About Cartool‎ > ‎Cartool Releases‎ > ‎

Cartool 4698 Springtime 2017 edition

posted May 3, 2017, 3:28 AM by Cartool Software FBMLab   [ updated Oct 9, 2017, 9:19 AM ]
This is the latest edition of Cartool, with a lot of changes in the micro-states segmentation, inverse matrices computation, etc...

What's new (long):

  • Optimally tuned Meta-Criterion for clustering:
    • Through extensive tests with synthetic data, the optimal clustering criteria have been selected (7 of them).
    • An optimal merging formula has also been devised, to produce a single synthetic Meta-Criterion.
    • Meta-Criterion works for ERPs, Resting States, either on individual or group analysis.
  • Segmentation in Inverse Space:
    • Introduced since awhile, the method has been improved and refined. Basically, it's the clustering just done in the Inverse Space instead of the EEG space.
    • For ERPs, RIS templates can be anything they want. But for Resting States, RIS templates are now forced to be non-overlapping, like independent regions.
    • It extracts both the clustering and the localization in the same process! It has been tested extensively on a lot of datasets, and produces consistent results.
  • Segmentation with a dual dataset !NEW!:
    • Within the EEG RS analysis, Cartool can now compute the corresponding RIS templates from the EEG templates. This allows to see what were the sources associated with all the EEG maps clustered together.
    • Conversely, it can compute the corresponding EEG templates from the RIS templates. This time it allows to see the EEG maps associated with all the Inverse data clustered together.
    • This is the optimal way to have the correspondance between the EEG and the Inverse Space templates.
  • Fitting will output more correlation data: the spatial correlation time courses, but also some cross-correlations of these time courses.
  • Frequency Analysis:
    • Frequency Analysis in the Inverse Space !NEW!:
      • Perform first the frequency analysis on the EEG, saving the results to Complex values.
      • Cartool can now apply the Inverse Matrix to these Complex data to obtain the resulting sources (Tools | Inverse | Computing RIS). This is done by applying the Real and Imaginary parts separately, then merging the results together.
    • You can now save the Phase from the wavelet analysis (S-Transform).
  • Correlation analysis:
    • 3 types of analysis are now available:
      Spatial Correlation; Time-course Correlation; Phase-Intensity Coupling.
      Most of these analysis can be done on tracks or on frequencies, with Cartool performing a lot of checks on the given data.
    • Spatial Correlation:
      • Given some EEG (or RIS) files, and some templates, computes the instantaneous spatial correlation with each template. This is the same as in the Fitting.
      • Results are time courses of these correlations.
    • Time-course correlation:
      • Given some time courses (whatever they are), compute the correlation between them. F.ex. to see the correspondance between templates spatial correlation and the solution points.
      • Result is a matrix of correlations.
    • Phase-Intensity Coupling !NEW!:
      • Given some phase and intensity files, computes the coupling between the two.
      • Result is a frequency file, showing the coupling for a given ROI / track with all the others.
  • Generating synthetic data for various tests, called from menu  Tools|Generating Synthetic Data:
    • There are currently 3 ways to generate data: EEG segments, RIS segments or dipole sources.
    • From a given number of core EEG maps, you can generate synthetic EEG segments. Correlation, noise levels and segments durations are controlled by the user.
    • Likewise in the Inverse Space, generating artificial inverse segmented data.
    • Any  number of oscillating dipoles, each with its own location, orientation, amplitude, frequency and phase. Results are given in both the EEG and Inverse spaces.
  • Inverse Display: you can now select the solution points and it will show up in the RIS tracks, or select some RIS tracks and it will show up in the inverse solution points. This is similar to the maps display, where you can select the electrodes.
  • Other stuff:
    • Utility to split freq files in any of its dimensions: per frequency, per eelctrode or per spectrum.
    • Refreshed the MRIs | Preprocessing utility. You can now explicitly reorient, downsample, coregister, extract brain + Bias Field correct.
    • Improved the topographic PCA, to be working in both EEG and RIS spaces.
    • Reading Analyze files with exotic formats, like float / doubles with possible NaN values...
    • Fixing old bugs (..and introducing new ones for the LOLs).