Recordings of the electrical activity of the human brain are the only way to measure the brain’s activity at the speed at which it occurs.
We use the term electrophysiology for a range of related techniques including scalp electroencephalography (EEG), magnetoencephalography (MEG) and implanted electrodes and electrode arrays.
Applications span the biomedical sciences, cognitive, behavioural, social sciences, sports science, human factors, and defence.
Data standards and analytic approaches for electrophysiology are typically developed in a siloed fashion. Advanced data management standards and processing pipelines developed in some disciplines (like AI and biomedical engineering) haven’t been widely adopted in other domains (such as cognitive, behavioural and social sciences).
This presents a barrier to the advancement of Australian electrophysiology research.
The Australian Electrophysiology Data Analytics Platform (AEDAPT) project is creating an open national platform for reproducible electrophysiology data analysis and sharing, accessible to all Australian researchers conducting electrophysiological research.
The project involves the following elements.
Electrophysiology data and metadata standards — AEDAPT supports open standards such as the Brain Imaging Data Structure (BIDS) format, which was recently extended from neuroimaging to electrophysiology. AEDAPT will include integrated tools to convert from proprietary electrophysiology formats to BIDS and manipulate, interrogate and share data with FAIR principles.
Electrophysiology and Crossmodal Container Library — An online library of containerised advanced electrophysiology analysis software, along with domain-specific and general machine learning libraries used in the field. Crucially, integration with AIS, BrainLife, DataLad and the BIDS standard will allow for the containerisation of crossmodal pipelines and tools providing: high-resolution electrophysiology modelling; integration of diffusion MRI measures of structural connectivity with electrophysiology measures of functional connectivity; and visualisation with state-of-the-art multimodal visualisation tools.
AEDAPT Core — The core is Neurodesk: a user-friendly desktop environment for electrophysiology and related behavioural and psychophysiological data analysis and processing that enables low-latency interactive analysis on user PCs, or deployment on HPC and other platforms. The modular architecture simplifies usage and the setup drastically lowers the barrier for entry for less-technical users by providing a desktop user interface to neuroscience applications within a standard browser, with built-in Python and MATLAB installation and easy connectivity with the user’s chosen data storage.
Neurodesk interoperability — AEDAPT will bring the tools to the researcher. Neurodesk will be interoperable with (AIS), Characterisation Virtual Lab (CVL) and BrainLife.io. Containerisation allows the system to run on existing infrastructure, from desktops to cloud VMs to HPCs, without the need to constrain users to any single platform while providing equivalent user-experience on these platforms, saving on migration costs and student/staff training.
Training and community engagement — Ease of use and community connection are key components of the AEDAPT project. Training workshops and interactive online tutorials will support researchers to easily transition from interactive desktop computing to containerised, fully reproducible HPC analysis.
Who Will Benefit
Researchers, clinicians and trainees will benefit from the project’s core features:
- data provenance — AEDAPT will adopt data provenance features of BIDS, as well as tools such as DataLad that record changes to files throughout processing. The integration will provide a versioned record of containers utilised for every analysis stage
- interoperability — Neurodesk will be interoperable with CVL and BrainLife. As it’s not tied to a specific platform, analysis workflows developed using our architecture are easily shared with international collaborators and industrial partners
- ease of use — AEDAPT is built around making the transition from interactive desktop computing to reproducible containerised workflows and High Performance Computing easy and intuitive.
- Swinburne University of Technology
- The University of Queensland
- University of Sydney
- National Imaging Facility
- Australian Brain Alliance
- Oracle for Research
By adapting an open-source, community-driven platform that was already applied to neuroimaging at The University of Queensland, and using containers for reproducible analysis pipelines, AEDAPT will make reproducible state-of-the-art analysis tools highly accessible to researchers from universities, industry and clinical settings wherever they are, including small laboratories, remote regions and internationally.
By making AEDAPT interoperable with other analysis platforms such as AIS, CVL and BrainLife, the platform will act as a catalyst for scaling up electrophysiology and multimodal neuroimaging research to large, national and international collaborative research projects addressing major challenges such as epilepsy, stroke, traumatic brain injury and dementia.
Crucially, it will bring high-performance analysis to researchers wherever they happen to be without the need for expensive infrastructure.