Introduction of Brain Database and Information Technology Platform

The Brain Database and Information Technology Platform of Guangdong Institute of Intelligent Science and Technology ( hereinafter referred to as " Brain Data Platform " ) is a set of unified brain data resource platform established by Guangdong Institute of Intelligent Science and Technology to promote brain science research and information engineering. The platform can provide the storage, sharing, processing and analysis services of multimodal brain data such as imaging, EEG, gene and clinical data. It is one of the three supporting technology platforms of the institute. The Brain Data Platform combines the research experience of the institute's members in the fields of neuroscience, molecular biology, brain-computer interface technology, brain-like chip technology, artificial intelligence technology, and integrates cutting-edge technologies such as supercomputing clusters, artificial intelligence, and cloud computing. It provides data and computing resource services for the institute and related cooperation units, and is committed to building a full-life cycle data ecological environment centered on brain and brain-like scientific data resources. Brain Data Platform mainly includes the following functions :

1.The Brain and Brain-like Multimodal Data Repository is a cloud data system that provides users with multimodal brain science data resource management, query, and retrieval functions. It can manage data of more than 10PB and manage and share hundreds of independent data sets.  

2. The Biomedical Data Processing System is a cloud-based online analytics system that combines supercomputing, artificial intelligence, and data mining technologies to manage, compute, and analyze hundreds of thousands of genomics data for users.

3, Brain-like Science Data Processing System is a cloud data analysis system, users can analyze EEG, near infrared, brain magnetic, eye movements and other types of brain science and brain-like data online.

4, Medical Image Data Processing System is a set of image data analysis system to support cloud visualization, based on the most advanced AI chip and algorithm library, users can real-time, fast, batch processing and analysis of human brain images, medical images, rat brain, monkey brain and other animal images.

Specifically, the brain data platform provides the following services :

1.Comprehensive data repository

The brain data repository uses a warehouse-style data management model to set up a separate storage space ( data warehouse ) for each data set, set up an independent administrator queue, and perform an independent shared authorization audit. The warehouse realizes the storage function through the Web interface and the automatic data transmission tool. At the same time, it can also carry out the raw data analysis process, extract the meta information and merge the raw data record entries according to the specific modal types of the data. The comprehensive data repository supports data in the three fields of neuroscience, biology and brain-like science, including gene sequencing data, EEG data, protein structure analysis data, near-infrared data, medical imaging data as well as other modalities. In the data warehouse, users can classify and manage the research subjects and sample cases, and can comprehensively organize, query and retrieve the multimodal original data. The integrated data repository also supports complex forms of management structures such as multi-site joint research projects. Combined with the three data processing systems of the Brain Data Platform, the data warehouse integrates the original data, the processed characteristic data and the data processing method into a resource pool that can be called and queried at any time, which is a reliable basis for the whole life cycle management of data.

2.Biological data processing system

The data processing methods of biological research are constantly updated, and professional researchers need to invest a lot of time in the process of mastering new processing methods. The biological data processing system pre-configures the cutting-edge data processing method as a data processing plug-in that can be directly selected by users, so that users can get out of the process of code writing and debugging and focus on the value discovery of data resources. The biological data processing system provides efficient preprocessing schemes for GWAS genome-wide and cryo-EM electron microscopy data, as well as preprocessing tools based on deep learning algorithms such as AlphaFold.

3.Brain science data processing system

Brain-like scientific data processing system is mainly aimed at data resources for brain-like scientific research and brain-like algorithm application engineering development, including EEG, MEG, eye tracking, functional near-infrared and other data that have been proved to be used to develop brain-computer interface and train deep neural network. Brain-like science data processing system can complete denoising, filtering and so on. At the same time, the brain-like science resource library is also specially optimized for brain-like science-related work. It has been strengthened in data preprocessing function, feature data authorization sharing generated by processing, and interface docking between data resources and AI algorithms. Users can quickly call AI computing power to carry out related training and verification work.

4.Medical image processing system

The medical image processing system supports batch and automated preprocessing of medical images and microscope images, including image denoising, deformation correction, and region of interest delineation. It also provides image processing algorithms and corresponding resources based on deep learning. At the same time, sensitive data containing user information can be desensitized.