WiMi Hologram Cloud Inc. (NASDAQ: WIMI) (“WiMi” or the “Company”), a leading global Hologram Augmented Reality (“AR”) Technology provider, today announced that a brain-computer interface (BCI) based on EEG-fNIRS multi-modal data integration is developed to improve the performance and accuracy of EEG-fNIRS multi-modal data integration.
Multi-modal data integration has been a hot topic in the field of artificial intelligence in recent years, and its main goal is to effectively combine data or information from different sources to provide a better basis for decision making than a single data source. Electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS) are two commonly used techniques for detecting neural signals in the brain, and each of them has its own advantages and limitations.
EEG can provide high-resolution brain neural activity information, but its spatial resolution is relatively low; Although fNIRS has low temporal resolution, it can provide high spatial resolution cerebral hemodynamic information. The team of WiMi has found that combining these two technologies can compensate for their respective shortcomings and provide more comprehensive and accurate brain neural information.
WiMi utilized a binary enhancement algorithm to achieve effective integration of EEG and fNIRS data. This is a deep learning model with a self-attention mechanism that automatically learns the intrinsic correlations of the data, improving the quality and efficiency of data integration. In addition, WiMi has designed a unique algorithmic framework that can handle large-scale multi-modal data and meet the application requirements in different scenarios.
The process can be divided into the following steps:
Data collection: First, we need to collect data on the same target at the same time using both an EEG device and an fNIRS device. the EEG device will record the electrical activity of the brain, while the fNIRS device will monitor the changes in blood flow in the brain.
Data pre-processing: The collected data need to be pre-processed for EEG and fNIRS data, including filtering, denoising, and de-artifacting to improve the data quality. This usually includes steps such as filtering and normalization. In addition, due to the different temporal resolutions of the EEG and fNIRS devices, a temporal alignment operation is also required.
Feature extraction: With the combination of data, we can extract richer and more accurate features of brain neural activity. Useful features are extracted from the pre-processed data. For EEG data, features such as time domain, frequency domain, and time-frequency domain can be extracted, such as average power spectral density, time domain features (e.g., mean, variance), wavelet transform coefficients, and so on. For fNIRS data are luminous flux variations, etc.
Data integration: In the EEG-fNIRS multi-modal data integration, features are combined to obtain a comprehensive multi-modal feature representation. Multi-modal feature integration is mainly to combine the features extracted from EEG and fNIRS data to get more comprehensive and accurate information about brain activities. Through the binary enhancement algorithm, a deep learning model based on the self-attention mechanism, it can automatically learn the intrinsic correlation of data, thus realizing the effective processing of high-dimensional and complex-structured data.
Model training: Model training process, using methods such as cross-validation for model parameter selection and performance evaluation.
Application realization: Based on the extracted features, various applications are realized. For example, using these features to train machine learning models for prediction and control of brain neural activity.
This technology will provide strong technical support for research and application in the fields of brain science, neural engineering, and clinical medical care. It can help researchers understand the law of brain nerve activity more deeply, provide clinicians with more accurate diagnosis and treatment basis, and can also be applied to brain-computer interfaces, virtual reality and other high-tech fields to promote their technological progress.
For more such updates, follow us on Google News Martech News
