symposium gathered the most prominent researchers in the field of brain computer interface (BCI) to present their most recent accomplishments and to discuss obstacles for future development of the field. This field of research faces challenges in trying to go from fundamental research into the field of development for commercial products. Milestones on this path were defined and formation of BCI society was suggested to improve the collaboration amongst research groups in the field. I got a chance to, not only learn about the newest research results presented in the talks, but also to talk to several lab leaders. These discussions gave me the insight on the inner workings of these successful labs. I also got the chance to promote software framework made in our lab.
Apart from the series of talks, a number of posters were presented and a small exposition was made by various companies that provide commercial BCI solutions and related hardware and software. Short demonstrations were made where one could have a glimpse at the newest product lines and learn the difference to the old, previously used equipment. Such example was the demonstration of Austrian company gTec that showcased their new amplifier, a newer and better version of the amplifier used in our experiments. This gave me the chance to discuss the issues that we had using the old amplifier and to get the impression whether these problems have been removed.
During the symposium, I managed to discuss my research with leaders of several leading research groups in ECoG BCI: Jean-Philippe Lachaux, Eric Leuthardt, Nick Ramsey and Gerwin Schalk. This gave me some important insights and new ideas that I will test in the short future.
I would also like to express utmost gratitude to Rudolf Magnus Institute for Neuroscience and the BrainGain Consortium for organising this symposium. In addition I would like to give my gratitude to the local organizing committee consisting of Nick Ramsey, Erik Aarnoutse, Mariska van Steensel and Femke Nijboer.
Finally I would like to thank International Neuroinformatics Coordinating Facility (INCF) for providing me with a travel grant to attend the BCI in 2011 symposium in Utrecht. Without their help I would not have been able to attend this wonderful event.
To attend the 8th BCI2000 workshop in Utrecht, the Netherlands, May 18-19, 2011.
Activities and Accomplishments:
The BCI2000 is a complete source code used for recording neural signals and doing BCI research, and it has already been used by more than 600 laboratories around the world and can be integrated with user’s own code. Hence the BCI2000 workshop provided me a great chance to learn this general-purpose software system and build a BCI with my own signal processing algorithms and paradigms.
In this year, the workshop was organized by Rudolf Magnus Institute for Neuroscience, the University of Utrecht, and the core team of BCI2000 developer (Dr. Gerwin Schalk and Dr. Jeremy Hill etc). It brought practical tutorials of the BCI2000 system and how to perform typical BCI experiments for humans. During the two days of training, there were 4 types of modules: 1) Introduction to the BCI2000 framework given by Dr. Gerwin Schalk. He described the main components currently implemented in the software, such as the data acquisition, signal processing and feedback paradigms etc, and he also showed some demos of using the BCI2000 to perform different types of experiments; 2) Tutorials given by Dr. Jeremy Hill about how to extend BCI2000 functionality with user’s own C++ code, and also how to incorporate Matlab functions and Python scripts into the BCI2000 system in a real time scenario, following 2 hours exercises about those programming; 3) Tutorials about using two BCI2000 related toolbox, one is PRSD Studio toolbox used for designing real-time BCI decoder, and another is FieldTrip toolbox used for implementing synchronous and asynchronous processing in BCI; 4) Two on-line BCI experiments, where the participants performed one experiment using BCI2000 and motor imagery to control a cursor moving in 1-D space, and another experiment using BCI2000 and P300 evoked potentials to make spelling of letters. Therefore, these introduction, exercise and experiment brought me a detailed overview on the software and also a chance to test some real BCI applications.
In summary, the workshop was very well organized and created an ideal atmosphere for learning the knowledge of the software and implementing an advanced BCI system, and it also provided a great chance for me to meet many other BCI researchers, whom are mainly from Europe this time. Moreover, I would really like to appreciate the INCF for funding my participation in this workshop.
INCF travel grant report for:
BCI2000 workshop in Utrecht, the Netherlands, May 18-19, 2011.
BCI2000 is software framework for signal acquisition, signal processing, brain activity decoding and application development for Brain Computer Interfaces (BCI). The framework is implemented in C++ programming language and supported for Windows operating system. It has a modular structure, i.e. the framework user can make use of already developed modules by the distributers as well he/she can develop his/her own modules for signal acquisition, signal processing and BCI
applications. The workshop was mainly composed of lectures and practical training sessions for benefiting from BCI2000 both as a user and a developer. It was organized by Gerwin Schalk (Wadsworth Center, New York), Erik Aarnoutse (Rudolf Magnus Institute, UMC, Utrecht) and Cristoph Guger (g.tec, Austria).
Lectures involved an overview of BCI2000, extending the BCI2000 with C++ code, Matlab-BCI2000 interfacing, BCI2000 for virtual reality and integrating Python scripts to BCI2000. The workshop also included one exercise session about using BCI2000 as a toolbox and about integrating external C++ code into it. BCI2000 is a modular software framework, so that one can remove and add modules/filters to the existing framework and just program the module that is newly added. Flexibly, the added modules can be programmed in Python, C++ and Matlab languages. The exercises focused on extension of the framework using the C++ language. Matlab and Python tools were covered during the lectures.
There were also two practical training sessions using a real EEG system. One of these was using BCI2000 to detect mu/beta rhythms for controlling a computer cursor on the screen using the EEG signals. This was done with the already existing BCI2000 modules ,i.e. without a need to extend them. The outcome of the sessions varied across different subjects. Some of them have succeeded to move the cursor towards the targets. The other practical training was about using BCI2000 for a P300 speller. P300 signal is an evoked EEG response after the recognition of an expected visual cue. It can be used for a spelling system based on brain signals. In this session majority of the naïve subjects were able to finish the spelling of a given text in reasonable time. P300 spelling task has also been performed with the already existing modules without a need for extending the core modules.
The workshop was generally good organized and able to give a nice overview of the basic BCI2000 system. It also provided the information for starting to extend the existing modules of the framework. I am planning to test the feasibility of the unsupervised BCI learning algorithms that we develop for EEG and ECoG data. BCI2000 workshop gave me a broad idea about how I can test this feasibility using BCI2000 in shortest time. I believe that I will benefit from this framework and
this workshop towards this goal. In the long term, this would also help BCI2000 in return, by extending its repository of available algorithms to other researchers. Therefore, I appreciate INCF for providing me the opportunity to attend this workshop.
M. Giugliano, Univ. Antwerpen, Belgium; J. Simeral, Brown University, US; and M. Spira, Hebrew University, Israel
Peter ÉRDI, Budapest Computational Neuroscience Group, Research Institute for Particle and Nuclear Physics, Hungarian Academy of Sciences, Budapest, Hungary, May 2011.
Aim of the lab visit:
The goal of our visit was to informally expand prior contacts and discussions, towards a possible joint submission to a funding agency (i.e. HFSP and FP7, FET-OPEN scheme). The topic of our discussions was focused on areas of overlapping interest related to the analysis and the interpretation of electrical neural activity as detected extracellularly by multielectrode arrays. In our interactions, such a common theme of activity clearly emerged as an obvious link across the participants’ expertise in these topics: 1) preclinical brain-machine-interfaces for decoding voluntary intracortical motor activity (Dr. SIMERAL), 2) reverse engineering and biophysical modeling of collective patterned activity, occurring in in vitro models of cortical networks – detected by substrate arrays of microelectrodes (Dr. GIUGLIANO), 3) in vitro novel technologies for non-invasive extracellular “in cell” recordings and stimulation mediated by peptide-cued electrode partial pinocytosis (Dr. SPIRA), and 4) the development of methods for improved current-source-density estimation from single-neuron multichannel extracellular recordings, in cortical and hippocampal in vivo and in vitro preparations (Dr. SOMOGYVÁRI).
The achievements of our visit included an initial scientific brainstorming, toward a common agreement and the intention to participate together in the planning and writing of a future joint research grant proposal. We have been discussing two opportunities: i) the HFSP, supporting Europe-US interactions, and ii) the EC-FP7 ICT FET-OPEN scheme, supporting only European and associated countries. We concluded that applying to the next open calls of the Human Frontier Program Organization for a letter of intent, though several months apart, might be a good way to explicitly include Dr. Simeral in a funded project.
Nevertheless, in applying for the EC-FP7 ICT FET-OPEN (August or November calls) and organizing a European consortium, we might still include Dr. Simeral as a non-funded partner with an official advisor role, possibly receiving from the EC a financial contribution but limited to travel support associated with the occasion of consortium events, collaborative visits and meetings. While we certainly will consider other opportunities, Dr. Simeral could be involved initially in some preliminary exploratory work, by sharing some traces of multielectrode recordings of sorted spiking activity, from non-human primates. The specific details and agreements will be defined at a later stage and reviewed.
The FP7 EC-ICT FET-OPEN scheme perspective appears therefore the most feasible short-term opportunity for us. With this in mind, we came out with the following confidential initial brainstorming plan and with clear proposals for a consortium building.
Yury V. Zaytsev, Bernstein Center Freiburg, Germany
To attend the 3rd G-Node Winter Course in Neural Data Analysis, March 2011.
Activities and Accomplishments:
The “G-Node Winter Course in Neural Data Analysis” is a very compact course focusing on the analysis of neurophysiological data hosted by LMU
Munchen. The course consists of four modules and on each day a different kind of data along with the most established ways to analyze it are presented.
Each module consists of an introductory lecture which is designed to compliment and extend the information acquired from the preparatory reading,
a questions session and the hands-on the exercises under the supervision of the course tutors. The evenings are reserved for social events and informal
interaction between students and the faculty.
This year, the topics ranged from the assessment of plasticity and reliability of synaptic events in slice preparations and analysis of variability of spike
trains from extracellular recordings to directional tuning in the motor cortex of awake monkeys and spectral analysis of spike data and graded potentials.
Even though the course was very intense, the atmosphere in general was friendly and relaxed. I particularly enjoyed discussing the data and the
alternative approaches to analyzing it with the faculty and the tutors both during the course hours and informally in the evenings.
In spite of the fact that Matlab was an official pre-requisite for completing the exercises, I had no troubles analyzing data using Python and would like to
express my gratitude to the faculty for being open-minded towards different programming environments and acknowledge the valuable help of all of the
course tutors (and particularly Christian Garbers).
To conclude, I would highly recommend this course to PhD students and even PostDocs that are willing to boost their skills in analyzing neurophysiological data. It is a rare opportunity to get a real hands-on experience in implementing diverse data analysis techniques in such a short time span. Also, I would like to thank INCF for providing the funding that made it possible for me attend this course!
Heidi Teppola, Computational Neuroscience Lab, Department of Signal Processing, Tampere University of Technology, Finland
To attend the G-node winter course in neural data analysis, 7th of March until 11th of March 2011.
Activities and Accomplishments:
My travel started form Freiburg at 6th of March and ended 12th of March. The course covered four modules each including lecture, matlab exercises and presentation of results. First module was lectured by Prof. Clemens Boucsein. During the module we analyzed intracellular voltage-clamp data, in particular post-synaptic events. We compute the reliability and plasticity of the data. Second module was held by Prof. Sunanne Ditlevesen. In this module we learn statistical analysis methods for interval and count variability in spike trains. This particular module was really beneficial when analyzing my own spike-train data. During the third module we got good introduction to directional tuning and decoding in motor cortex held by Prof. Martin Nawrot. Last module included spectral analysis of graded potentials and spike responses. This module was guided by Dr. Jan Grewe.
I was very glad to be able to attend this particular course in neural data-analysis which gave me new ideas and knowledge how to progress with my own data and Ph.D thesis. I am very thankful for INCF supporting my participation to this course with this travel grant.
Stephan Waldert, Bernstein Center Freiburg, Germany
To attend the “FENS-IBRO-Hertie Winter School: The systems neuroscience of primate hand function: models, mechanisms, rehabilitation and mirror systems” in Obergurgl, Austria, January 2001.
Activities and Accomplishments:
The winter school brought together world leading experts (Prof. Lemon / Rizzolatti / Luppino / Schieber), experienced and young scientists in the field of neural control of hand movements. This inspiring environment together with the talks and poster presentations allowed for contacts on all levels, ranging from intense discussions with senior researchers to exchange of experiences between young scientists. For example, the school was a great opportunity to intensify my contact to the group of Prof. Roger Lemon (London) and to establish close contact to the group of Prof. Hans Scherberger (Göttingen).
The school was very well organized, the program started at 8:15am and finished around 10pm, leaving time in the night for social gathering. The program included talks, poster presentations and training of young researchers in workshops in the afternoon. Especially the rule that the first three questions directly after the talk had to come from students, only afterwards questions from senior researchers were taken, allowed for vivid discussions between the presenter, senior and young researchers.
The topics covered in this school contributed significantly to my research. It has been an ideal preparation for my research as a post doc. Topics covered many aspects of neural control of hand movements: organization and function of brain areas involved in grasping (AIP, LIP, VIP, PF, PFG, PMv, PMd, M1, MDL, ...), interactions in the networks created by these areas (e.g. dorsal and ventral stream), organization of M1, involvement of brain stem in grasping, spinal cord, biomechanics of the hand, recording techniques (intracortical and non-invasive), stimulation of brain areas in order to investigate their function in grasping (ICMS, TMS), grasping related Brain-Machine Interfaces, behavioural grasping studies and models (small and large scale) of neural hand control. These topics provided a detailed overview on the state-of-the-art knowledge and research and on yet unpublished new scientific findings.
In summary, the FENS Winter School was very well organized, significantly extended my knowledge in the field of neural control of hand movements and allowed for intense contacts, especially with European researchers (mainly UK, Italy, and France).By this report, I would like to thank the INCF for funding my participation in this school.
Attend the FENS-IBRO-Hertie Winter School 2010/11 in Obergurgl, Austria, January 9th - 16th, 2011. The theme was “The Systems Neuroscience Of Primate Hand Function: Models, Mechanisms, Rehabilitation And Mirror Systems”
Activities and Accomplishments:
I wish to express utmost gratitude to FENS, IBRO and Hertie foundations for funding and organising the Obergurgl winter school. In addition I would like to give my gratitude to Alois Saria, chairman of the local organizing committee, for organising a spotless event, Roger Lemon and Giacomo Rizzolatti, school’s scientific organizers, for selecting such an amazing list of speakers and Patricia Schane and Sigrid Habringer for helping in the organisation of the school.
Obergurgl is a small village in the Oetztal valley about a 100km from Innsbruck. It is also one of the ski-lift bottom stations for the famous Obergurgl-Hochgurgl skiing complex. University of Innsbruck Alpine Research Centre is situated in Obergurgl, and that’s where the winter school was held. Research Centre has several seminar rooms, a big lecture hall and a large number of sleeping rooms, enough to house around 200 attendees of the school and make them feel pleasant and welcome.
The Winter school started on Sunday, 9th of January, with a small reception followed by a dinner and an introduction lecture by Alois Saria, chairman of the local organizing committee, and Roger Lemon and Giacomo Rizzolatti, school’s scientific organizers. Lecture was followed by the self presentation of every student attending the school and by short introduction of all the speakers.
In addition, workgroups were presented and students were asked to join one of the school workgroups. Suggested workgroups were:
1. How are sensory inputs and motor outputs co-ordinated during skilled grasp?
2. How do sensorimotor networks interact during grasp?
3. How do sensorimotor networks interact with spinal mechanisms for skilled grasp?
4. What do mirror neurons do?
5. How can a brain-machine-interface be used to control skilled grasp?
6. What are the therapeutic possibilities for restoration of skilled grasp after spinal injury or stroke? What role will animal models play?
Idea behind the workgroup was to offer one important question connected to the topic of the winter school, and to try to offer an answer based on the lectures from the winter school, discussion with the workgroup tutors and on the offered literature. Workgroup should then give a 40 minute presentation followed by a 10 minute discussion on Saturday, last day of the winter school.
Before every scientific lecture speaker was presented by one of the students. Lectures were held in the format of 50 minute presentation followed by a 20 minute discussion. Every day consisted of 4 to 5 lectures (two in the morning, and rest in the evening slot), apart the last day when 6 workgroups had to give one lecture each. On the first two days poster sessions were held (two hours each) that gave the ample opportunity for young scientists to present and discuss their work. Posters remained in their places for the rest of the winter school, and interested students and speakers often asked the poster presenters to present the poster outside these allotted times.
During the rest of the week, workgroups would meet, often several times a day, to discuss and prepare their presentations. I chose the workgroup titled “How can a brain-machine-interface be used to control skilled grasp?” and had a lot of fresh idea through the discussion with my tutors, Hans Scherberger and Marc Maier, and my fellow students, Claudio Galletti and Stefan Schaffelhofer.
In between the scientific program, there were several recreational activities organised. Thursday featured a ski race where everyone could join and compete, while, due to overwhelming interest, snow-shoe hike was held both on the Thursday and Friday evening. Winter school attendees were also invited to ski or snowboard on the slopes of the Obergurgl-Hochgurgl ski-lift complex during every lunch break, with discounts for ski and snowboard lessons and ski and snowboard rentals. In addition, lunch was organised in the David’s hutte, an autochthon Austrian mountain hut, where one could savour the traditional Austrian meals. At the end of each day one could also socialize with the other school attendees in the open bar of the Research Centre. These evening gave rise to many friendships and useful scientific contacts, one that will hopefully help me in finding a postdoc position, maybe even in one of the labs led by one of the speakers.
I would like to single out the lecture from Andrew Jackson that gave a talk on “A systems neuroscience approach to upper-limb neural prostheses”. His talk gave a great overview of the current advances in the field of the brain-machine interface, and how close or far current cortical prostheses are from realizing skilled reaches and grasps. Since this is the research field I am working on, it was fruitful to be remained on the state of the field and presented with the several projects I didn’t know about from before.
I also liked the talk from Alexander Kraskov titled “Corticospinal mirror neurons” that presented a new method for singling out pyramidal tract neurons in the online intra-cortical recordings and a new method for establishing that one records from the identical neuron on subsequent days. In addition he, for the first time, showed evidence of the “mirror suppressed neurons”, and that these neurons project to the spinal cord, therefore arising questions on the purpose of these neurons and the projections they make.
Finally I would like to thank International Neuroinformatics Coordinating Facility (INCF) for providing me with a travel grant to attend the FENS-IBRO-Hertie Winter School in Obergurgl. Without their help I would not have been able to attend this wonderful event.
Jean-Baptiste Poline, Neurospin-I2BM-CEA, Gif-sur-Yvette Cedex, France
The main purpose of this trip. October 2010, was to integrate NiPyPE and PyXNAT and enable running NiPyPE on the Neurospin cluster. This was seen a necessary practical first step towards metadata gathering and establishing metadata standards for neuroimaging analysis.
The trip lasted 5 days. It involved a short talk, joint programming sessions and discussions with members of Neurospin about NiPyPE and PyXNAT.
During the visit, I worked with Yannick Schwartz towards better integration of NiPyPE and PyXNAT. In particular, we refined the XNAT interfaces in NiPyPE to be consistent with the file-system based data sink and source interfaces. This uniform semantics simplify user interaction with these interfaces. We were also able to create an SPM-based workflow that pulled data from the XNAT server for a number of subjects, ran the SPM preprocessing, first level analysis and surface-based group analysis and pushed the data back to the server. This was an important first step before starting to push provenance and meta-data back to the XNAT server.
Computational Neurophysics Group at the RIKEN Brain Science Institute in Wako, Saitama, Japan.
Purpose and results:
The INCF travel grant facilitated my research visit to the Computational Neurophysics Group at the RIKEN Brain Science Institute in Wako, Saitama, Japan (mainly Sept 2010). This visit is part of a collaborative project that involves members of the NEST Initiative (www.nest-initiative.org) from different countries. In this project, we develop a new design for the fundamental data structures of our neural network simulation technology NEST to enable brain-scale simulations on supercomputers such as the JUGENE Bluegene/P architecture. Preliminary results have been first presented at the 2nd INCF Congress of Neuroinformatics in September 2009.
During my research visit, we focused on the serial data structures in NEST that are key to providing access to local neurons and synapses. In the context of clusters of up to O(100) processors, a linear data structure that provides direct access by index is not just the most straightforward but also the most efficient design choice. However, when large numbers of processors are employed, such a data structure contains many empty entries, which results in substantial serial overhead. We investigated to what extent sparse data structures constitute an alternative in terms of memory usage and performance. This work enables us to identify a regime where an 'all-purpose' simulator design is no longer feasible and where model-oriented design solutions have to be considered.
I am grateful to the INCF for approving this generous travel grant and I also want to thank all members of the Computational Neurophysics and Statistical Neuroscience Group for their warm hospitality and the stimulating scientific atmosphere.
The G-Node, the center for Mind/Brain Sciences and the Fondazione Bruno Kessler
The aim to apply for the travel grant was to attend the "Advanced Scientific Proramming in Python" Autumn School, October 2010 in Trento, Italy.
The purpose of this autumn school was to introduce researchers to advanced programming techniques suitable to their aims of scientific computing, and hence saving time and energy.
This was a great opportunity for me to learn more about this programming language and its many applications within science. Not only did I want to deepen my knowledge of Python because it is a free (and therefore mobile) programming tool with a big user community worldwide. But also this course would provide new, helpful techniques and tools for Python in special and also for programming practice in general.
The autumn school lasted 5 full days. It comprised lectures and exercises about different topics as manifold as object oriented programming, best practices in programming, version control, software carpentry, scientific tools for Python, data visualization and more.
An additional challenge was a group programming competition. On the one hand it trained the coordinated teamwork and on the other hand exercised several tasks and subjects covered in the lectures. For this project a Pac man game had to be implemented. At the last day a final spiritful tournament was held to find the best team.
Furthermore several social events were organized for the evenings to allow the students to socialize outside the school and contract friendships or even collaborations.
During this autumn school I learned much more than my intended goals. I got to know advanced methods within python and helpful techniques for programming in general. But I also gained knowledge beyond pure programming. The lecture about version control introduced powerful tools to keep track and organize long-lasting projects in general, which is also helpful e.g. for writing a paper or the PhD thesis. The lecture "data visualization" did not only deal with tools to illustrate data, but also with the general rules and traps presenting data, like misleading diagrams, etc.
Furthermore the group exercises and the programming project were a salutary exercise to successfully deal with group psychology and efficient team work. During this project we had a lot of fun within our group which could also be observed for the other teams.
The social events helped to create a warm atmosphere between students as well as organizers, too. Especially the last day showed that during the whole course a team spirit and bonding between every participant was created.
All in all this autumn school was not only a successful and informative course but also a lot of fun. I enjoyed it very much and thank INCF for making this experience possible.