PROJECT 5: BCI System Based on Neural Decoding of EEG/FMRI
To develop methods for modifying pathological forms of representing information in brain areas and circuits in selected brain disorders by using fMRI-based neurofeedback and its translation into more expandable EEG-based methodologies.
There is a long-term tradition in neuroscience in using neurofeedback by measuring brain signals with electroencephalogram (EEG). The results of these investigations, although encouraging, are not conclusive or fully replicated, and there is no clear consensus about the features to be used as a basis for feedback. Part of the problem lies in the difficulty of unambiguously and accurately associating the characteristics of EEG with structures and functions of brain areas, thus using the specific pathophysiology and pathology of each targeted brain dysfunction as the critical guide.
Recently, the use of neurofeedback based on functional magnetic resonance imaging (fMRI) has emerged as an alternative. Here, the signals detected can be located precisely in the brain and contrasted with a broad knowledge about normal and abnormal brain function. There are encouraging results in this field with neurological and psychiatric diseases, but so far studies have been limited to modify the blood oxygenation level dependent signal (BOLD) of known structures or the functional connectivity between targeted areas (which only measures the associated fluctuations in net activity of regions). However, evidence from studies in a group of brain disorders indicates that the problem lies actually on abnormal representations of information, rather than differences in overall activations. Thus, it would be critical to modify these abnormal representations by using neurofeedback. This remediation is now feasible by using the multivoxel pattern analysis (MVPA) approach to fMRI, which has been proven to be more sensitive than the traditional fMRI analysis.
CNEURO accumulated several years of research in various pathologies, including dyscalculia and affective disorders, which are suitable targets for neurofeedback. Regarding these disorders, CNEURO has access to large patient databases in collaboration with the Cuban Ministry of Education and Ministry of Public Health, fully characterized and accessible for clinical research. Importantly, there is significant experience in CNEURO on the application of rapid methods for MVPA processing, and connectivity from MVPA, with several important results. These methods significantly improve execution times when processing fMRI data-MVPA, allowing their deployment in real-time situations. Finally, CNEURO has the necessary technological basis to conduct studies of neurofeedback. It is also finalizing a software technology transfer and expertise in real-time fMRI and fMRI-based neurofeedback.
Under joint study for future presentation due to the innovative nature of the technology.
SCAN Lab have already successfully developed this technology in the MRI Center at UESTC and have already completed a proof of concept study using healthy subjects demonstrating that subjects can learn voluntary control of specific regions of their brain and that this has beneficial effects on their emotional behavior. Furthermore they have shown that this learned control can continue even in the absence of further feedback.
In CNEURO, the following stages have also been completed:
- The Matlab-based massive algorithms for MVPA calculations are packed in a toolbox, which has been extensively tested.
- The final results from a MVPA study exploring the affective processing in a sample of healthy controls indicate that it is possible to separate mental representations responsible for familiarity and valence processing. This constitutes the conceptual proof validating the use of MVPA approach to monitor brain areas involved in these representations (see Figure 1).
Figure 1 Regions of interest analysis evincing areas involved in valence and familiarity processing from human faces in healthy participants.
Further steps of the project include:
- Deployment of real-time fMRI processing algorithms.
- Introduction, testing and generalization of CNEURO’s MVPA processing routines for neurofeedback.
- Joint EEG/fMRI recordings.
- Translation of fMRI-based MVPA results into EEG signals susceptible to modification by neurofeedback.
- Clinical validation.
REQUESTED KIND OF COLLABORATION:
This will involve a collaboration between UESTC scientists together with the Department of Psychiatry in the Sichuan Provincial People’s Hospital in Chengdu and Professor Rainer Goebel in the University of Maastricht who is a leading international expert in this field and has developed this neurofeedback technology. We also hope to work closely with the Chengdu-based Company Alltech, who design and manufacture MRI machines, and to work with them to incorporate this technology into the MRI machines which they supply to hospitals.
Additional areas of collaboration will include:
• Co-development and testing of new fast massive methods for analyzing fMRI-based MVPA.
• Co-development and testing of translational research in fMRI-to-EEG neurofeedback.
• Expansion to different recording platforms/clinical samples.
COMPETITIVE ADVANTAGES AND MILESTONES:
The main expected outcome of this project is to help anxiety and depression patients recover to be able to return to leading normal lives, particularly those who are unresponsive to traditional drug treatment. And extend the results to other countries.
Neurosciences Center of Cuba. Calle 190 e/ 25 y 27, CUBANACAN, Playa. La Habana. CP 11600, Phone: (+53) 7263-7100.
Key Lab for Neuroinformation. University of Electronic Sciences and Technology of China UESTC. Sichuan, Chengdu 610054, China.
Mitchell Joseph Valdes-Sosa, M.D.,
PhD. Professor, Senior Researcher and General Director, CNEURO
Cognitive Neuroscience Department & General Direction
Keith M. Kendrick, Ph.D.
Head of Social, Cognitive and Affective Lab SCAN,
Key Lab Neuroinformation UESTC
Ben Becker, PhD
Social, Cognitive and Affective Lab
Key Lab Neuroinformation UESTC
Xia Yang, Professor
Ministry of Education Key Laboratory of Neural Information, UESTC