Artikel
Unified control and data analysis system for real time functional Magnetic Resonance Imaging (rfMRI) experiments
Suche in Medline nach
Autoren
Veröffentlicht: | 6. September 2007 |
---|
Gliederung
Text
Introduction: rfMRI allows to optimise learning strategies, neuro-feedback or communication processes [Ref. 1],[Ref. 2],[Ref. 3]. Usually, separated software controls MRI measuring sequence, stimulus presentation, and statistical analysis rendering user communication and synchronisation difficult. In a new approach, an XML-based experiment description language (EDL) was developed that controlled in real-time the paradigm presentation, the rfMRI application, and the data analysis.
Methods: Pre-defined information about the paradigm, the measurement parameters, and the statistical analysis, were stored in a central EDL-file. For the statistical analysis, students t-test and correlation analysis were integrated into the software. The consistency of the EDL syntax and range of experimental parameters, i.e. the experimental control and data analysis, was checked using public XML tools.
Results: rfMRI measurements were performed on a 3T (SIEMENS Trio) and a 7T (SIEMENS) MR scanner. The standard BOLD-EPI-sequence, and the scanner-based image post-processing were modified to export in real-time each single 3D dataset to an external PC. After written consent and approval by the local ethic committee, seven right handed volunteers were examined performing a motor paradigm (right and left hand finger tapping: 10/10 images baseline/activation; TR=2000ms, TE= 29ms [3T] / 20ms [7T], 64x64x31 [3T] or 64x64x16 [7T]). First, 40 images ([5 baseline, 5 right tapping, 5 left tapping]*2) served to extract the localization (ROIs) of the activation in the left and right motor-cortex. In the following real-time measurement, the software performed a statistical analysis in the pre-selected ROIs while measurements were ongoing. Then, the software classified automatically and successfully which hand was moved. Results were displayed to both experimenter and volunteer in real-time.
Conclusion: Requiring about 1s for the data analysis, our approach is limited only by the hemodynamic response function. Additionally, the concept allows for an unsupervised activation-dependent adaptation of stimulus and measurement process enabling new dynamic real-time paradigms.