Hope you are doing well. Can you please let me know the answers to the following queries:
1) Is the formula to calculate ICA activations (time series) [1-3] using the components (weights and sphere) the same for ORICA as other standard ICA techniques?
ICA activations = Weights x Sphere x Data
[1] 2015c, Tracking non-stationary EEG sources using adaptive online recursive independent component analysis.
[2] 2016, Real-time adaptive EEG source separation using online recursive independent component analysis.
[3] 2018, Online Automatic Artifact Rejection using the Real-time EEG Source-mapping Toolbox (REST).
2) If yes, should I use the final outputs of weights and sphere from ORICA and multiply them with the entire input data? OR Should I use the same equation but only for weight and sphere outputs for the samples within the variable blockSizeICA -> Repeat the same process for the entire input data for each block -> Append all the activations calculated for each block to get final activations for the entire input data?
3) Is the rule of thumb for standard ICA techniques to decide the amount of input data i.e. kN^2 [4] where k > 25, N: channels applicable to ORICA also?
[4] 2014, Online recursive independent component analysis for real-time source separation of high-density EEG.
Hi Shawn,
Hope you are doing well. Can you please let me know the answers to the following queries:
1) Is the formula to calculate ICA activations (time series) [1-3] using the components (weights and sphere) the same for ORICA as other standard ICA techniques? ICA activations = Weights x Sphere x Data
[1] 2015c, Tracking non-stationary EEG sources using adaptive online recursive independent component analysis. [2] 2016, Real-time adaptive EEG source separation using online recursive independent component analysis. [3] 2018, Online Automatic Artifact Rejection using the Real-time EEG Source-mapping Toolbox (REST).
2) If yes, should I use the final outputs of weights and sphere from ORICA and multiply them with the entire input data? OR Should I use the same equation but only for weight and sphere outputs for the samples within the variable blockSizeICA -> Repeat the same process for the entire input data for each block -> Append all the activations calculated for each block to get final activations for the entire input data?
3) Is the rule of thumb for standard ICA techniques to decide the amount of input data i.e. kN^2 [4] where k > 25, N: channels applicable to ORICA also?
[4] 2014, Online recursive independent component analysis for real-time source separation of high-density EEG.