Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Both sides previous revision Previous revision
Next revision
Previous revision
example:ecog_ny [2018/07/11 13:00]
robert
example:ecog_ny [2018/08/04 19:54] (current)
arjen [Data analysis]
Line 17: Line 17:
 <​code>​ <​code>​
 %% load electrode locations %% load electrode locations
-fid = fopen('​NY394_fmri_coor_T1.txt'​);​+fid = fopen('​NY394_MRI_coor.txt'​);​
 elec_info = textscan(fid,'​%s %f %f %f %s'); elec_info = textscan(fid,'​%s %f %f %f %s');
 fclose(fid);​ fclose(fid);​
Line 28: Line 28:
  
 %% load pial surface %% load pial surface
-load('NY394_fmri_rh_pial_surf.mat'​);​+load('NY394_MRI_rh_pial_surface.mat'​);​
  
 % create FieldTrip surface mesh structure % create FieldTrip surface mesh structure
 mesh      = []; mesh      = [];
-mesh.pos ​ = coords+mesh.pos ​ = surface.pos
-mesh.tri ​ = faces;+mesh.tri ​ = surface.tri;
 mesh.unit = '​mm';​ mesh.unit = '​mm';​
  
Line 52: Line 52:
 === 1. Data preprocessing === === 1. Data preprocessing ===
  
-First, we will load the data and segment them into trials using **[[/​reference/​ft_preprocessing]]**. Event information ​have already been extracted from the trigger channels and stored ​in in NY394_trl.mat. The segmentation of continuous data based on triggers is described in detail in one of the [[tutorial:​preprocessing|preprocessing tutorials]].+First, we will load the data and segment them into trials using **[[/​reference/​ft_preprocessing]]**. Event information ​has already been extracted from the trigger channels and stored in NY394_trl.mat. The segmentation of continuous data based on triggers is described in detail in one of the [[tutorial:​preprocessing|preprocessing tutorials]].
  
 <​code>​ <​code>​
Line 85: Line 85:
 {{:​tutorial:​ny394_bad_channel.png?​400|}} {{:​tutorial:​ny394_bad_channel.png?​400|}}
  
-For rejecting artifact trials, we will use the '​summary'​ method in **[[/​reference/​ft_rejectvisual]]**. Identifying artifact trials in ECoG is similar to EEG analysis and can be done according to the  tutorial on [[tutorial:​visual_artifact_rejection|visual artifact rejection]]. Note, that ECoG data typically have higher amplitudes and better signal-to-noise ratios compared with data from scalp EEG, because they are recorded directly from the cortex. Still, a number of  technical and physiological artifacts can be present in the data. Due to the clinical - and therefore less rigorously controlled - environment during the recording process, technical artifacts are quite common. The present dataset is relatively clean and, hence, does not need much rejection. Some moderate outliers can be found for the metrics: maxabs, zvalue and maxzvalue.+For rejecting artifact trials, we will use the '​summary'​ method in **[[/​reference/​ft_rejectvisual]]**. Identifying artifact trials in ECoG is similar to EEG analysis and can be done according to the tutorial on [[tutorial:​visual_artifact_rejection|visual artifact rejection]]. Note, that ECoG data typically have higher amplitudes and better signal-to-noise ratios compared with data from scalp EEG, because they are recorded directly from the cortex. Still, a number of technical and physiological artifacts can be present in the data. Due to the clinical - and therefore less rigorously controlled - environment during the recording process, technical artifacts are quite common. The present dataset is relatively clean and, hence, does not need much rejection. Some moderate outliers can be found for the metrics: maxabs, zvalue and maxzvalue.
  
 <​code>​ <​code>​
Line 100: Line 100:
  
 == 3.1 calculate and plot ERPs == == 3.1 calculate and plot ERPs ==
-The analysis of event-related potentials is done in accordance with the standard [[tutorial:​preprocessing_erp|ERP tutorial]]. Here, we will calculate and compare the ERPs of two conditions ('​object'​ and '​face'​). In the parameters of **[[/​reference/​ft_timelockanalysis]]** we use the preprocessing options to apply filters to the data (30 Hz low-pass, 1 Hz high-pass). The high-pass filter reduces slow drifts while the low-pass filter eliminates high frequency noise. Note that the exact filter setting ​depend ​on the ERP component under investigation. Baseline correction is subsequently done with respect to the time interval of -300 ms to - 50 ms.+The analysis of event-related potentials is done in accordance with the standard [[tutorial:​preprocessing_erp|ERP tutorial]]. Here, we will calculate and compare the ERPs of two conditions ('​object'​ and '​face'​). In the parameters of **[[/​reference/​ft_timelockanalysis]]** we use the preprocessing options to apply filters to the data (30 Hz low-pass, 1 Hz high-pass). The high-pass filter reduces slow drifts while the low-pass filter eliminates high-frequency noise. Note that the exact filter setting ​depends ​on the ERP component under investigation. Baseline correction is subsequently done with respect to the time interval of -300 ms to - 50 ms.
  
 <​code>​ <​code>​
Line 126: Line 126:
 </​code>​ </​code>​
  
-For plotting the data we select channel '​IO_03',​ located in or in close proximity of the fusiform face area, which is known to strongly respond to face stimuli. In agreement with the literature, the ERPs appear to be larger for '​face'​ compared to ‘object’ stimuli. Before plotting, we need to average the data across trials, because we kept the individual trials when initially calling ​ft timelockanalysis.+For plotting the data we select channel '​IO_03',​ located in or in close proximity of the fusiform face area, which is known to strongly respond to face stimuli. In agreement with the literature, the ERPs appear to be larger for '​face'​ compared to ‘object’ stimuli. Before plotting, we need to average the data across trials, because we kept the individual trials when initially calling ​ft_timelockanalysis.
  
 <​code>​ <​code>​