The Neurophysiological Biomarker Toolbox (NBT)


Table of Contents

Compute Biomarkers

Once you have cleaned all the data from bad channels, transient artifacts and other physiological artifacts (with ICA), you can start computing the biomarkers. The biomarker analysis will be automatically saved in an analysis file with the same name of the original file (i.e. NBT.S0006.081215.ECR1_analysis.mat).

NBT is provided with several biomarkers, e.g.,:

You can see a complete overview of biomarkers here.

Other biomarkers can be added to the list. If you want to develop a new biomarker look at the section Implementing new NBT biomarkers.

Here we present the steps for computing biomarkers.

Open NBT GUI:

  • If you want to compute biomarkers on a specific signal you have cleaned:
    1. go to File | Load NBT Signal, select from the window that will appear RawSignalInfo. You need to select the RawSignal and not continue with the ICASignal since the ICASignal has been down-sampled for speeding up visualization. The cleaning and ICA changes you have made to the recording will be loaded automatically when running the biomarker computation, or in some cases you may be prompted to select the ICASignal at a later stage.
    2. go to Compute Biomarkers | For current NBT Signal and select the biomarker you want to compute.
  • If you want to compute biomarkers on multiple files ( so called batch mode processing ), you do not need to load the signals
    1. go to Compute Biomarkers | For multiple NBT Signals and select the biomarker you want to compute.
    2. in response to the question Which Signal do you want to use?, type the signal on which you want to calculate the biomarker, i.e. ICASignal (which should contain the re-referenced cleaned signal)
    3. select the folder where the cleaned signals are stored (we assume that you have all the data saved in the same folder)

For some biomarkers you will be asked to fill specific parameters:

  1. for the coherence and the phase locking value you will be asked to set the frequency range and the time interval you want to analyze. The frequency range you will input depends on the frequency band of interest (for example if you are interested in looking at Coherence in the classical frequency bands, you would select 1 45)
  2. for the DFA you will be asked to set frequency range, filter order, overlapping of the windows used for the segmentation of the signal, bins of the logarithmic scale used for the segmenting windows, and fitting interval (for most of these values default settings are also provided)

When you run the analysis in batch mode you do not need to insert the above mentioned parameters for each file which is processed, but only for the first. The following files in the folder will then be processed together using the same setting parameters.

For this Demoset, we computed the biomarkers: Correlation, Amplitudes and DFA in the classical frequency bands. For the DFA we used default settings for filter order, windows overlap and number of bins for logarithmic window scale. Although Fitting interval and Calculation interval are automatically set (Fitting interval from 5 sec to 10% of the signal duration; Calculation Interval from 0.1 sec to the entire signal duration ), you can still modify these parameters.

When you run a DFA analysis in batch mode make sure that each signal has similar length or use a fitting interval that can be valid for all your signals.

Once you load your cleaned signal you can check which biomarkers have already been computed on the signal exploring the menu Compute Biomarkers|List of biomarkers in current signal. A window like this will appear showing the list of computed biomarkers:

Next step: Test your cleaning

Go back to the Tutorials

tutorial/compute_biomarkers.txt · Last modified: 2017/05/06 14:40 by Mia Thomaidou
The NBTwiki platform - version 2.8 - 9 May 2013
Copyright (C) 2008-2015