Embedded Files
For now this page will document what I (ELN) have learned about how iEEG data should / could be handled. I hope to clean it up to be a tutorial in how we handle it as a lab.
Problem: How do we store the data in a systematic fashion that is in line with modern views of open science?
Problem: How do we store the data in a systematic fashion that is in line with modern views of open science?
Rationale: it is increasingly important that data be stored in a highly systemized fashion that enables transparent record-keeping about what is in the dataset and to promote reuse. This is relevant within the lab and even within a project. By using a systematic data organization and storage plan, it ensures that there is no drift over time in standards and thereby reduces risk of process breakdown due to poorly organized data. Moreover, it is becoming standard that data used in a project be made broadly available and doing so requires using an established standard formatting.
Possible solutions:
NWB - Neurodata Without Borders.
Pros:
A standard that is being widely adopted for rodent electrophysiological studies.
We are likely to adopt the same for our rodent work, using it for the iEEG project will reduce the total number of formats we use.
It uses HDF5 formats at its core, which are optimized for random access of large files. This is a particularly important detail for this project.
It is supported by Matlab and Python APIs, allowing the data to be used more flexibly. This might allow us to use existing matlab pipelines.
Cons:
It is designed to handle neuronal spiking as well as LFP. We have no current plans to have spiking data. Depending upon how extensive the support for spiking data is, we may be using the wrong tool for our application. HOWEVER: it is possible we will start to acquire single unit data from humans, so this may not be bad to have available.
Uncertainties:
It isn't clear to me which aspects of the data NWB will cover. For example, reading about iEEG-BIDS suggests that NWB might just cover the storage of the functional data and that another package (e.g., iEEG-BIDS) would be required to organize the full data set, including other details like the localizer scans, details about the electrodes, stimuli used in testing, etc.
iEEG-BIDS
A standard for the organization and internal documentation of iEEG data extended from a format originally defined for the fMRI community.
I don't know what it would take to get our data into this structure. I saw that Brainstorm can export into this format. But I don't really know what it takes to get our data into Brainstorm either.
Page updated
Report abuse