Electroretinography analysis in Brainstorm?

Requests
,
1

Dear experts,

I'm a very new user interested in using Brainstorm to organize, process and analyze electroretinography (ERG) data simultaneously collected from the left and right eyes of individual mice as part of a large study. I didn't find any threads in this forum dealing with ERG. I have several questions (I am happy to share data if it makes the discussion easier):

  1. How can I correct for heartbeat artifacts? I imported raw ERG data as well as the first derivative of the data calculated in Excel, which provided a convenient and robust way to identify heartbeat events based on Events>Detect events above threshold using the pipeline editor. However, I was unable to apply any of the heartbeat SSP or ICA correction tools without generating errors. It seems most of these tools are designed for multiple ECG sensors that are built into MEG or EEG devices. I would like advice on how to achieve this for ERG data without accompanying ECG sensors, if possible.

  2. When I import data via a tab-delimited .txt file containing a column for time and a column for ERG voltage (time x channel array), I end up with two series, E01 and E02. E01 is a line at voltage = 0. Why is this?

  3. Normal mice yield a reproducible ERG signal with several negative and positive voltage peaks that rise and fall in succession after a flash of light. These peaks are understood to reflect the activities of light-sensitive photoreceptor neurons and secondary neuronal populations, followed by non-neuronal cell activities. An ultimate goal would be to reduce the time series data (currently 20,000 points) to a small number of parameters that can be used for statistical comparison of normal and mutant mouse lines. What approaches are available in Brainstorm for this purpose? A common approach used for ERG is to derive the voltage and time coordinates for each of the peaks that are elicited by the flash. However, this approach might benefit from peak fitting or smoothing to reduce variation in these coordinates due to noise and artifacts. Any recommendations to learn about the strategies used in Brainstorm for curve fitting or smoothing would be appreciated.

  4. My goals don't require the sophisticated capabilities of Brainstorm to map functional data to brain structures. Are there alternatives to Brainstorm that have the same electrophysiology data processing and analysis capabilities without the structural mapping capabilities?

Thanks for your time.

Cheers,
Mark Krebs

2

How can I correct for heartbeat artifacts? ... It seems most of these tools are designed for multiple ECG sensors that are built into MEG or EEG devices. I would like advice on how to achieve this for ERG data without accompanying ECG sensors, if possible.

SSP and ICA are based on arrays of sensors where we can identify topographies of values representing the artifact. This represents a form of spatial filtering of the signals. If you have only one or two signals, you can't apply techniques like these.
I'm sorry, I don't think Brainstorm offers anything to help you subtract the cardiac artifacts from your signals.

When I import data via a tab-delimited .txt file containing a column for time and a column for ERG voltage (time x channel array), I end up with two series, E01 and E02. E01 is a line at voltage = 0. Why is this?

It imports two time series: the time and the voltage. There is no simple text format with explicit time measures available by default in Brainstorm. But you can easily reformat these text files with a text editor, excel or openoffice.

An ultimate goal would be to reduce the time series data (currently 20,000 points) to a small number of parameters that can be used for statistical comparison of normal and mutant mouse lines. What approaches are available in Brainstorm for this purpose? A common approach used for ERG is to derive the voltage and time coordinates for each of the peaks that are elicited by the flash. However, this approach might benefit from peak fitting or smoothing to reduce variation in these coordinates due to noise and artifacts. Any recommendations to learn about the strategies used in Brainstorm for curve fitting or smoothing would be appreciated.

There are no such tools in Brainstorm. You'd need to do your analysis from separate Matlab scripts.

My goals don't require the sophisticated capabilities of Brainstorm to map functional data to brain structures. Are there alternatives to Brainstorm that have the same electrophysiology data processing and analysis capabilities without the structural mapping capabilities?

@mpompolas @Sylvain Do you know programs that would be more adapated for processing these ERG signals?

3

Hi Francois,

Thanks for your rapid and helpful responses. I'll keep looking!

Cheers,
Mark