OSCR

FREQ-NESS Reveals Age-Related Differences in Frequency-Resolved Brain Networks During Auditory Recognition and Resting State.

Code ↔ Paper

10 matches between paragraphs of the paper and lines of its authors' code, computed by the harvester (lexical-v1). Click a colored paragraph or line to see its counterpart.

The 10 matches · 1 of them tie a paragraph to a whole file, not to given lines: a weak match, whose lines are not tinted
  1. [1] § Methods › Source Reconstruction ↔ MEG_SR_Beam_LBPD.m, lines 1–96 · score 0.92 · leadfield model, FieldTrip, single shell, MNI152 T1, MEG channels, MEG sensors
  2. [2] § Methods › Source Reconstruction ↔ MEGSourceReconstruction_LeadFieldModel_Workshop.m, lines 28–71 · score 0.90 · leadfield model, FieldTrip, MNI152 T1, MEG channels, source reconstruction, MEG sensors
  3. [3] § Methods › Source Reconstruction ↔ MEG_SR_Beam_LBPD.m, lines 1–96 · score 0.72 · leadfield model, beamforming algorithm, MEG sensor, source reconstruction, brain source, weights
  4. [4] § Methods › Source Reconstruction ↔ MEGSourceReconstruction_LeadFieldModel_Workshop.m, lines 28–71 · score 0.69 · leadfield model, source reconstruction, MEG sensor, brain source, orientations
  5. [5] § Methods › Data Acquisition ↔ MEG_sensors_MCS_plottingclusters_LBPD_D.m, lines 58–128 · score 0.65 · channel Elekta Neuromag, TRIUX, Position, MEG
  6. [6] § Methods › Frequency Analysis › Morlet Wavelet Transform ↔ MEG_sensors_MonteCarlosim_LBPD_D.m, lines 1–69 · score 0.63 · Monte Carlo, original clusters, binarizing, permutation, threshold, computational
  7. [7] § Methods › Frequency Analysis › Morlet Wavelet Transform ↔ oneD_MCS_LBPD_D.m, the whole file · a weak match · score 0.63 · Monte Carlo, original clusters, binarizing, permutation, threshold
  8. [8] § Methods › Preprocessing ↔ MEG_SR_Beam_LBPD.m, lines 288–357 · score 0.63 · planar gradiometers, FieldTrip, head, segmented, SPM, position
  9. [9] § Results › Overview on Experimental Design and Analysis ↔ FREQNESS_Toolbox/FREQNESS_Functions/FREQNESS_NetworkEstimation.m, lines 281–367 · score 0.53 · narrowband covariance matrix, spatial activation patterns, eigenvectors, broadband, eigenvalues, filter
  10. [10] § Methods › Generalized Eigenvector Decomposition ↔ FREQNESS_Toolbox/FREQNESS_Functions/FREQNESS_NetworkEstimation.m, lines 1–91 · score 0.53 · source reconstructed brain, multichannel, covariance matrices, eigenvalue, vector, voxels

Paper

Loaded from Europe PMC by your browser, not stored by OSCR: doi.org · Europe PMC

The paper is loaded when this pane is shown.

The authors' code

MATLAB · 910 lines · 53 KB · GPL-3.0 · 3 matches

  1. function [ OUT ] = MEG_SR_Beam_LBPD( S )
  3. % Function to run the several steps required for source reconstruction
  4. % using a beamforming algorithm.
  5. % The algorithms used here are based on work of other brilliant people and
  6. % I do not want to take any credit for that. This function was created simply
  7. % to handle several steps more easily, make small different choices, try various solutions,
  8. % and also use such algorithms in connection to the cluster of computers of Aarhus University
  9. % (for this particular purpose, set the correspondent label below).
  10. % Furthermore, writing these codes allowed me to deeply study the beamforming as well as to prepare
  11. % materials for teaching purposes. Specifically, this function is based on
  12. % a combination of codes and functions from FieldTrip, OSL and SPM
  13. % (beamforming toolbox), plus some in-house-built solutions.
  14. % Thus, to run this function you should first download OSL (https://github.com/OHBA-analysis/osl-core),
  15. % which comprises also the required functions from FieldTrip and SPM. Then,
  16. % you should also start OSL up so that all functions paths are known by Matlab.
  17. % If you work with Aarhus University (and in some cases Oxford University)
  18. % facilities, I would suggest you to contact me to set everything up.. best way could be using the set of functions named: LBPD.
  19. % Leonardo Bonetti, [email hidden]
  21. % The function handles the following steps:
  22. % I) NORMALIZING MEG SENSORS, MAINLY FOR COVARIANCE ESTIMATION (OPTIONAL BUT STRONGLY RECOMMENDED)
  23. % II) GRID (i), LEADFIELD MODEL (ii), MEG SENSORS DATA EXTRACTION (iii), SPATIAL FILTERS COMPUTATION (iv)
  24. % III) FULL INVERSION
  25. % IV) SPATIAL AND/OR TEMPORAL SMOOTHING (OPTIONAL) (TO BE UPDATED SINCE I HAVE TO CHECK WHETHER THE NEW MASK IS COMPATIBLE WITH THE ORIGINAL ONE IN TERMS OF MNI COORDINATES, ETC.)
  26. % V) CREATION OF NIFTI IMAGES (OPTIONAL)
  31. % INPUT: -S.norm_megsensors (I)
  32. % S.norm_megsensors.zscorel_cov: 1 = z-score normalization; 0 = no normalization;
  33. % (SUGGESTED 1!)
  34. % S.norm_megsensors.workdir: working directory (where computation will be made and output stored; highly recommended
  35. % to have a different workdir per subject.
  36. % S.norm_megsensors.MEGdata_e: path to MEG data (epoched)
  37. % S.norm_megsensors.freq: frequency range in Hertz to be used for the source reconstruction (e.g. [2 8]); leave empty [] to not apply any filter
  38. % S.norm_megsensors.MEGdata_c: path to MEG data (continuous).
  39. % This must be passed only if you ask more filtering (so if S.norm_megsensors.freq is not empty).
  40. % This is done since you should do the filtering on the continuous (and not on the epoched) data.
  41. % S.norm_megsensors.forward: forward solution for leadfield; at the moment should be: 'Single Shell'
  43. % -S.beamfilters (II)
  44. % S.beamfilters.sensl: 1 = magnetometers; 2 = gradiometers; 3 = both MEG sensors (mag and grad) (SUGGESTED 3!)
  45. % S.beamfilters.maskfname: path to brain mask: (e.g. 8mm MNI152-T1: '/projects/MINDLAB2017_MEG-LearningBach/scripts/Leonardo_FunctionsPhD/External/MNI152_T1_8mm_brain.nii.gz')
  47. % -S.inversion (III)
  48. % S.inversion.znorml: 1 for inverting MEG data using the zscored normalized data.
  49. % 0 to normalize the original data with respect to maximum and minimum of the experimental conditions if you have both magnetometers and gradiometers.
  50. % 0 to use original data in the inversion if you have only mag or grad (while e.g. you may have used zscored-data for covariance'matrix)
  51. % (SUGGESTED 0 IN BOTH CASES!)
  52. % (PLEASE, NOTE THAT THE COVARIANCE MATRIX IS ALWAYS CALCULATED BEFORE WITH THE OPTION PROVIDED WITH: "S.norm_megsensors.zscorel_cov")
  53. % S.inversion.timef: data-points to be extracted (e.g. 1:300); leave it empty [] for working on the full length of the epoch (both for covariance matrix and for ERF data to be multiplied by weights W)
  54. % S.inversion.conditions: cell with characters for the labels of the experimental conditions (e.g. {'Old_Correct','New_Correct'})
  55. % S.inversion.bc: specify the time-samples to be used for baseline correction (as usual, subtraction of the mean of the baseline from the whole timeseries; independently done for each brain source).
  56. % e.g. [1 15] means first 15 time-samples (e.g. 100ms if you have sampling rate = 150Hz).
  57. % Leave empty [] for non calculating it.
  58. % S.inversion.abs: 1 for absolute values of brain sources values; 0 otherwise.
  59. % Please, note that the sign of the reconstructed signal is not really meaningful or at least it may be normalized after this function if you have knowledge of the polarity of the signal you are reconstructing.. THIS LAST STATEMENT MUST BE CHECKED WITH SOME TESTS!! IT SEEMED FINE IN OAT, BUT I'M NOT SURE YET HERE..
  60. % (SUGGESTED 1!)
  62. % S.inversion.effects: Defining solution for computing main effects and contrasts:
  63. % -Main effects of experimental conditions:
  64. % 1 = simple mean over trials
  65. % 2 = mean divided by standard deviation
  66. % 3 = mean divided by square root of standard deviation
  67. % 4 = mean divided by (standard deviation divided by square root of number of trials) (actual t-value..)
  68. % 5 = single trials (i.e. all trials independently)
  69. % 6 = custom data (matrix: MEG channels x time-points x (new) conditions); OBS!! This requires "S.inversion.alternative_data"
  70. % -Contrasts between two experimental conditions:
  71. % 1 = simple difference of means over trials
  72. % 2-3-4 = two-sample t-tests,
  73. % S.inversion.alternative_data: cell array: {1} = new data (double matrix: MEG channels x time-points x (new) conditions)
  74. % {2} = new conditions label (cell array with characters)
  76. % -S.smoothing (IV)
  77. % S.smoothing.spatsmootl: 1 for spatial smoothing; 0 otherwise
  78. % S.smoothing.spat_fwhm: spatial smoothing fwhm (suggested = 100)
  79. % S.smoothing.tempsmootl: 1 for temporal smoothing; 0 otherwise
  80. % S.smoothing.temp_param: temporal smoothing parameter (suggested = 0.01)
  81. % S.smoothing.tempplot: vector with sources indices to be plotted (original vs temporally smoothed timeseries; e.g. [1 2030 3269]).
  82. % Leave empty [] for not having any plot.
  84. % -S.nifti (V): 1 for plotting nifti images of the reconstructed sources of the experimental conditions
  85. % Please, note that no statistics is computed; that will be done by other functions/scripts.
  87. % -S.Aarhus_cluster: 1 for using the cluster of computers (Haydes) of Aarhus (CFIN-MIB).
  88. % Please, note that to use this you need to have access to the Aarhus' facilities.
  89. % Please, contact me, Leonardo Bonetti ([email hidden]), for further information.
  90. % -S.out_name: name (character) for saving output results and nifti images (conditions name is automatically detected and added) on disk
  93. % OUTPUT: -OUT: struct (both outputted and saved in S.norm_megsensors.workdir) with:
  94. % -MEG sensor data (ERFs)
  95. % -source reconstructed data (data before and after smoothing is stored)
  96. % -S structure with information used for the computations
  103. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
  106. % [email hidden]
  107. % Leonardo Bonetti, Aarhus, DK, 19/02/2021
  110. %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
  118. if S.Aarhus_cluster == 1
  119. %starting LBPD up (and therefore also OSL and a subset of SPM and FieldTrip functions)
  120. pathl = '/projects/MINDLAB2017_MEG-LearningBach/scripts/Leonardo_FunctionsPhD'; %path to stored functions
  121. addpath(pathl);
  122. LBPD_startup_D(pathl);
  123. end
  125. %preparing output structure
  126. OUT = [];
  127. OUT.S = S;
  128. %checking input for baseline correction
  129. if ~isempty(S.inversion.bc) && length(S.inversion.bc) ~= 2
  130. error('S.inversion.bc (baseline correction) must be either empty [] or comprise two values (e.g. [1 15])')
  131. end
  134. %%%%% I - MEG SENSORS NORMALIZATION %%%%%
  135. disp('%%%%% I - MEG SENSORS NORMALIZATION %%%%%')
  136. %loading SPM object (D)
  137. workdir = S.norm_megsensors.workdir;
  138. if ~exist(workdir,'dir') %creating working folder if it does not exist
  139. mkdir(workdir)
  140. end
  141. %filtering, if requested
  142. if ~isempty(S.norm_megsensors.freq) %if you have provided a frequency range
  143. if ~isfield(S.norm_megsensors,'MEGdata_c') || isempty(S.norm_megsensors.MEGdata_c) %checking if you have provided continuous data for filtering purposes
  144. error('if you want to apply the bandpass filter you need to provide continuous data (in S.norm_megsensors.MEGdata_c)')
  145. end
  146. disp('filtering MEG sensor data')
  148. %%% original (for some reasons, in some cases the bandpass filter did not work and I had to use a low pass filter which in this particular case was also fine
  149. S3 = [];
  150. S3.D = S.norm_megsensors.MEGdata_c; %continuous data
  151. S3.freq = S.norm_megsensors.freq; %frequency range
  152. S3.band = 'bandpass';
  153. % S.prefix = 'f_LBPD';
  154. D = spm_eeg_filter(S3); %actual function
  155. %%%
  157. %%% temporary
  158. % S3 = [];
  159. % S3.D = S.norm_megsensors.MEGdata_c; %continuous data
  160. % S3.freq = [3.5]; %frequency range
  161. % S3.band = 'low';
  162. % % S.prefix = 'f_LBPD';
  163. % D = spm_eeg_filter(S3); %actual function
  164. %%%
  167. disp('epoching MEG filtered sensor data with epochinfo from your previously epoched data')
  168. D_e = spm_eeg_load(S.norm_megsensors.MEGdata_e); %loading previously computed epoched data to get epochinfo (info about trials specification, labels, etc.)
  169. epochinfo = D_e.epochinfo; %getting the epochinfo information
  171. %%% TEMPORARY!! for communications biology..
  172. %1000ms
  173. % epochinfo.trl(:,1) = epochinfo.trl(:,1) - 135;
  174. % epochinfo.trl(:,3) = ones(length(epochinfo.trl),1).*(-150);
  175. %500ms
  176. % epochinfo.trl(:,1) = epochinfo.trl(:,1) - 60;
  177. % epochinfo.trl(:,3) = ones(length(epochinfo.trl),1).*(-75);
  178. %2000ms
  179. % epochinfo.trl(:,1) = epochinfo.trl(:,1) - 285;
  180. % epochinfo.trl(:,3) = ones(length(epochinfo.trl),1).*(-300);
  181. %%%
  183. %building structure for spm_eeg_epochs
  184. S3 = [];
  185. S3.D = D; %filtered continuous data
  186. S3.trl = epochinfo.trl; %trial information from epochinfo
  187. S3.conditionlabels = epochinfo.conditionlabels; %condition labels
  188. S3.prefix = 'e_LBPD'; %prefix to clearly identify that I am doing a further epoching
  189. D = spm_eeg_epochs(S3); %actual function
  190. %store the epochinfo structure inside the D object
  191. D.epochinfo = epochinfo;
  192. %switch the montage to 1 (ICA denoised data; before was 0 (non-ICA-denoised data) since OSL people did it.. however should not really matter.. what is important in my understanding is that here it is switched back to 1)
  193. D = D.montage('switch',1);
  194. D.save(); %saving on disk
  195. Dfilte = D; %copying file for potential later usage when extracting MEG sensors for inversion
  196. clear D_e
  197. else
  198. D = spm_eeg_load(S.norm_megsensors.MEGdata_e); %otherwise loading non-filtered (presumably broad band) data
  199. end
  201. %cloning SPM object
  202. [workdiroriginal, fname, ext] = fileparts(D.fullfile); %getting information about the path of the data
  203. if S.norm_megsensors.zscorel_cov == 1
  204. Dnew = D.clone( fullfile(workdir,['Clone' fname ext]) );
  205. idxMEGc1 = find(~cellfun(@isempty,strfind(D.chanlabels,'MEG'))); %elaborated way to get indices of MEG channels (even though I am mainly interested in the first MEG channel index)
  206. % dat = D(idxMEGc1,:,:); %extracting only MEG channels
  207. % for jj = 1:size(dat,3) %over trials
  208. % dat(:,:,jj) = zscore(dat(:,:,jj),[],2);
  209. % disp(['normalizing trial ' num2str(jj) ' / ' num2str(size(dat,3))])
  210. % end
  211. % dat = zeros(length(idxMEGc1),size(D,2),size(D,3));
  212. for jj = 1:size(D,3) %over trials
  213. Dnew(idxMEGc1,:,jj) = zscore(D(idxMEGc1,:,jj),[],2);
  214. % dat(:,:,jj) = zscore(D(idxMEGc1,:,jj),[],2);
  215. disp(['normalizing trial ' num2str(jj) ' / ' num2str(size(D,3))])
  216. end
  217. % Dnew(idxMEGc1,:,:) = dat; %storing normalized MEG channels
  218. Dnew.save();
  219. D = Dnew;
  220. %%% OBS!!! NOTE THAT AFTER CLONE, HERE WE TAKE DATA FROM AFTER
  221. %%% AFRICA (SO THE ICA-DENOISED DATA) BUT WE HAVE TO STORE IT IN
  222. %%% THE MONTAGE = 0 OF THE CLONED SPM OBJECT.. SO IN THE CLONE
  223. %%% MONTAGE = 0 SHOULD CONTAIN THE ICA-DENOISED DATA AFTER ZSCORE
  224. %%% NORMALIZATION..
  225. % else %otherwise just passing the data
  226. % Dnew(:,:,:) = D(:,:,:);
  227. % Dnew.save();
  228. end
  229. % clear D Dnew
  230. % if S.norm_megsensors.zscorel_cov == 1 %this is just to make sure to properly work on normalized or non-normalized MEG data
  231. % D = spm_eeg_load([workdir '/Clone' fname ext]);
  232. % else
  233. if S.norm_megsensors.zscorel_cov ~= 1 %this is just to make sure to properly work on normalized or non-normalized MEG data
  234. if exist('Dfilte','var')
  235. D = Dfilte; %getting filtered file, if you previously computed it
  236. else %otherwise original epoched data provided by the user
  237. D = spm_eeg_load(S.norm_megsensors.MEGdata_e);
  238. end
  239. end
  240. S2 = [];
  241. S2.forward_meg = S.norm_megsensors.forward; %- Specify forward model for MEG data ('Single Shell' or 'MEG Local Spheres')
  242. D = osl_forward_model(D,S2);
  243. D.save();
  244. %previous specifications from SPM beamforming toolbox
  245. space = 'MNI-aligned';
  246. val = 1;
  247. gradsource = 'inv';
  248. %retrieving data from SPM object
  249. BF.data = spm_eeg_inv_get_vol_sens(D, val, space, gradsource);
  250. idxMEGc1 = find(~cellfun(@isempty,strfind(D.chanlabels,'MEG'))); %elaborated way to get indices of MEG channels (even though I am mainly interested in the first MEG channel index)
  251. BF.data.MEG.sens.label = D.chanlabels(idxMEGc1)'; %to prevent possible discrepancies between MEG channel labels..
  252. %emulating SPM beamforming toolbox
  253. BF.data.D = D;
  254. BF.data.mesh = D.inv{val}.mesh;
  255. BF.data.zscorel_cov = S.norm_megsensors.zscorel_cov; %storing information about normalization of MEG data
  256. %1 BF file for each subject
  257. save([workdir '/BF_' fname '.mat'],'BF');
  261. %%%%% II - GRID (i), LEADFIELD MODEL (ii), MEG SENSORS DATA EXTRACTION (iii), SPATIAL FILTERS COMPUTATION (iv) %%%%%
  262. disp('%%%%% II - GRID (i), LEADFIELD MODEL (ii), MEG SENSORS DATA EXTRACTION (iii), SPATIAL FILTERS COMPUTATION (iv) %%%%%')
  263. sensl = S.beamfilters.sensl; %getting information about MEG sensors to be used
  264. % timef = S.beamfilters.timef; %getting time-points to be used
  266. %%% 1) COMPUTING GRID (8-MM MNI152 STANDARD MASK TRANSFORMED INTO SPACE WHERE WE DO THE BEAMFORMING)
  267. disp('computing grid')
  268. %loading prepared data (preprocessed, epoched, coregistered (rhino), (maybe normalized), forward model done (vol in ft))
  269. load([workdir '/BF_' fname '.mat']);
  270. [ mni_coords, ~ ] = osl_mnimask2mnicoords(S.beamfilters.maskfname);
  271. %emulating SPM beamforming toolbox
  272. S2 = [];
  273. S2.pos = mni_coords;
  274. % transform MNI coords in MNI space into space where we are doing the beamforming
  275. M = inv(BF.data.transforms.toMNI);
  276. gridv.pos = S2.pos;
  277. %actual transformation of MNI coordinates into the space where we are doing the beamforming
  278. res = ft_transform_geometry(M, gridv);
  279. % establish index of nearest bilateral grid point (for potential use in lateral beamformer); NOT REALLY USED HERE..
  280. res.bilateral_index = zeros(size(S2.pos,1),1);
  281. for jj = 1:size(S2.pos,1)
  282. mnic = S2.pos(jj,:);
  283. mnic(1) = -mnic(1);
  284. res.bilateral_index(jj) = nearest_vec(S2.pos,mnic);
  285. end;
  286. BF.sources.mni = res;
  288. %%% 2) COMPUTING LEADFIELD MODEL
  289. disp('computing leadfield model')
  290. %simulating SPM beamforming toolbox
  291. BF.sources.pos = BF.sources.mni.pos;
  292. BF.sources.ori = [];
  293. nvert = size(BF.sources.pos, 1); %number of dipoles
  294. modalities = {'MEG', 'EEG'};
  295. reduce_rank = 2;
  296. %getting some information
  297. chanind = indchantype(BF.data.D, {'MEG', 'MEGPLANAR'}, 'GOOD'); %channels index
  298. % chanunits = units(BF.data.D, chanind); %channels unit
  299. %preparing data for computation of lead field model
  300. [vol, sens] = ft_prepare_vol_sens(BF.data.(modalities{1}).vol, BF.data.(modalities{1}).sens, 'channel', chanlabels(BF.data.D, chanind));
  301. %getting locations of dipoles (derived from original MNI coordinates from standard space mask)
  302. pos = BF.sources.pos;
  303. %plotting together grid in the brain, vertices and triangles from segmentation and MEG channels
  304. %here I think I am plotting the native space with the dipoles derived from standard mask in MNI coordinates transformed into native space
  305. figure
  306. ft_plot_vol(vol, 'edgecolor', [0 0 0], 'facealpha', 0);
  307. hold on
  308. plot3(pos(:, 1), pos(:, 2), pos(:, 3), '.r', 'MarkerSize', 10);
  309. hold on
  310. plot3(BF.data.MEG.sens.chanpos(:, 1), BF.data.MEG.sens.chanpos(:, 2), BF.data.MEG.sens.chanpos(:, 3), '.b', 'MarkerSize', 10);
  311. rotate3d on;
  312. axis off
  313. axis vis3d
  314. axis equal
  315. %getting label indices (the order of the triplets is fine since the
  316. %leadfield function reads it from sens.label and now I am also reading that
  317. %from sens.label for extracting proper channels (e.g. magnetometers only)
  318. %from leadfield computation
  319. cntmag = 0; cntgrad2 = 0; cntgrad3 = 0; IDXMAG = zeros(102,1); IDXGRAD2 = zeros(102,1); IDXGRAD3 = zeros(102,1);
  320. for gg = 1:length(sens.label) %over MEG channels
  321. if strcmp(sens.label{gg}(end),'1') %if it is a magnetometer
  322. cntmag = cntmag + 1;
  323. IDXMAG(cntmag) = gg; %storing index
  324. elseif strcmp(sens.label{gg}(end),'2') %same concept for gradiometers (first of the couple)
  325. cntgrad2 = cntgrad2 + 1;
  326. IDXGRAD2(cntgrad2) = gg;
  327. elseif strcmp(sens.label{gg}(end),'3') %second of the couple
  328. cntgrad3 = cntgrad3 + 1;
  329. IDXGRAD3(cntgrad3) = gg;
  330. end
  331. end
  332. IDXGRAD = sort(cat(1,IDXGRAD2,IDXGRAD3)); %combining indices of planar gradiometers (according to progressive order in sens.label)
  333. %preallocating space for lead field
  334. L = cell(1,nvert);
  335. Ltot = cell(1,nvert);
  336. %actual computation of lead field
  337. for ii = 1:nvert
  338. dumL = ft_compute_leadfield(BF.sources.pos(ii, :), sens, vol, 'reducerank', reduce_rank(1)); %computation of the leadfield using FieldTrip function
  339. if sensl == 1 %magnetometers
  340. %previous
  341. % L{ii} = dumL(1:3:306,:);
  342. %new
  343. L{ii} = dumL(IDXMAG,:);
  344. elseif sensl == 2 %gradiometers
  345. %previous
  346. % dumdumL = zeros(204,size(dumL,2));
  347. % dumdumL(1:2:204,:) = dumL(2:3:306,:);
  348. % dumdumL(2:2:204,:) = dumL(3:3:306,:);
  349. % L{ii} = dumdumL;
  350. L{ii} = dumL(IDXGRAD,:);
  351. else %all MEG sensors
  352. L{ii} = dumL;
  353. end
  354. Ltot{ii} = dumL; %storing full leadfield for later plotting computations
  355. % L{i} = ft_compute_leadfield(BF.sources.pos(ii, :), sens, vol, 'reducerank', reduce_rank(1),'normalize', 'yes'); %trying to use the normalize option to reduce the bias towards the center of the head
  356. disp(['computing leadfield model.. source: ' num2str(ii) ' / ' num2str(nvert)])
  357. end
  361. %%%%%% CONSIDER TO REMOVE THIS IF EVERYTHING WORKS FINE
  362. % %%% 3) EXTRACTING MEG SENSORS DATA
  363. % disp('extracting MEG sensors data')
  364. % %loading and extracting data (you need to decide whether the normalization should be used for all processes of source reconstruction or only for computing the covariance matrix)
  365. % %z-scored normalized data (if you previously asked for normalization, otherwise simply clone of the original MEG data)
  366. % Dc = spm_eeg_load([workdir '/Clone' fname ext]);
  367. % if ~exist('timef','var')
  368. % warning('timef was not specified, so working on full length of the epoch.. you may want to check whether you need a subset of the time-window or the full epoch..')
  369. % timef = 1:size(D,2);
  370. % end
  371. % %extracting data (only MEG sensors and first timef time-points)
  372. % idxMEGc1 = find(strcmp(Dc.chanlabels,'MEG0111')); %getting extreme channel index (first MEG channel, then I add 305 to get them all)
  373. % chans = Dc.chanlabels(idxMEGc1:idxMEGc1+305); %extracting labels for MEG channels
  374. % if sensl == 1 %magnetometers
  375. % datameg = Dc(idxMEGc1:3:idxMEGc1+305,timef,:); %data
  376. % elseif sensl == 2 %gradiometers
  377. % datameg = zeros(204,length(timef),size(Dc,3));
  378. % datameg(1:2:204,:,:) = Dc(idxMEGc1+1:3:idxMEGc1+305,timef,:); %data
  379. % datameg(2:2:204,:,:) = Dc(idxMEGc1+2:3:idxMEGc1+305,timef,:); %data
  380. % else
  381. % datameg = Dc(idxMEGc1:idxMEGc1+305,timef,:); %data
  382. % end
  383. % time = Dc.time(timef);
  384. % ERFmean = mean(datameg(:,:,1:80),3);
  385. % ERFmedian = median(datameg(:,:,1:80),3);
  386. %%%%%%% UNTIL HERE
  388. %RESHAPING ORDER OF MNI COORDINATES (FROM ORIGINAL STANDARD BRAIN TO BRAIN LAYOUT BUILT FOR BETTER VISUALIZATION PURPOSES, you find it here: /projects/MINDLAB2017_MEG-LearningBach/scripts/Leonardo_FunctionsPhD/External/MNI152_8mm_coord_dyi.mat)
  389. %this is important for printing the nifti image with the results to be visualized
  390. disp('reshaping order of MNI coordinates.. just for handling better some visualization purposes..')
  391. % if ~exist('MNI8','var') %if not already loaded..
  392. warning('loading MNI152-T1 8mm brain newly sorted set of coordinates.. remember that if you want a different spatial resolution (e.g. 2mm), you need to create a new mask and update this line of code!!')
  393. load('/projects/MINDLAB2017_MEG-LearningBach/scripts/Leonardo_FunctionsPhD/External/MNI152_8mm_coord_dyi.mat') %loading coordinates for the new IDs
  394. % end
  395. % if ~exist('mni_coords','var') %if not already loaded..
  396. % [ mni_coords, ~ ] = osl_mnimask2mnicoords('/scratch5/MINDLAB2017_MEG-LearningBach/MNI152_T1_8mm_brain_PROVA.nii.gz');
  397. % end
  398. MNI_sorted_index = zeros(length(mni_coords),1);
  399. for ii = 1:length(MNI8) %over coordinates (ii = coordinates from file: MNI152_8mm_coord_dyi)
  400. coorddum = MNI8(ii,:); %getting new IDs coordinates
  401. source = find((double(mni_coords(:,1)==coorddum(1)) + double(mni_coords(:,2)==coorddum(2)) + double(mni_coords(:,3)==coorddum(3))) == 3); %finding old ID coordinates (from leadfield) corresponding to new IDs coordinates (image MNI152_8mm_coord_dyi.nii.gz)
  402. MNI_sorted_index(ii,1) = source; %storing correspondance between MNI8 and mni_coords
  403. end
  404. L2 = L(MNI_sorted_index); %sorting leadfield model (selected MEG channels only)
  405. Ltot2 = Ltot(MNI_sorted_index); %sorting leadfield model (all MEG channels)
  406. pos2 = pos(MNI_sorted_index,:); %sorting coordinates of brain sources for later plotting purposes
  407. %storing outputs
  408. OUT.L = L2; %leadfield model only for requested MEG channels (either magnetometers or gradiometers)
  409. OUT.Ltot = Ltot2; %all MEG channels leadfield model (useful for later plotting purposes)
  410. OUT.pos_brainsources_MNI8 = pos2; %position of coordinates
  411. OUT.vol = vol; %volume conduction model (vol) from FieldTrip
  412. OUT.sens = sens; %information about MEG channels
  413. OUT.transforms = BF.data.transforms; %transform matrix from and to MNI and native
  414. OUT.IDXMAG = IDXMAG; %storing magnetometers indices
  415. OUT.IDXGRAD = IDXGRAD; %storing gradiometers indices
  418. %%% 3) COMPUTING SPATIAL FILTERS
  419. disp('extracting MEG sensors data for covariance estimation')
  420. conditions = S.inversion.conditions;
  421. %%% EXTRACTING MEG SENSORS DATA FOR COVARIANCE ESTIMATION
  422. % disp('extracting MEG sensors data')
  423. %loading and extracting data (z-scored normalized data)
  424. if S.norm_megsensors.zscorel_cov == 1 %this is just to make sure to properly work on normalized or non-normalized MEG data
  425. Dc = spm_eeg_load([workdir '/Clone' fname ext]);
  426. else
  427. warning('you are computing covariance matrix on MEG sensors which are not zscore-normalized.. so you should expect much stronger contribution of magnetometers than gradiometers in the reconstructed signal')
  428. warning('please, note that even if you use only magnetometers or gradiometers, I still suggest you to zscore-normalize the data for covariance matrix computation')
  429. if exist('Dfilte','var')
  430. Dc = Dfilte; %getting filtered file, if you previously computed it
  431. else %otherwise original epoched data provided by the user
  432. Dc = spm_eeg_load(S.norm_megsensors.MEGdata_e);
  433. end
  434. end
  435. % Dc = spm_eeg_load([workdir '/Clone' fname ext]);
  436. if isempty(S.inversion.timef)
  437. warning('timef was not specified, so working on full length of the epoch.. you may want to check whether you need a subset of the time-window or the full epoch..')
  438. timef = 1:size(D,2);
  439. else
  440. timef = S.inversion.timef;
  441. end
  442. %getting indices of conditions
  443. condcat = [];
  444. for ii = 1:length(conditions) %over condition labels
  445. condcat = cat(2,condcat,find(strcmp((Dc.conditions),conditions{ii}))); %getting the condition numbers ordered according to the specification of the user (concatenated together since the covariance matrix is computed on all experimental conditions trials)
  446. end
  447. %extracting data (only MEG sensors and first timef time-points)
  448. %previous
  449. % idxMEGc1 = find(strcmp(Dc.chanlabels,'MEG0111')); %getting extreme channel index (first MEG channel, then I add 305 to get them all)
  450. % if isempty(idxMEGc1)
  451. % warning('there is no channel named MEG0111, is your data from Elekta-Neuromag?')
  452. % warning('trying to load MEG 0111, in case it is simply a problem of labels..')
  453. % idxMEGc1 = find(strcmp(Dc.chanlabels,'MEG 0111')); %getting extreme channel index (first MEG channel, then I add 305 to get them all)
  454. % end
  455. %new
  456. % idxMEGc1 = find(strcmp(Dc.chanlabels,'MEG0111')); %getting extreme channel index (first MEG channel, then I add 305 to get them all)
  457. idxMEGc1 = find(~cellfun(@isempty,strfind(Dc.chanlabels,'MEG'))); %elaborated way to get indices of MEG channels (even though I am mainly interested in the first MEG channel index)
  458. % chans = Dc.chanlabels(idxMEGc1:idxMEGc1+305); %extracting labels for MEG channels
  459. if S.beamfilters.sensl == 1 %magnetometers
  460. %previous
  461. % datameg = Dc(idxMEGc1:3:idxMEGc1+305,timef,condcat); %data
  462. %new
  463. datameg = Dc(idxMEGc1(1)-1 + IDXMAG,timef,condcat); %mag indices previously computed, here adding idxMEGc1(1), in case you have other channels before MEG channels; here I assume that once you have the first MEG channel then you have all the other ones, without any interruption
  464. elseif sensl == 2 %gradiometers
  465. %this solution would be better.. here and in general..
  466. %gradi = sort(cat(2,[2:3:306],[3:3:306])); magi = 1:3:306;
  467. %datameg = Dc(idxMEGc1+gradi,timef,condcat);
  468. %primitive codes.. previous
  469. % datameg = zeros(204,length(timef),length(condcat));
  470. % datameg(1:2:204,:,1:length(condcat)) = Dc(idxMEGc1+1:3:idxMEGc1+305,timef,condcat); %data
  471. % datameg(2:2:204,:,1:length(condcat)) = Dc(idxMEGc1+2:3:idxMEGc1+305,timef,condcat); %data
  472. %new
  473. datameg = Dc(idxMEGc1(1)-1 + IDXGRAD,timef,condcat); %mag indices previously computed, here adding idxMEGc1(1), in case you have other channels before MEG channels; here I assume that once you have the first MEG channel then you have all the other ones, without any interruption
  474. else
  475. datameg = Dc(idxMEGc1,timef,condcat); %data
  476. end
  477. time = Dc.time(timef);
  478. OUT.time = time; %storing time for later purposes
  480. disp('computing spatial filters')
  481. %YOU MAY CONSIDER TO TRY ALSO WITH COVARIANCE OVER TRIALS INSTEAD THAT OVER TIME.. BUT THIS IS MORE EXPERIMENTAL AND I WOULD DO THAT LATER..)
  482. %please, note that you compute the covariance matrix on the same (amount of) data that you want to invert later
  483. %THE SOLUTION BELOW (meancovl = 1 or meancovl = 3) SEEMS TO PROVIDE ALMOST THE SAME RESULTS (COVARIANCE OF CONCATENATED TRIALS SHOULD BE BETTER IN PRINCIPLE AND IT IS ALSO SLIGHTLY FASTER TO COMPUTE)
  484. %conversely, as expected, meancovl = 2 is not fine
  485. %I do not delete this since it may be useful to read it later on
  486. % meancovl = 3; %THIS SHOULD BE = 3
  487. % if meancovl == 1 %mean of trials covariance
  488. % Ct = zeros(size(datameg,1),size(datameg,1),size(datameg,3));
  489. % Cinvt = zeros(size(datameg,1),size(datameg,1),size(datameg,3));
  490. % for cc = 1:size(datameg,3) %over trials
  491. % Ct(:,:,cc) = cov(datameg(:,:,cc)'); %computing covariance matrix (time-sample x MEG channels, so covariance of MEG channels according to Matlab cov function)
  492. % Cinvt(:,:,cc) = pinv_plus(Ct(:,:,cc),50); %Mark Woolrich inv; 50 is the pca_dim related to Maxfilter computation (which makes the matrix of the data with a rank = 50)
  493. % end
  494. % C = mean(Ct,3); %mean over trials covariance
  495. % Cinv = mean(Cinvt,3); %mean over inv of trials coavariance
  496. % elseif meancovl == 2 %covariance calculated over mean of the trials
  497. % C = cov(ERFmean');
  498. % Cinv = pinv_plus(C,50); %Mark Woolrich inv; 50 is the pca_dim related to Maxfilter computation (which makes the matrix of the data with a rank = 50)
  499. % elseif meancovl == 3
  502. %computing covariance between MEG sensors (trials concatenated)
  503. datacov = reshape(datameg,size(datameg,1),size(datameg,2)*size(datameg,3))'; %data was extracted from SPM object previously loaded (2 sections above)
  504. C = cov(datacov); %computing covariance matrix (time-sample x MEG channels, so covariance of MEG channels according to Matlab cov function)
  505. Cinv = pinv_plus(C,50); %Mark Woolrich inv; 50 is the pca_dim related to Maxfilter computation (which makes the matrix of the data with a rank = 50)
  506. %this function uses at most X (in this case 50) PCs to estimate inverted matrix; in this case 50 should stabilize the computation of pinv
  507. % end
  508. W = cell(1,length(L2)); %initializing spatial filters
  509. WW = zeros(length(L2),size(L2{1},1));
  510. %computing the spatial filters (independently for each MEG source)
  511. for ii = 1:length(L2) %over brain sources
  512. llf = L2{ii}; %extracting brain source ii
  513. %combining dipole orientations
  514. tmp = llf' * Cinv * llf; %temporary estimation of covariance between sources
  515. nn = size(llf,2); %number of orientations
  516. [u, ~] = svd(real(pinv_plus(tmp(1:nn,1:nn),2,0)),'econ'); %single value decomposition, code from Mark Woolrich and Robert Oostenveld; (2 would be equal to the "reduce_rank" variable..)
  517. eta = u(:,1);
  518. llf = llf * eta; %actual reduction of the 3 orientations to 1 global vector
  519. %compuation of weights
  520. W{ii} = pinv_plus(llf' * Cinv * llf,2,0) * llf' * Cinv; %actual computation of spatial filter (Mark Woolrich pinv); I guess equivalent to: W{ii} = pinv_plus(llf' * Cinv * llf) * llf' * Cinv; pinv_plus(data,2,0) should not use 2 as maximum number of PCs.. anyway, I reproduce previous codes here..
  521. W{ii} = W{ii}./sqrt(W{ii}*W{ii}'); %applying a weight normalization to counterbalance the bias towards the centre of the head, as suggested by Woolrich, Huang, Van Veen
  522. % W{ii} = pinv(llf' * Cinv * llf) * llf' * Cinv; %actual computation of spatial filter
  523. WW(ii,:) = W{ii}(1,:);
  524. % disp(['computing spatial filters.. source: ' num2str(ii) ' / ' num2str(length(L2))])
  525. end
  526. %plotting images of filters (with weight normalisation)
  527. figure
  528. imagesc(WW)
  529. colorbar
  530. %storing output
  531. OUT.W = W;
  535. %%%%% III - FULL INVERSION %%%%%
  536. disp('%%%%% III - FULL INVERSION %%%%%')
  537. %%% EXTRACTING MEG SENSORS DATA
  538. disp('extracting MEG sensors data for actual inversion')
  539. %loading and extracting data
  540. if S.inversion.znorml == 1
  541. %z-scored normalized data
  542. Dc = spm_eeg_load([workdir '/Clone' fname ext]);
  543. else
  544. %original data (non-normalized)
  545. if exist('Dfilte','var')
  546. Dc = Dfilte; %getting filtered file, if you previously computed it
  547. else %otherwise epoched data provided by the user
  548. Dc = spm_eeg_load(S.norm_megsensors.MEGdata_e);
  549. end
  550. end
  551. %previous codes
  552. %extracting data (only MEG sensors and first timef time-points)
  553. % idxMEGc1 = find(strcmp(Dc.chanlabels,'MEG0111')); %getting extreme channel index (first MEG channel, then I add 305 to get them all)
  554. % if isempty(idxMEGc1)
  555. % warning('there is no channel named MEG0111, is your data from Elekta-Neuromag?')
  556. % warning('trying to load MEG 0111, in case it is simply a problem of labels..')
  557. % idxMEGc1 = find(strcmp(Dc.chanlabels,'MEG 0111')); %getting extreme channel index (first MEG channel, then I add 305 to get them all)
  558. % end
  559. % % chans = Dc.chanlabels(idxMEGc1:idxMEGc1+305); %extracting labels for MEG channels
  560. % if S.beamfilters.sensl == 1 %magnetometers
  561. % datameg = Dc(idxMEGc1:3:idxMEGc1+305,timef,:); %data
  562. % elseif sensl == 2 %gradiometers
  563. % datameg = zeros(204,length(timef),size(Dc,3));
  564. % datameg(1:2:204,:,:) = Dc(idxMEGc1+1:3:idxMEGc1+305,timef,:); %data
  565. % datameg(2:2:204,:,:) = Dc(idxMEGc1+2:3:idxMEGc1+305,timef,:); %data
  566. % else
  567. % datameg = Dc(idxMEGc1:idxMEGc1+305,timef,:); %data
  568. % end
  569. %new codes
  570. idxMEGc1 = find(~cellfun(@isempty,strfind(Dc.chanlabels,'MEG'))); %elaborated way to get indices of MEG channels (even though I am mainly interested in the first MEG channel index)
  571. if S.beamfilters.sensl == 1 %magnetometers
  572. datameg = Dc(idxMEGc1(1)-1 + IDXMAG,timef,:); %mag indices previously computed, here adding idxMEGc1(1), in case you have other channels before MEG channels; here I assume that once you have the first MEG channel then you have all the other ones, without any interruption
  573. elseif sensl == 2 %gradiometers
  574. datameg = Dc(idxMEGc1(1)-1 + IDXGRAD,timef,:); %mag indices previously computed, here adding idxMEGc1(1), in case you have other channels before MEG channels; here I assume that once you have the first MEG channel then you have all the other ones, without any interruption
  575. else
  576. datameg = Dc(idxMEGc1,timef,:); %data
  577. end
  578. time = Dc.time(timef);
  579. %getting indices of conditions (Old_Correct and New_Correct)
  580. conds = cell(1,length(conditions));
  581. for ii = 1:length(conditions) %over condition labels
  582. conds{ii} = find(strcmp((Dc.conditions),conditions{ii})); %getting the condition numbers ordered according to the specification of the user in S.conditions (sum to avoid the problem of zero-vector occurring if it does not find the match between the specified label and the data label.. of course if you specify two times the same label in the structure S it crashes as well.. but you would need to be quite peculiar to do that so I assume you will not do that..)
  583. end
  584. %meanll = 1 or = 0 seem to return very similar (almost identical) results; this is slightly surprising, I would expect similar but not almost identical results.. anyway.. the procedure seems correct, so I'll just live with it..
  585. % meanll = 1; % (SUGGESTED 1!) 1 for source reconstruction of mean over trials; 0 for mean over single-trials source reconstruction
  586. %I did not delete these lines since it may useful to read this information in the future
  587. % if meanll == 1 %mean over trials and then source reconstruction
  588. %computing mean of ERFs (consider to compute also median later..)
  589. if S.inversion.effects == 6 %custom data provided by the user
  590. ERFs = cell(1,size(S.inversion.alternative_data,3));
  591. for ii = 1:size(S.inversion.alternative_data{1},3) %over 3rd dimension of the custom data (which is supposed to be the new "condition")
  592. dumdim = S.inversion.alternative_data{1}; %extracting the data
  593. ERFs{ii} = dumdim(:,:,ii); %placing the data in the required format
  594. conditions = S.inversion.alternative_data{2}; %extrating the new conditions label
  595. end
  596. else
  597. ERFs = cell(1,length(conditions));
  598. for ii = 1:length(conditions) %over experimental conditions
  599. if ~isempty(conds{ii}) %it mis empty if there are is no match between condition requested by user and condition in MEG data (it may happen for a few subjects..)
  600. if S.inversion.effects == 1 %simple mean over trials
  601. ERFs{ii} = mean(datameg(:,:,conds{ii}),3);
  602. elseif S.inversion.effects == 2 %mean divided by st
  603. ERFs{ii} = mean(datameg(:,:,conds{ii}),3) ./ std(datameg(:,:,conds{ii}),0,3);
  604. elseif S.inversion.effects == 3 %mean divided by square root of st
  605. ERFs{ii} = mean(datameg(:,:,conds{ii}),3) ./ sqrt(std(datameg(:,:,conds{ii}),0,3)); %(variation of t-value, dividing mean over trials by their standard error
  606. elseif S.inversion.effects == 4 %mean divided by (st divided by square root of number of trials) (actual t-value..)
  607. ERFs{ii} = mean(datameg(:,:,conds{ii}),3) ./ (std(datameg(:,:,conds{ii}),0,3)./sqrt(length(conds{ii}))); %t-value (sort of), dividing mean over trials by their standard error
  608. elseif S.inversion.effects == 5 %all trials
  609. ERFs{ii} = datameg(:,:,conds{ii});
  610. end
  611. end
  612. end
  613. end
  614. %this should be ok, but a further check could be suggested..
  615. if S.beamfilters.sensl == 3 && S.inversion.znorml ~= 1 && S.inversion.effects ~= 6 %custom data provided by the user
  616. %if you have both MEG sensors and they should be normalized with reference to their overall maximum value
  617. %getting maximum and minimum values experimental conditions together
  618. maxx = zeros(length(conditions),1);
  619. minn = zeros(length(conditions),1);
  620. rmaxx = zeros(length(conditions),1);
  621. rminn = zeros(length(conditions),1);
  622. indg = sort([2:3:306 3:3:306]); %index of gradiometers
  623. for cc = 1:length(conditions) %over conditions
  624. if S.inversion.effects < 5
  625. maxx(cc) = max(max(abs(ERFs{cc}))); %maximum for each conditions (all MEG sensors, so magnetometers..)
  626. minn(cc) = min(min(abs(ERFs{cc}))); %minimum for each condition (all MEG sensors, so magnetometers..)
  627. rmaxx(cc) = max(max(abs(ERFs{cc}(indg,:)))); %maximum for each conditions (only gradiometers)
  628. rminn(cc) = min(min(abs(ERFs{cc}(indg,:)))); %minimum for each condition (only gradiometers)
  629. else
  630. maxx(cc) = max(max(max(abs(ERFs{cc})))); %maximum for each conditions (all MEG sensors, so magnetometers..)
  631. minn(cc) = min(min(min(abs(ERFs{cc})))); %minimum for each condition (all MEG sensors, so magnetometers..)
  632. rmaxx(cc) = max(max(max(abs(ERFs{cc}(indg,:,:))))); %maximum for each conditions (only gradiometers)
  633. rminn(cc) = min(min(min(abs(ERFs{cc}(indg,:,:))))); %minimum for each condition (only gradiometers)
  634. end
  635. end
  636. rmax = max(maxx); %global maximum over all conditions (all MEG sensors, so magnetometers..)
  637. rmin = min(minn); %global minimum over all conditions (all MEG sensors, so magnetometers..)
  638. rmaxg = max(rmaxx); %global maximum over all conditions (only gradiometers)
  639. rming = min(rminn); %global minimum over all conditions (only gradiometers)
  640. for cc = 1:length(conditions) %over conditions
  641. %this should be important since we want the experimental conditions
  642. %to maintain their different relative strengths (if you zscore
  643. %normalized (X-mean(X)./std(X), you should loose their relative
  644. %different strengths.. thus now:
  645. % 1) COVARIANCE MATRIX COMPUTED ON Z-SCORE NORMALIZED MEG DATA
  646. % 2) INVERSION COMPUTED ON MEG DATA SCALED WITH RESPECT TO RELATIVE DIFFERENCES BETWEEN EXPERIMENTAL CONDITIONS
  647. if S.inversion.effects < 5
  648. dumones = ones(size(ERFs{cc},1),size(ERFs{cc},2));
  649. dumones(ERFs{cc}<0) = -1; %getting a matrix with values -1 and 1 according to the original sign in ERFs{cc}
  650. ERFs{cc} = abs(ERFs{cc}); %absolute value
  651. ERFs{cc}(indg,:) = (ERFs{cc}(indg,:)-rming)./(rmaxg-rming).*(rmax-rmin) + rmin; %normalizing gradiometers on the basis of magnetometers (with respect to the conditions different strengths)
  652. ERFs{cc} = ERFs{cc} .* dumones; %assigning the original sign to the data
  653. else
  654. dumones = ones(size(ERFs{cc},1),size(ERFs{cc},2),size(ERFs{cc},3));
  655. dumones(ERFs{cc}<0) = -1; %getting a matrix with values -1 and 1 according to the original sign in ERFs{cc}
  656. ERFs{cc} = abs(ERFs{cc}); %absolute value
  657. ERFs{cc}(indg,:,:) = (ERFs{cc}(indg,:,:)-rming)./(rmaxg-rming).*(rmax-rmin) + rmin; %normalizing gradiometers on the basis of magnetometers (with respect to the conditions different strengths)
  658. ERFs{cc} = ERFs{cc} .* dumones; %assigning the original sign to the data
  659. end
  660. end
  661. end
  662. %until here
  665. disp('computing inversion')
  666. if S.inversion.effects ~= 5 %aggregated trials (e.g. averaged)
  667. sources_ERFs = zeros(length(W),length(time),length(conditions));
  668. for cc = 1:length(conditions) %over conditions
  669. if ~isempty(ERFs{cc}) %if there was no condition cc in MEG data
  670. for ii = 1:length(W) %over spatial filters (brain sources)
  671. for tt = 1:length(time) %over time-points
  672. sources_ERFs(ii,tt,cc) = W{ii} * ERFs{cc}(:,tt); %actual inversion
  673. end
  674. disp(ii)
  675. end
  676. end
  677. end
  678. else %single trials independently
  679. % sources_ERFs = zeros(length(W),length(time),length(conditions));
  680. s_tr_ERFs = cell(length(conditions),1); %there may be a different number of trials in different conditions (e.g. in learningbach-recogminor), so you need to sotre different conditions in cells
  681. for cc = 1:length(conditions) %over conditions
  682. if ~isempty(ERFs{cc}) %if there was no condition cc in MEG data
  683. sources_ERFs = zeros(length(W),length(time),size(ERFs{cc},3));
  684. for ii = 1:length(W) %over spatial filters (brain sources)
  685. for tt = 1:length(time) %over time-points
  686. for rr = 1:size(ERFs{cc},3) %over trials
  687. sources_ERFs(ii,tt,rr) = W{ii} * ERFs{cc}(:,tt,rr); %actual inversion
  688. end
  689. end
  690. disp(['spatial filter ' num2str(ii) ' / ' num2str(length(W)) ' - condition ' num2str(cc) ' / ' num2str(length(conditions))])
  691. end
  692. s_tr_ERFs{cc} = sources_ERFs;
  693. end
  694. end
  695. end
  696. %baseline correction (subtracting mean activity in the baseline)
  697. if ~isempty(S.inversion.bc)
  698. disp('computing baseline correction')
  699. if S.inversion.effects < 5
  700. for cc = 1:length(conditions) %over conditions
  701. if ~isempty(ERFs{cc}) %if there was no condition cc in MEG data
  702. for ii = 1:length(W) %over spatial filters (brain sources)
  703. sources_ERFs(ii,:,cc) = sources_ERFs(ii,:,cc) - mean(sources_ERFs(ii,S.inversion.bc(1):S.inversion.bc(2),cc),2);
  704. end
  705. end
  706. end
  707. else %single trials independently
  708. for cc = 1:length(conditions) %over conditions
  709. if ~isempty(ERFs{cc}) %if there was no condition cc in MEG data
  710. sources_ERFs = s_tr_ERFs{cc};
  711. for ii = 1:length(W) %over spatial filters (brain sources)
  712. for rr = 1:size(s_tr_ERFs{cc},3) %over trials
  713. sources_ERFs(ii,:,rr) = sources_ERFs(ii,:,rr) - mean(sources_ERFs(ii,S.inversion.bc(1):S.inversion.bc(2),rr),2);
  714. end
  715. disp(['baseline correction - condition ' num2str(cc) ' / ' num2str(length(conditions))])
  716. end
  717. s_tr_ERFs{cc} = sources_ERFs;
  718. end
  719. end
  720. end
  721. end
  722. % else %source reconstruction of single-trials and then mean
  723. % %extracting single-trial data
  724. % ERF_Old_st = datameg(:,:,conds{1}); %condition OLD
  725. % ERF_New_st = datameg(:,:,conds{2}); %condition NEW
  726. % sources_OLD = zeros(length(W),length(time));
  727. % sources_NEW = zeros(length(W),length(time));
  728. % for ii = 1:length(W) %over spatial filters (brain sources)
  729. % for tt = 1:length(time) %over time-points
  730. % %OLD condition
  731. % temptr = zeros(size(ERF_Old_st,3),1);
  732. % for rr = 1:size(ERF_Old_st,3) %over trials (OLD)
  733. % temptr(rr,1) = W{ii} * ERF_Old_st(:,tt,rr); %actual inversion
  734. % end
  735. % sources_OLD(ii,tt) = mean(temptr); %mean over trials for source ii and time-point tt
  736. % %NEW condition
  737. % temptr = zeros(size(ERF_New_st,3),1);
  738. % for rr = 1:size(ERF_New_st,3) %over trials (NEW)
  739. % temptr(rr,1) = W{ii} * ERF_New_st(:,tt,rr); %actual inversion
  740. % end
  741. % sources_NEW(ii,tt) = mean(temptr); %mean over trials for source ii and time-point tt
  742. % end
  743. % disp(['source ' num2str(ii) ' / ' num2str(length(W))])
  744. % end
  745. % end
  746. %scatter plotting (to check the distribution of the sources at a specific time-point)
  747. % figure
  748. % scatter(1:length(W),sources_OLD(:,timeselected),'o','r')
  749. % hold on
  750. % scatter(1:length(W),sources_NEW(:,timeselected),'o','b')
  752. %storing outputs
  753. OUT.data_MEG_sensors = ERFs;
  754. if S.inversion.effects < 5
  755. if S.inversion.abs == 1 %absolute values, if requested
  756. sources_ERFs = abs(sources_ERFs);
  757. end
  758. OUT.sources_ERFs = sources_ERFs; %no absolute values
  759. else
  760. if S.inversion.abs == 1 %absolute values, if requested
  761. for cc = 1:length(conditions) %over conditions
  762. s_tr_ERFs{cc} = abs(s_tr_ERFs{cc});
  763. end
  764. end
  765. OUT.sources_ERFs = s_tr_ERFs; %no absolute values
  766. end
  769. %%% MUST BE UPDATED SINCE NOW I GUESS IT WORKS WITH PREVIOUS MNI COORDINATES..
  770. %%%%% IV - SMOOTHING %%%%%
  771. %actual computations
  772. %spatial smoothing
  773. % if S.smoothing.spatsmootl == 1
  774. % disp('%%%%% IV - SMOOTHING %%%%%')
  775. % fwhm = S.smoothing.spat_fwhm; %spatial smoothing parameter
  776. % x1 = S.smoothing.temp_param;
  777. % %spatial smoothing
  778. % %preparing inputs
  779. % disp('computing spatial smoothing')
  780. % ERFs_spat_smooth = zeros(size(sources_ERFs,1),size(sources_ERFs,2),size(sources_ERFs,3));
  781. % for cc = 1:size(sources_ERFs,3) %over conditions
  782. % vol_as_matrix = sources_ERFs(:,:,cc); %data
  783. % %plotting original image
  784. % figure; imagesc(vol_as_matrix); title('original')
  785. % mask_fname = S.beamfilters.maskfname; %path to standard brain mask (MAYBE HERE THE NEW MASK??)
  786. % %actual function
  787. % ERFs_spat_smooth(:,:,cc) = smooth_vol_osl(vol_as_matrix, mask_fname, fwhm);
  788. % %plotting smoothed image
  789. % figure; imagesc(ERFs_spat_smooth(:,:,cc)); title(['spatially smoothed - condition ' S.inversion.conditions{cc}])
  790. % end
  791. % %storing outputs
  792. % OUT.spatial_smoothing = ERFs_spat_smooth;
  793. % sources_ERFs = ERFs_spat_smooth; %assigning spatially smoothed matrix to previously computed "full inversion" matrix since you want the spatially smoothed data to be passed to the next steps (temporally smoothing and printing nifti images)
  794. % else
  795. % OUT.spatial_smoothing = [];
  796. % end
  797. %temporal smoothing
  799. if S.smoothing.tempsmootl == 1 && S.inversion.effects < 5
  800. disp('computing temporal smoothing')
  801. %preparing inputs (same for all experimental conditions)
  802. x2 = 1; %dimension 1 of matrix requested by osl_gauss
  803. x3 = size(sources_ERFs,2); %dimension 2 of matrix requested by osl_gauss
  804. f = fftshift(osl_gauss(x1/(time(2)-time(1)),x2,x3)'); %current_level.time_smooth_std/tres,1,length(datstd))');
  805. for cc = 1:size(sources_ERFs,3) %over conditions
  806. dat = sources_ERFs(:,:,cc); %data (condition cc)
  807. %plotting original image
  808. figure; imagesc(dat); title('original')
  809. ERFs_temp_smooth = zeros(size(sources_ERFs,1),size(sources_ERFs,2),size(sources_ERFs,3));
  810. for ii = 1:size(sources_ERFs,1) %over brain voxels
  811. ERFs_temp_smooth(ii,:,cc) = fftconv(dat(ii,:)',f);
  812. end
  813. %plotting temporally smoothed image
  814. figure; imagesc(ERFs_temp_smooth(:,:,cc)); title(['temporally smoothed - condition ' S.inversion.conditions{cc}])
  815. if ~isempty(S.smoothing.tempplot) %plotting difference between original and temporal smoothed data (only for requested brain voxels)
  816. for pp = 1:length(S.smoothing.tempplot)
  817. figure
  818. plot(dat(S.smoothing.tempplot(pp),:))
  819. hold on
  820. plot(fftconv(dat(S.smoothing.tempplot(pp),:)',f))
  821. end
  822. end
  823. end
  824. %storing output
  825. OUT.ERFs_temp_smooth = ERFs_temp_smooth;
  826. sources_ERFs = ERFs_temp_smooth; %assigning temporally smoothed matrix to previously computed "full inversion" matrix since you want the printing nifti images to be run on temporally (and maybe also spatially) smoothed data
  827. else
  828. OUT.ERFs_temp_smooth = [];
  829. if S.smoothing.tempsmootl == 1 && S.inversion.effects == 5
  830. warning('spatial and temporal smoothing are implemented for averaged ERFs only')
  831. end
  832. end
  837. %%%%% V - PRINTING NIFTI IMAGE(s) %%%%%
  838. if S.nifti == 1
  839. if S.inversion.effects < 5
  840. disp('%%%%% V - PRINTING NIFTI IMAGE(s) %%%%%')
  841. % %getting maximum and minimum values experimental conditions together (used for scaling images)
  842. % maxx = zeros(length(conditions),1);
  843. % minn = zeros(length(conditions),1);
  844. % for cc = 1:length(conditions)
  845. % maxx(cc) = max(max(abs(sources_ERFs(:,:,cc)))); %maximum for each conditions
  846. % minn(cc) = min(min(abs(sources_ERFs(:,:,cc)))); %minimum for each condition
  847. % end
  848. % rmax = max(maxx); %global maximum over all conditions
  849. % rmin = min(minn); %global minimum over all conditions
  850. warning('loading MNI152-T1 8mm brain newly sorted set of coordinates.. remember that if you want a different spatial resolution (e.g. 2mm), you need to create a new mask and update this line of code!!')
  851. maskk = load_nii('/projects/MINDLAB2017_MEG-LearningBach/scripts/Leonardo_FunctionsPhD/External/MNI152_8mm_brain_diy.nii.gz'); %getting the mask for creating the figure
  852. for iii = 1:length(conditions)
  853. fnamenii = [workdir '/' S.out_name '_cond_' conditions{iii} '_norm' num2str(S.inversion.znorml) '_spatsmoot_' num2str(S.smoothing.spatsmootl) '_tempsmoot_' num2str(S.smoothing.tempsmootl) '_abs_' num2str(S.inversion.abs) '.nii.gz']; %path and name of the image to be saved
  855. % %normalization of nifti image
  856. % SO = (abs(sources_ERFs(:,:,iii))-rmin)./(rmax-rmin).*(100-0) + 0; %normalizing according to some scale (e.g. 0-100)
  857. % if S.inversion.abs ~= 1 %absolute values, if requested
  858. % dumones = ones(size(sources_ERFs,1),size(sources_ERFs,2)); %matrix of ones with dimensions of sources_ERFs(:,:,iii)
  859. % dumones(sources_ERFs(:,:,iii)<0) = -1; %getting a matrix with values -1 and 1 according to the original sign in sources_ERFs(:,:,iii)
  860. % SO = SO .* dumones; %rebuilding the original sign of the matrix with scaled values
  861. % end
  862. %without normalization
  863. SO = sources_ERFs(:,:,iii);
  864. %building nifti image
  865. SS = size(maskk.img);
  866. dumimg = zeros(SS(1),SS(2),SS(3),length(time));
  867. for ii = 1:size(sources_ERFs,1) %over brain sources
  868. dum = find(maskk.img == ii); %finding index of sources ii in mask image (MNI152_8mm_brain_diy.nii.gz)
  869. [i1,i2,i3] = ind2sub([SS(1),SS(2),SS(3)],dum); %getting subscript in 3D from index
  870. dumimg(i1,i2,i3,:) = SO(ii,:); %storing values for all time-points in the image matrix
  871. end
  872. nii = make_nii(dumimg,[8 8 8]); %making nifti image from 3D data matrix (and specifying the 8 mm of resolution)
  873. nii.img = dumimg; %storing matrix within image structure
  874. nii.hdr.hist = maskk.hdr.hist; %copying some information from maskk
  875. disp(['saving nifti image - condition ' num2str(iii)])
  876. save_nii(nii,fnamenii); %printing image
  877. end
  878. else
  879. warning('printing nifti images is implemented only for averaged ERFs')
  880. end
  881. end
  883. %saving output on disk
  884. disp('saving results')
  885. %for the future better saving assuming the all files are big
  886. if S.inversion.effects < 5 %normally small(ish) file
  887. save([S.norm_megsensors.workdir '/' S.out_name '_norm' num2str(S.inversion.znorml) '_abs_' num2str(S.inversion.abs) '.mat'],'OUT','-v7.3')
  888. else %single trial data can be way bigger..
  889. save([S.norm_megsensors.workdir '/' S.out_name '_norm' num2str(S.inversion.znorml) '_abs_' num2str(S.inversion.abs) '.mat'],'OUT','-v7.3')
  890. end
  892. %deleting Clone data and variable BF
  893. disp('deleting Clone data, BF and, if any, filtered MEG sensor data')
  894. delete([workdir '/BF_' fname '.mat']) %BF
  895. delete([workdir '/Clone' fname '.mat']) %Clone .mat
  896. delete([workdir '/Clone' fname '.dat']) %Clone .dat
  897. if ~isempty(S.norm_megsensors.freq) %if you have provided a frequency range
  898. delete([workdiroriginal '/' fname(7:end) '.mat']) %deleting the filtered MEG sensor data for saving space on disk (this does not affect the source reconstruction) .mat file
  899. delete([workdiroriginal '/' fname(7:end) '.dat']) %deleting the filtered MEG sensor data for saving space on disk (this does not affect the source reconstruction) .dat file
  900. delete([workdiroriginal '/' fname '.mat']) %epoched data by LBPD .mat
  901. delete([workdiroriginal '/' fname '.dat']) %epoched data by LBPD .dat
  902. end
  904. if S.Aarhus_cluster == 1
  905. OUT = []; %Aarhus cluster usually do not have output like normal Matlab functions, but saves results on disk
  906. end
  909. end

MEG_SR_Beam_LBPD.m at commit 17ce460, under GPL-3.0 · at the source

Overview

Authors: Chiara Malvaso1,2, Gemma Fernández‐Rubio1, Mattia Rosso1, Elisa Serra1,3,4, Vera Rudi1,3,4, Peter Vuust1, Morten L Kringelbach1,3,4, Claudia Testa2, Leonardo Bonetti1,3,4
ORCID iDs: Chiara Malvaso, Gemma Fernández‐Rubio, Leonardo Bonetti
  1. Center for Music in the Brain, Department of Clinical Medicine, Aarhus University & The Royal Academy of Music, Aarhus/Aalborg, Denmark
  2. Department of Physics and Astronomy, University of Bologna, Bologna, Italy
  3. Centre for Eudaimonia and Human Flourishing, Linacre College, University of Oxford, Oxford, UK
  4. Department of Psychiatry, University of Oxford, Oxford, UK
Institutions: University of Bologna (Italy); Aarhus University (Denmark); Royal Academy of Music (Denmark); Ghent University (Belgium); University of Oxford (United Kingdom)
Journal: Annals of the New York Academy of Sciences, volume 1562, issue 1, article e70349
Dates: received 9 September 2025; accepted 21 May 2026; published online 30 July 2026; in print August 2026
Type: Research article · Language: English
License: CC BY
Identifiers: DOI 10.1111/nyas.70349 · PMID 42529957 · PMCID PMC13419915 · OpenAlex W4412024099
Open access: hybrid, a free copy (OpenAlex)
Preprint: doi:10.1101/2025.06.30.662094
Status: code verified
Categories: MEG (modality), human (organism), cognitive (subfield)
Methods: Spectral & time-frequency, Connectivity, Statistics, Smoothing, state filtering, decompositions, Machine learning, Preprocessing, fMRI & imaging, Physiology & signal measures
MeSH: Aging*, Auditory Perception*, Brain*, Nerve Net*, Rest*, Acoustic Stimulation, Adolescent, Adult, Aged, Brain Mapping, Female, Humans, Magnetoencephalography, Male, Middle Aged, Young Adult (* major topic)
Topic: Functional Brain Connectivity Studies (Cognitive Neuroscience, Neuroscience), according to OpenAlex
Funding: Pettit Foundation; Society for Education, Music and Psychology Research (SEMPRE); Nordic Mensa Fund; Lundbeck Foundation (R469-2024-1573); Carlsberg Foundation; Luxembourg National Research Fund (FNR) (17906488); Købmand herman Sallings Fond; Independent Research Fund Denmark (DFF) (10.46540/5253-00003B); Danish National Research Foundation (DNRF117); Aker Scholarship; Danmarks Grundforskningsfond (DNRF117)
Citations: not cited yet (Europe PMC); 82 references in the paper

Abstract

Understanding how brain networks operate across different frequencies during cognitive tasks, and how these dynamics change with age, remains a central challenge in cognitive neuroscience. While previous studies have focused on resting‐state activity and passive listening, less is known about frequency‐specific brain dynamics during event‐related tasks that require active memory engagement. In this study, we extend the recently developed FREQ‐NESS analytical pipeline by adapting it to event‐related task and resting‐state source‐reconstructed magnetoencephalography (MEG) data from 140 healthy participants. This method quantified the variance explained by frequency‐specific brain networks, their spatial organization, and associated time‐resolved power estimates. We found significant effects of age, condition, and their interaction in the variance explained by leading components at 8.6, 10.0, and 20.0 Hz. Older adults exhibited peaks at 8.6 and 10.0 Hz across both rest and task, while younger adults displayed a task‐related reduction, suggesting a different organization of brain networks during memory processing with age. Time–frequency analysis revealed age‐ and condition‐dependent desynchronization in the alpha and beta bands (7.1–22.9 Hz). These findings demonstrate the effectiveness of the adapted FREQ‐NESS pipeline for event‐related tasks and highlight the importance of frequency‐resolved network analysis for characterizing age‐related changes in active auditory memory processing.

Reproduced under the paper's license (CC BY), from the paper cited above.

Repositories

Its files are read in the Code ↔ Paper reader above, with 10 matches between paragraphs and lines of code.

leonardob92/LBPD-1.0

License: GPL-3.0
State: the link answers, verified on 26 September 2026
Evidence: files inventoried
Commit: 17ce460e8036eb77f2d81dc2fa0f24c0a8ecb6b1, 2 June 2025
Languages: MATLAB (166), Java (1)
Size: 201 files, 167 scripts
Software Heritage: not archived
Found in: “Data Availability Statement”
Holds: README, license file
Not found: CITATION.cff, environment file, tests, continuous integration, documentation
Tools: OHBA Software Library (OSL) (32 files), SPM (19 files), Statistics and Machine Learning Toolbox (18 files), Tools for NIfTI and ANALYZE image (MATLAB) (15 files), FieldTrip (10 files), Brain Connectivity Toolbox (1 file), export_fig (1 file), FSL (1 file), Image Processing Toolbox (1 file), Signal Processing Toolbox (1 file), Connectome Workbench (1 file)
Availability: 1 check, the latest on 26 September 2026: the link answers
  • 26 September 2026: the link answers
168 files
At the source: github.com/leonardob92/LBPD-1.0.git

malvasochiara/FREQ-NESS-for-age-related-differences-in-brain-networks-during-auditory-recognition-and-resting

License: MIT
State: the link answers, verified on 26 September 2026
Evidence: files inventoried
Commit: e798c9ff809bda4c6f72d84cde2f9a056dc050ff, 23 April 2026
Languages: MATLAB (12)
Size: 13 files, 12 scripts
Software Heritage: not archived
Found in: “Data Availability Statement”
Holds: license file
Not found: README, CITATION.cff, environment file, tests, continuous integration, documentation
Tools: Signal Processing Toolbox (2 files), FSL (1 file), Statistics and Machine Learning Toolbox (1 file), Tools for NIfTI and ANALYZE image (MATLAB) (1 file), OHBA Software Library (OSL) (1 file), SPM (1 file)
Availability: 1 check, the latest on 26 September 2026: the link answers
  • 26 September 2026: the link answers
13 files
At the source: github.com/malvasochiara/FREQ-NESS-for-age-related-differences-in-brain-networks-during-auditory-recognition-and-resting

mattiarosso92/frequency-resolved_brain_network_estimation_via_source_separation_freq-ness

License: none: the authors keep all their rights
State: the link answers, verified on 26 September 2026
Evidence: files inventoried
Commit: 49e4b7815a0458e5f7e0e3904e3bf5c9efbd2f90, 9 September 2026
Languages: MATLAB (116), Python (32), Jupyter (1)
Size: 170 files, 149 scripts
Software Heritage: not archived
Found in: “Data Availability Statement”
Holds: README, environment (FREQNESS_Toolbox/python/pyproject.toml), tests, continuous integration, 1 notebook
Not found: license file, CITATION.cff, documentation
Tools: NumPy (31 files), Tools for NIfTI and ANALYZE image (MATLAB) (19 files), Matplotlib (15 files), SciPy (10 files), Statistics and Machine Learning Toolbox (5 files), SPM (4 files), FSL (3 files), OHBA Software Library (OSL) (3 files), NiBabel (2 files), Image Processing Toolbox (1 file)
Availability: 1 check, the latest on 26 September 2026: the link answers
  • 26 September 2026: the link answers
150 files
At the source: github.com/mattiaRosso92/Frequency-resolved_brain_network_estimation_via_source_separation_FREQ-NESS/tree/main/FREQNESS_Toolbox

The paper's code and data availability statement is in the Data section.

Tracing map

Proposed by the machine: these links were found in the paper and verified at the source, without human review. The map will receive a Zenodo DOI once one of the paper's authors has validated it with their ORCID.

What the map holds:

  • 3 repositories of the authors' code, each at its verified commit, with its license and how the link was found in the paper;
  • 328 scripts, each with its path and the digest of its content;
  • 10 matches between paragraphs of the paper and lines of the code (method lexical-v1);
  • neither the text of the paper nor the code itself.

Its JSON (tracing-map.json) is deposited on Zenodo with its DOI once the map is validated.

Data

No dataset and no data link were found in the paper.

Data Availability Statement

The code for the preprocessing is available at the following link: https://github.com/leonardob92/LBPD‐1.0.git (https://github.com/leonardob92/LBPD-1.0.git)

The pipeline for the current study is available at the following link: https://github.com/malvasochiara/FREQ‐NESS‐for‐age‐related‐differences‐in‐brain‐networks‐during‐auditory‐recognition‐and‐resting (https://github.com/malvasochiara/FREQ-NESS-for-age-related-differences-in-brain-networks-during-auditory-recognition-and-resting)

The FREQ‐NESS Toolbox is available at the following link: https://github.com/mattiaRosso92/Frequency‐resolved_brain_network_estimation_via_source_separation_FREQ‐NESS/tree/main/FREQNESS_Toolbox (https://github.com/mattiaRosso92/Frequency-resolved_brain_network_estimation_via_source_separation_FREQ-NESS/tree/main/FREQNESS_Toolbox)

Reproduced under the paper's license (CC BY), from the paper cited above.

Versions

The history of this record: each version stored by the harvester or made by a correction of its authors or of the maintainers of its code, and what changed in its facts. The texts of the paper (its abstract, its availability statements) are not part of it; versions that changed only those are not listed.

Version 3, 28 September 2026

  • Publisher: n/a → Wiley

Version 1, 27 September 2026: the first record

Recorded: type, language, journal, volume, issue, pages, dates, 9 authors, 16 MeSH terms, 11 funders, 79 references.

Cite

This paper

Malvaso, C., Fernández‐Rubio, G., Rosso, M., Serra, E., Rudi, V., Vuust, P., Kringelbach, M. L., Testa, C., & Bonetti, L. (2026). FREQ-NESS Reveals Age-Related Differences in Frequency-Resolved Brain Networks During Auditory Recognition and Resting State. Annals of the New York Academy of Sciences, 1562(1), e70349. https://doi.org/10.1111/nyas.70349

BibTeX

@article{malvaso2026freq,
author = {Malvaso, Chiara and Fernández‐Rubio, Gemma and Rosso, Mattia and Serra, Elisa and Rudi, Vera and Vuust, Peter and Kringelbach, Morten L and Testa, Claudia and Bonetti, Leonardo},
title = {{FREQ-NESS Reveals Age-Related Differences in Frequency-Resolved Brain Networks During Auditory Recognition and Resting State}},
journal = {Annals of the New York Academy of Sciences},
year = {2026},
month = aug,
volume = {1562},
number = {1},
pages = {e70349},
publisher = {Wiley},
issn = {0077-8923},
doi = {10.1111/nyas.70349},
url = {https://doi.org/10.1111/nyas.70349},
pmid = {42529957},
pmcid = {PMC13419915}
}

RIS

TY - JOUR
AU - Malvaso, Chiara
AU - Fernández‐Rubio, Gemma
AU - Rosso, Mattia
AU - Serra, Elisa
AU - Rudi, Vera
AU - Vuust, Peter
AU - Kringelbach, Morten L
AU - Testa, Claudia
AU - Bonetti, Leonardo
TI - FREQ-NESS Reveals Age-Related Differences in Frequency-Resolved Brain Networks During Auditory Recognition and Resting State
T2 - Annals of the New York Academy of Sciences
J2 - Ann N Y Acad Sci
PY - 2026
DA - 2026/08/01
VL - 1562
IS - 1
SP - e70349
SN - 0077-8923
PB - Wiley
DO - 10.1111/nyas.70349
UR - https://doi.org/10.1111/nyas.70349
LA - en
ER -

CSL-JSON

{
"id": "10.1111/nyas.70349",
"type": "article-journal",
"title": "FREQ-NESS Reveals Age-Related Differences in Frequency-Resolved Brain Networks During Auditory Recognition and Resting State",
"container-title": "Annals of the New York Academy of Sciences",
"author": [
{
"family": "Malvaso",
"given": "Chiara"
},
{
"family": "Fernández‐Rubio",
"given": "Gemma"
},
{
"family": "Rosso",
"given": "Mattia"
},
{
"family": "Serra",
"given": "Elisa"
},
{
"family": "Rudi",
"given": "Vera"
},
{
"family": "Vuust",
"given": "Peter"
},
{
"family": "Kringelbach",
"given": "Morten L"
},
{
"family": "Testa",
"given": "Claudia"
},
{
"family": "Bonetti",
"given": "Leonardo"
}
],
"container-title-short": "Ann N Y Acad Sci",
"volume": "1562",
"issue": "1",
"page": "e70349",
"DOI": "10.1111/nyas.70349",
"PMID": "42529957",
"PMCID": "PMC13419915",
"ISSN": "0077-8923",
"publisher": "Wiley",
"URL": "https://doi.org/10.1111/nyas.70349",
"language": "en",
"issued": {
"date-parts": [
[
2026,
8,
1
]
]
}
}

The tracing map gets a citation of its own once an author has validated it and it has a DOI.

Similar papers

The papers with a page that share the most with this one: the tools found in their code, their categories, datasets, cited references and authors, the rarest counting most.

[1] doi:10.1007/s00429-025-03012-5 [code]
The neurophysiology of healthy and pathological aging: a comprehensive systematic review
Journal: n/a
In common: MEG, 30 references, 2 authors
[2] doi:10.1002/advs.77857 [code]
Brain Network Dynamics of Local and Global Predictive Processing in Aging.
Journal: Advanced science (Weinheim, Baden-Wurttemberg, Germany)
In common: OHBA Software Library (OSL), export_fig, Brain Connectivity Toolbox, 9 other tools, cognitive, 19 references, 2 authors
[3] doi:10.1111/nyas.70314 [code]
Neural Oscillatory Dynamics in Joint Action: Dissociable Roles of Entrainment and Beta Modulation in Self-Other Integration.
Journal: Annals of the New York Academy of Sciences
In common: FieldTrip, Signal Processing Toolbox, Statistics and Machine Learning Toolbox, cognitive, 9 references
[4] doi:10.7554/elife.108408 [code]
Frequency and laminar profile of feature-specific visual activity revealed by interleaved EEG-fMRI.
Journal: eLife
In common: Connectome Workbench, Tools for NIfTI and ANALYZE image (MATLAB), FieldTrip, 9 other tools, 2 references
[5] doi:10.1038/s41467-026-73668-y [code]
Convergent and divergent brain-cognition development in early adolescence.
Journal: Nature communications
In common: Connectome Workbench, Tools for NIfTI and ANALYZE image (MATLAB), FieldTrip, 9 other tools, 1 reference
[6] doi:10.1038/s41398-026-04025-2 [code]
Brain energetic landscapes shape state dysregulation in major depressive disorder: a morphological network controllability perspective.
Journal: Translational psychiatry
In common: Brain Connectivity Toolbox, Connectome Workbench, Tools for NIfTI and ANALYZE image (MATLAB), 9 other tools
[7] doi:10.1038/s41467-026-71151-2 [code]
Common and distinct neural correlates of social interaction processing and theory of mind in narratives.
Journal: Nature communications
In common: Brain Connectivity Toolbox, Connectome Workbench, FieldTrip, 9 other tools, cognitive
[8] doi:10.1038/s41467-026-76452-0 [code]
Music evokes shared neural representations of imagined narratives across sensory modalities.
Journal: Nature communications
In common: Connectome Workbench, Tools for NIfTI and ANALYZE image (MATLAB), FieldTrip, 8 other tools, cognitive, 1 reference
[9] doi:10.1038/s41467-026-76011-7 [code]
Human cortex organizes dynamic co-fluctuations along the sensorimotor-association axis.
Journal: Nature communications
In common: Brain Connectivity Toolbox, Connectome Workbench, FieldTrip, 9 other tools
[10] doi:10.1016/j.crmeth.2026.101473 [code]
AmygdalaGo-BOLT for boundary-aware segmentation of the human amygdala.
Journal: Cell reports methods
In common: Brain Connectivity Toolbox, Connectome Workbench, FieldTrip, 9 other tools

Contribute

The authors of this paper can claim it, correct its record and validate its tracing map, and the maintainers of its code (its owner, or a public member of its organization) correct what it says of their repository; anyone signed in can ask for its removal. Every request goes to OSCR's own machine, which answers it; your account page follows them.

Sign in with ORCID to claim this paper as one of its authors, correct its record or validate its tracing map: when the paper's metadata lists your ORCID iD, you are recognized at once. Maintainers of its code: sign in with GitHub, then claim the repository on your account page.

Request its removal

To ask OSCR to remove this record, the copies of its authors' scripts or its tracing map, use the removal request page: signed in, you say who you are, what to remove and why, then review and confirm the request. Published rules decide every request (how).

Discussion, reproductions, activity

Discussion: questions and error reports about this paper and its code, from signed-in readers and its authors. It opens with sign-in.

Reproductions: reports from readers who ran the authors' code: what they reproduced, with which environment, commit and data. It opens with sign-in.

Activity: what happens around this paper: new versions of its record, its map's validation, discussions and reproductions. It opens with sign-in.