GFS_GCS_fMRI

This software is for the paper [1], whose abstract is as follows.

Since functional Magnetic Resonance Imaging (fMRI) signals are a group of sparse signals, and its autocorrelation matrix contains limited information, it is difficult to accurately locate the brain activation area directly using traditional blind separation algorithms. For the issue, this paper proposes a method with inverse-sparse transform and second order blind identification (SOBI) for the separation of the activations. The contribution of this paper is to achieve the separation of sparse brain map signals and have lower computational complexity than higher-order statistical BSS. In experiments, we use both simulated and measured fMRI data to evaluate our method. The experimental results show that the proposed method’s running time is only 1/30 of a higher-order statistical independent component analysis (ICA) algorithm, while its separation errors is close to ICA and less than half of a traditional SOBI algorithm.

Dependencies:

1. SimTB, address:http://trendscenter.org/trends/software/simtb/index.html
2. GIFT, address:https://trendscenter.org/trends/software/gift/index.html

Simulaed data experiment steps

1. Produce simulated data via Simtb software (1) Move files, 'run_NoMotion.m' and 'experiment_params_aod_NoMotion.m' into the folder '\simtb_v18\examples'. Note that the different between 'experiment_params_aod_NoMotion.m' and the original file, 'experiment_params_aod.m' is as follows: i) 'out_path' in lin.22 ii) 'M=3' in lin. 40 iii) 'SM_source_ID = [ 2 4 6 8 10 15 18 19 29 ]' in line. 51 iv) 'D_motion_FLAG = 0' in lin.244 v) 'saveNII_FLAG = 1' in lin. 26 (2) Run simtb.m , and then run 'run_NoMotion.m'. (3) Find the output folder and move '.nii' into created subfolders 'sub0' (4) Change 'experiment_params_aod_NoMotion.m' parameters i) 'out_path = 'D:\simtb_data_hcnr'' to 'out_path = 'D:\simtb_data_lcnr'' in lin.22 ii) 'minCNR = 5; maxCNR = 5.5;' to 'minCNR = 0.65; maxCNR = 2;' in lin. 238 (5) run 'run_NoMotion.m'. (6) Find the output folder and move '.nii' into created subfolders 'sub0'
2. Complexity analysis of four algorithms Add a folder '\algorithm' to 'set path'. (1) Time complexity analysis Find a file ‘icatb_icaAlgorithm.m’ in a folder ‘GroupICATv4.0b\icatb\ icatb_analysis_functions’, and add the following codes to lin. 112 mento = 500; TimeComplexity(data,ICA_Options,mento); Run 'gift.m', and then click 'Setup ICA annaalysis' to set up parameters and then click 'Run Analysis'. (2) Multiplicative times analysis Find a file ‘icatb_icaAlgorithm.m’ in a folder ‘GroupICATv4.0b\icatb\ icatb_analysis_functions’, and add the following codes to lin. 112 MULtimes(data,ICA_Options); Run a file, ‘icatb_runAnalysis.m’ in a folder ‘gift\GroupICATv4.0b\icatb\icatb_analysis_functions’
3. Separation error i) FastICA Find a file ‘icatb_calculateICA.m’ in a folder ‘GroupICATv4.0b\icatb\ icatb_analysis_functions’, and add the following codes to lin. 252 locerror(icasig); Run a file, ‘icatb_runAnalysis.m’ in a folder ‘gift\GroupICATv4.0b\icatb\icatb_analysis_functions’ ii) GFS Find a file ‘icatb_calculateICA.m’ in a folder ‘GroupICATv4.0b\icatb\ icatb_analysis_functions’, and add the following codes to lin. 252 locerror(icasig); Find a file ‘icatb_icaAlgorithm.m’ in a folder ‘GroupICATv4.0b\icatb\ icatb_analysis_functions’, and add the following codes to lin. 112(Comment codes in lin. 113-116) [A, W, icasig_tmp] = fsobi(data, size(data,1));
   icasig_tmp = real(icasig_tmp);
   W = real(W);
   A = real(A); Run a file, ‘icatb_runAnalysis.m’ in a folder ‘gift\GroupICATv4.0b\icatb\icatb_analysis_functions’ iii) GCS Find a file ‘icatb_calculateICA.m’ in a folder ‘gift\GroupICATv4.0b\icatb\ icatb_analysis_functions’, and add the following codes to lin. 252 locerror(icasig); Find a file ‘icatb_icaAlgorithm.m’ in a folder ‘gift\GroupICATv4.0b\icatb\ icatb_analysis_functions’, and add the following codes to lin. 112(Comment codes in lin. 113-116) [A, W, icasig_tmp] = csobi(data, size(data,1));
   icasig_tmp = real(icasig_tmp);
   W = real(W);
   A = real(A); Run a file, ‘icatb_runAnalysis.m’ in a folder ‘gift\GroupICATv4.0b\icatb\icatb_analysis_functions’ iv) SOBI Find a file ‘icatb_calculateICA.m’ in a folder ‘gift\GroupICATv4.0b\icatb\ icatb_analysis_functions’, and add the following codes to lin. 252 locerror(icasig); Find a file ‘icatb_icaAlgorithm.m’ in a folder ‘gift\GroupICATv4.0b\icatb\ icatb_analysis_functions’, and add the following codes to lin. 112(Comment codes in lin. 113-116) [A, W, icasig_tmp] = sobi(data, size(data,1));
   icasig_tmp = real(icasig_tmp);
   W = real(W);
   A = real(A); Run a file, ‘icatb_runAnalysis.m’ in a folder ‘gift\GroupICATv4.0b\icatb\icatb_analysis_functions’

Measured data experiment steps

1. The download address of measured data is https://trendscenter.org/trends/software/gift/data/example_subjects.zip

2. Time complexity analysis Find a file ‘icatb_icaAlgorithm.m’ in a folder ‘gift\GroupICATv4.0b\icatb\ icatb_analysis_functions’, and add the following codes to lin. 112 mento = 500; TimeComplexity(data,ICA_Options,mento);

3. The separation results of the four algorithms i) FastICA ii) GFS Find a file ‘icatb_icaAlgorithm.m’ in a folder ‘gift\GroupICATv4.0b\icatb\ icatb_analysis_functions’, and add the following codes to lin. 112(Comment codes in lin. 113-116) [A, W, icasig_tmp] = fsobi(data, size(data,1));
   icasig_tmp = real(icasig_tmp);
   W = real(W);
   A = real(A); iii) GCS Find a file ‘icatb_icaAlgorithm.m’ in a folder ‘gift\GroupICATv4.0b\icatb\ icatb_analysis_functions’, and add the following codes to lin. 112(Comment codes in lin. 113-116) [A, W, icasig_tmp] = csobi(data, size(data,1));
   icasig_tmp = real(icasig_tmp);
   W = real(W);
   A = real(A); iv) SOBI Find a file ‘icatb_icaAlgorithm.m’ in a folder ‘gift\GroupICATv4.0b\icatb\ icatb_analysis_functions’, and add the following codes to lin. 112(Comment codes in lin. 113-116) [A, W, icasig_tmp] = sobi(data, size(data,1));
   icasig_tmp = real(icasig_tmp);
   W = real(W);
   A = real(A);

   Run a file, ‘icatb_runAnalysis.m’ in a folder ‘gift\GroupICATv4.0b\icatb\icatb_analysis_functions’

Reference [1]H. Wu, D. Li, M. Lu, Y. Zeng. fMRI Activations via Low-complexity Second-Order Inverse-Sparse-Transform Blind Separation.Digital Signal Processing，(2021):103137. [2]E.B. Erhardt, S. Rachakonda, E.J. Bedrick, E.A. Allen, T. Adali, V.D. Calhoun. Comparison of multi‐subject ICA methods for analysis of fMRI data. Human brain mapping, 32(12) (2011): 2075-2095. [3]E.B. Erhardt, E.A. Allen, Y. Wei, T. Eichele, V.D. Calhoun. SimTB, a simulation toolbox for fMRI data under a model of spatiotemporal separability. Neuroimage, 59(4) (2012): 4160-4167. [4]D. Nuzillard, J.M. Nuzillard. Second-order blind source separation in the Fourier space of data. Signal processing, 83(3) (2003): 627-631.