2015 Fiscal Year Research-status Report
Learning to control brain activity pattern using real-time functional MRI: A feasibility study
| Project/Area Number |
26350993
|
|---|
| Research Institution | Nagoya University |
|---|
Principal Investigator |
Bagarinao E. 名古屋大学, 脳とこころの研究センター, 特任准教授 (00443218)
|
|---|
| Project Period (FY) |
2014-04-01 – 2017-03-31
|
| Keywords | real-time functional MRI / neurofeedback / support vector machine / brain machine interface / machine learning |
|---|
| Outline of Annual Research Achievements |
Overall, the research went as plan. The second year of the project was devoted to performing the actual experiments in which participants were trained to control their brain activity pattern using real-time functional MRI neurofeedback. To this end, a support vector machine (SVM) was used to classify on-going brain activation patterns. A correct match of the participant's brain activity to the target pattern would move a displayed arrow to the right (representing the feedback signal). Thus, the better the participant was in reproducing the correct activation pattern, the further the arrow would move.
Before the end of the last academic year (AY2015), 30 participants were recruited and scanned. Initial analysis of the acquired datasets from 17 participants was also carried out. The initial results showed that trained SVMs attained classification accuracy greater than 90% using data from the training scans. On the other hand, real-time classification performance during feedback scans varied depending on the task used: 81% for imagined finger tapping task, 88% for word generation task, and 85% for serial subtraction task. Overall, this high classification accuracy enabled participants to successfully control the movement of the arrow used as feedback in the study.
Finally, the results of the initial analysis were presented in domestic as well as international conferences.
|
| Current Status of Research Progress |
Current Status of Research Progress
2: Research has progressed on the whole more than it was originally planned.
Reason
The second year of the project progressed rather smoothly. We were able to recruit and scanned 30 participants in total, more than the initial target. This is in spite of the center's MRI scanner problem happening at the latter part of 2015. The result of the initial analysis was also very promising. Classification accuracy of the trained SVM was quite high (more than 80%) suggesting the feasibility of the used method.
|
| Strategy for Future Research Activity |
For this academic year (AY2016), additional scans will be conducted with a minimum target of 3 more participants. After all the scans are performed, analysis of the full dataset will be undertaken. The results of the analysis will be presented in upcoming domestic and international conferences. The main findings will also be compiled into a manuscript and will be submitted to some journals for possible publication.
|
| Causes of Carryover |
Most of the scans went smoothly that instead of the allotted 3-hour scan per session, the scans usually finished within two hours or less. This gave us more scan time for additional participants. Unfortunately, some participants were not scanned last academic year due to the problem with the Brain and Mind Research Center's MRI scanner in the latter part of 2015. The scanner's gradient coil needed to be replaced. There was also some delay in the recruitment of the project's research assistant. Because of these factors, not all of the fund were disbursed completely during the academic year.
|
| Expenditure Plan for Carryover Budget |
The remaining amount will be used to cover the expenses for the scanning of additional participants. It will also be used for the dissemination of the research output such as presentation in domestic conferences as well as publication expenses.
|
