Background

Similar to the Rodent Sleep Scoring detector, the Rodent Sleep Scoring 2 detector uses one or two EEG signals, along with EMG and activity data to automatically assign a vigilance stage to a user defined epoch. However, this module provides greater user control over how the algorithm scores the recording for greater accuracy in scoring rodent sleep/wakefulness.

Unlike the original Rodent Sleep Scoring detector, which calculates the periodogram over the entire user defined epoch, the Rodent Sleep Scoring 2 detector calculates the periodogram over 1 second sub-epochs, assigning each a unique sleep stage. Once all 1 second epochs are scored, a sleep stage will be determined for the entire user defined epoch (e.g. 10sec or 30sec) and scored based on the stage with the majority of 1 second epochs. 

The algorithm will assess each individual second of the recording and determine the probability of Wake, Paradoxical Sleep (PS) and Slow Wave Sleep (SWS) based on the following criteria:

• Delta Power (Delta/Total)
  Used to identify SWS.
• Theta Power (Theta/Delta)
  Used to identify PS.
• EMG Power (Root Mean Square)
  Used to identify Wake.
• Contribution factors (User-defined %)
• Stage transition probabilities (User-defined %).
• Epochs with Activity >0 are scored as Active Wake.

If the EEG or EMG amplitude exceeds the artifact threshold or has any bad samples (i.e. signal dropout) the 1 second epoch is scored as Artifact.

Understanding the Detection Level Sliders

The Delta Power Levels will be used as an example to better understand the use of the Detection Level Sliders.

1. First, the algorithm will compute the Delta/Total power ratio for each individual second of the recording.
2. Then, it will calculate the average Delta/Total ratio, after discarding the lowest 10% and the highest 30%.
3. In the example below, the probability of SWS will be maximal if the Delta/Total power of a 1-second epoch is greater than 1.45 times the average that we have calculated above.

 

This same concept holds true for the Theta Power Levels and EMG Power Levels.

 

Understanding Contribution Factors

It is possible to modify the contribution factor of Delta/Total, Theta/Delta and EMG:

• Default is 100%.
• Example: If the algorithm is overestimating PS, decreasing Theta/Delta contribution factor may help.

Understanding Stage Transitions

Stage transition probabilities also come into play:

• If epoch N is Active Wake, Epoch N+1 is more likely to be Active Wake (100%) than Paradoxical Sleep (70%).
• If you see too many PS epochs during Active Wake, you may decrease the probability to transition from AW to PS from 70% to a lower value.

Automatic scoring: 

All 1 second sub-epochs are summarized to determine which sleep stage is most present within the user defined epoch to be marked. E.g., for a 10 second epoch, each 1 second scoring epoch represents 10% of the total epoch. So if there are four 1 second epochs of PS then the PS has a 40% coverage for the epoch. The total epoch will be scored based on the criteria below.

The scoring order is Active Wake > Wake > PS > SWS.

In the screenshot below (default values), the sleep epoch will be considered as:

• Artifact if 50% of the 1s sub-epochs are Artifact,
• Active Wake if 30% of the 1s sub-epochs are Active Way,
• Wake if 50% of the 1s sub-epochs are Wake,
• PS if 50% of the 1s sub-epochs are PS,
• SWS if 50% of the 1s sub-epochs are SWS.

If none of the conditions above are met, the highest number wins.

• For example if during one 10s epoch there are 4 PS, 3 SWS and 3 W, the epoch will be scored as PS.
• If two numbers are tied (for example 4 SWS, 4 PS and 2 W), then the scoring order (Wake > PS > SWS) is used again (and the epoch will be scored as PS).

Please see the NeuroScore User Manual for more information.

The following is a link to a video of how to use the Rodent Scoring 2 with Visual Tuning Tool.

https://support.datasci.com/hc/en-us/articles/232067648