Advantages
- Discrimination system based on new features found by time series analysis of brain signals.
- Unique Features that cannot be extracted by conventional Fourier transform
- Highly accurate identification of subjects’ brain activity is expected.
- This system can discriminate at high speed.
- Adaptable to various time-series signal data, not limited to brain signals, and is expected to identify all types of oscillatory phenomena.
Technology Overview & Background
Analysis of brain signals is expected to have applications in the diagnosis of diseases such as Alzheimer’s disease and epilepsy, and Brain-Machine Interface (BMI), and various methods are being researched and developed.
Brain functions are produced by the interaction and coordination of many neurons, and brain signals are an extremely complex, high-dimensional. Analyzing a high-dimensional signal as it is difficult due to the enormous amount of computation involved, and an approach that removes the dimension becomes necessary. Recently, Dynamic Mode Decomposition (DMD) has been attracting attention as a method of dimensionality reduction for high-dimensional signals. The researchers have previously developed a method for decoding brain information using DMD for ECoG signals, a type of brain signal, and reported that the accuracy of estimating motor content was improved by extracting the components characteristic of movement in the phase relationship between electrodes, which could not be extracted using FFT.
The researchers’ group improved the method of decoding brain information using DMD. They found that the information extracted by the improved method are a feature value that well representing the characteristics of brain activity. This information is expected to identify each of the multiple brain activities. While this system is a useful brain activity identification system, it can be applied to various time-series signals for improving the accuracy of signal analyses.
Patents
A patent application has been submitted but is not yet published.
Principal Investigator & Academic Institution
Prof. Takufumi Yanagisawa MD-Ph.D. (Institute of Advanced Co-Creation studies and a neurosurgeon in Department of Neurosurgery, Osaka University)
Development Stages & Plans
- Brain activity was identified by this system using brain signals from healthy subjects and from Alzheimer’s patients. The identification accuracy was about 80%.
- Brain activity was identified by this system using brain signals from a healthy subject and from an epilepsy patient. The discrimination accuracy was about 95%.
- Adaptation to time series data other than brain signals will be verified in the future.
- There is another system that estimates the results of amyloid PET test from the subject’s brain signals, although it is a different method from this system.
We are looking for companies that are interested in utilizing this system and in the practical application of disease diagnostic technology using this system. If you are interested, we would be pleased to provide you with additional information or to arrange a meeting with the researcher as the next step.
Project No. TT-04731