Clinical Applications Workshop

Place: University of Groningen, Bernoulliborg, Nijenborgh 9, faculty room (12:30-13:00), room 253 (after 13:00), directions
Date: 5 october 2012, (12:30 reception) 13:00 - 17:00
Register: by 27 september, here

The aim of the workshop on Clinical Applications of Neuroinformatics is to bring together neuroinformatics researchers and clinicians, and see where advances in neuroinformatics (image/data analysis, databasing, large scale computing) can be put to clinical practice (diagnosis and treatment).

The workshop consists of a discussion, preceeded by a series of talks of about 35 minutes each, in which considerable time is reserved for feedback from the audience.

Detailed program:

12.30-13:00 Reception with coffee and tea in the faculty room (signposted at entrance)
13.00-13:15 Move to room BB0253, welcome by program chair Jos Roerdink
13:15-13:50 Christian F. Beckmann
Statistical Imaging Neuroscience: From unsupervised learning techniques to clinical applications
Abstract: Independent Component Analysis and related unsupervised learning techniques have been used extensively to estimate hidden source signals in functional imaging data and have proven successful in characterising patterns of resting functional connectivity. While early work concerned the characterisation of these effects at the single group level, increasingly often researchers now use such tools in order to characterise the differences between groups of subjects.
This talk will present a review of the neuroinformatics approaches most commonly applied in this rapidly advancing field, such as seed-based correlation analysis and independent component analysis, along with examples of their use at the individual subject and group analysis levels and a discussion of practical and theoretical issues. We describe the similarities and differences across these varied statistical approaches to processing resting-state functional magnetic resonance imaging signals. In particular, we will focus on the issue of biased between-group analysis and discuss the biased nature of group inferences derived both in the context of a seed-based analysis and an ICA/PCA. We will demonstrate how these approaches can be used in clinical investigations of large-scale systems.
13:50-14:15 Hilleke Hulshoff Pol
Early detection of psychosis. Can neural network analysis help?
14:15-14:40 Hugo Schnack
MRI brain image classification in health and disease: Towards a diagnostic tool?
Abstract: In the past decades, numerous magnetic resonance brain imaging studies have shown that the brains of psychiatric patients differ on average from those of healthy people. The detected brain abnormalities are, however, small, and could only be found between large groups of patients and healthy subjects. Recently, machine learning techniques have been invoked to go beyond statistical description, to make individual predictions based on MRI scans. The research project ‘structural magnetic resonance brain imaging classification in health and disease’ aims to use machine learning in MRI to separate healthy individuals from those with a psychiatric disease. In a recent study we built a structural-MRI-based schizophrenia classification model and tested its predictive capacity in an independent test sample. Using two large data sets, we confirmed the feasibility to use structural MRI for individualized prediction whether a subject is a schizophrenia patient or a healthy control, with an accuracy of 70.4%. Although scientifically interesting, the clinical use is limited: these classification models become really useful if they can predict a subject’s future status, or its current status if this cannot be determined by other means. more...
14:40-15:00 Tea break
15:00-15:35 Natasha Maurits
EEG analysis in the 21st century: clinical and research applications of neuroinformatics
Abstract: Electroencephalography (EEG) is the oldest neuroimaging technique available and since its initial application in humans by Hans Berger in the 1920s, advancements have been huge. Two important novelties of the last decades are the development of high-density EEG (using upto 256 electrodes instead of the clinically more commonly used 19 or 32 electrodes) and the simultaneous recording of EEG and functional magnetic resonance imaging (fMRI). These developments pose both technical and analytical challenges. In this presentation I will discuss three examples of how neuroinformatics can be of help to provide solutions to these challenges. I will show how an inverse problem must be solved as part of the pre-surgical work-up of a tumour patient, how network analysis helps to deal with the enormous amount of data generated by high-density EEG recordings and what type of methods are suited to deal with the complementary spatially and temporally informative data obtained during a simultaneous EEG-fMRI recording.more...
15:35-16:10 Nico Leenders
GLIMPS: diagnosis of Neurodegenerative Brain Diseases
Brain degenerative diseases are common and still badly understood. They usually commence slowly at a middle or older age and progress inexorably resulting in many fysical and cognitive disabilities destroying the patients’ lifes. At the early stages of these disease conditions it is often difficult to be sure about the diagnosis on clinical grounds alone.
Brain imaging has been proven to be of diagnostic help. This is particularly true for the functional brain imaging scans like the PET-scans using the radiotracer FDG (fluoro-deoxyglucose). The outcome measure of this method is the quantification per pixel in terms of the first step of the glycolysis (the 6-phosphate kinase). This measure reflects accurately the energy requirements of the underlying brain tissue. In the case of damage of neuronal systems in the brain by a disease process a complex adaptation of glucose use – sometimes subtle – of many brain regions is the consequence.
Using a multivariate covariance method – Scaled Subprofile Model/ Principal Component Analysis (SSM/PCA) – patterns of regional covariance can be calculated which are specifically disease related and which can be expressed for each subject separately.
Within the new GLIMPS-project of the UMCG/RUG we now have developed a number of patterns which can be tested in many patients. The analysis is done via website data entry of patient data and FDG-scan data from various centers in The Netherlands into a database (developed by TARGET/RUG). The subject scores processed centrally are reported to the clinicians and Nuclear Medicine physicians. Follow-up studies will be undertaken to compare the final clinical outcome with the early brain pattern determinations. Examples and preliminary results will be shown.
16:10-17:00 Discussion: how to effectively put neuroinformatics into clinical practice. We ask participants to help us:
  • think about clinicians whom you know might have a contribution from the "demand" side, and invite them to participate
  • formulate a question of the type "Wouldn't it be possible to ... (use neuroinformatics to achieve) ..."
  • formulate a question of the type "Couldn't neuroinformatics be applied to solve ... (a clinical use scenario) ..."
  • think about such a supply or demand question yourself and submit it to the discussion (mail to:
17:00-18:00 Drinks, served in room BB0106
Friday, October 5, 2012 - 13:00
University of Groningen, Bernoulliborg, Nijenborgh 9, 9747AG Groningen, room 253
Date status: