The first task of this progress report was selection of the project. Our group decided to pursue a project on classification of REM sleep based on brain activity.
For this progress report, we focused on:
point 1: loading data into MATLab
point 2: meta-data available
point 3: plotting some of this data in the time and frequency domains
The dataset we selected contains readings from electrodes implanted in the brains of male Long Evans rats. There are several days worth of data for each of several rats, but we started with one day's worth of readings from one rat. The data we used is publicly available on DANDI at this link, and is in the Neurodata Without Borders (NWB) format.
Data was first loaded into MATLab using the publicly available matnwb library, an interface for working with NWB data. We created an NWB tree of the data and from this tree, we were able to extract different variables recorded during the trial period. We started by creating a table identifying which stage of sleep the subject was in during each period of the data collection. The first several rows of this table are shown below.
To examine neural activity, local field potential (LFP) data was examined. LFPs are electrical signals that are the principal measurements of brain activity. This LFP data demonstrating neural activity was plotted over time.
Additionally, the LFP data was converted into the frequency domain to note trends in the brain activity at different frequencies. From this we saw peaks in brain activity especially at very high and low frequencies.