Reconstructing Brain-Wide Neural Activity from Local ECoG
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current neural recording technologies, such as electrocorticography (ECoG), are limited in their ability to non-invasively capture brain-wide neural activity due to the invasive nature of implanting electrodes, which restricts the spatial coverage and prevents simultaneous recordings from distributed cortical networks, leaving the rich information content of surface potentials underexploited.
Innovation Solution
The use of a recurrent neural network model that processes local electrophysiological recordings from graphene electrodes to virtually reconstruct brain-wide neural activity with pixel-level spatial resolution, leveraging multimodal data from simultaneous wide-field calcium imaging to decode cortex-wide brain activity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If electrodes are implanted to record neural activity, then measurement precision is improved, but device complexity and invasiveness increase
Solution Approach 1:
The patent uses wide-field calcium imaging to create a virtual copy of brain-wide neural activity that can be decoded from local ECoG recordings. This copying approach allows the system to infer global brain activity without physically implanting electrodes across the entire brain surface, thereby maintaining measurement precision through the calcium imaging proxy while avoiding the complexity of widespread electrode implantation.
Solution Approach 2:
The patent introduces a recurrent neural network decoder as an intermediary that translates local ECoG signals into brain-wide activity maps. This intermediary component enables the system to bridge the gap between limited local recordings and comprehensive brain-wide monitoring, achieving high measurement precision without direct physical contact across the entire brain surface.
2Device complexity
If electrodes are implanted on limited cortical regions, then device complexity is reduced, but spatial coverage and information content are limited
Solution Approach 1:
The patent creates a virtual representation of brain-wide neural activity by training a recurrent neural network on simultaneous ECoG and calcium imaging data. This virtual copy allows the system to recover information about distributed cortical networks from limited local recordings, preventing information loss while keeping the electrode array simple and localized.
Solution Approach 2:
The patent transforms the parameter space by using a recurrent neural network to map from the limited dimensional space of local ECoG recordings to the high-dimensional space of brain-wide activity patterns. This parameter transformation enables the system to extract maximal information content from minimal physical sensors.
3Object-affected harmful factors
If local recordings are used, then invasiveness is reduced, but spatial reach and ability to capture distributed networks is limited
Solution Approach 1:
The patent uses wide-field calcium imaging to create a comprehensive spatial map of neural activity that serves as a training target for the decoder. This copying approach allows the system to learn the relationship between local ECoG signals and global brain activity patterns, enabling minimal invasive recordings to achieve maximal spatial coverage through computational reconstruction.
Solution Approach 2:
The patent replaces the mechanical approach of physically distributing electrodes across the entire brain surface with a computational approach using recurrent neural networks. This substitution maintains spatial reach and the ability to capture distributed networks while minimizing physical invasiveness by confining electrodes to local regions.
Data Source
AI summary
Methods and devices for computationally constructing brain potentials across whole brain using electrocorticography signals recorded from a small region on the brain surface are disclosed. In some embodiments of the disclosed technology, a method includes obtaining a plurality of locally recorded surface potentials from a plurality of first cortical areas of a brain surface; and performing a virtual reconstruction of an average brain activity for individual cortical areas and a pixel-level cortex-wide brain activity for a plurality of cortical areas of the brain surface including the plurality of first cortical areas based on the plurality of locally recorded surface potentials.