|
Notes:
CONN main release (19c)
Installation
instructions: https://web.conn-toolbox.org/resources/installation
Changes:
19.c CUMULATIVE PATCH UPDATE FOR PREVIOUS 19.b RELEASE
patch to fix logged description of denosing procedure parameters
(thanks to James Ford for bringing this issue up; https://www.nitrc.org/forum/forum.php?th...)
patch to display group-labels when using ROIs from atlas.nii file
('CONN atlas apriori ROI order/clusters' in ROI-to-ROI second-level
results) and networks.nii file ('CONN networks apriori ROI
order/clusters' option in ROI-to-ROI second-level results)
patch to fix error when using conn_module('glm'...,'mask',...) to
run ROI-to-ROI second-level analyses with masking
patch to fix error when using analysis masks close to image borders
(thanks for Jan Maloman for bringing up this issue; https://www.nitrc.org/forum/forum.php?th...)
patch to fix error when importing BIDS datasets without zero-padded
subject ID numbers (thanks to Hongru Zhu for bringing up this
issue; https://www.nitrc.org/forum/forum.php?th...)
(*) patch to fix 'mask with grey matter' option in Setup.ROIs tab
to mask within thresholded grey matter mask instead of original
grey matter mask (see conn_batch binary_threshold and
binary_threshold_type for additional options; change erosion
settings "binarization threshold" field to 0 for
back-compatibility)
patch to add 'graphic options' button in ROI-to-ROI seccond-level
results GUI (shows/hides advanced graphic-options menubar) (thanks
to Joey Contreras for bringing up this issue; https://www.nitrc.org/forum/forum.php?th...)
patch to add 'export data' option in ROI-to-ROI second-level
results (exports ROI-to-ROI connectivity matrices in matrix NIFTI
format, which can then be entered as input to
conn_module('glm',...))
patch to add ability to read subject-specific freesurfer annotation
.annot files (thanks to Amy Bouchard for bringing up this issue; https://www.nitrc.org/forum/forum.php?th...)
patch to add Yeo2011 7-networks template (in conn/utils/otherrois)
as ROI or as template for ICA component identification (spatial
match to template) (thanks to Yiquan for suggesting this addition https://www.nitrc.org/forum/message.php?...)
patch to add 'merge (not-in-union)' option to "conditions tool"
menu allowing to compute an exclusive conjunction of several
conditions (those scans/timepoints not already covered by one or
several other conditions; e.g. to explicitly define a
reference/baseline effect from an event related design) (thanks to
Alex for suggesting this addition https://www.nitrc.org/forum/forum.php?th...)
CONN 19.b changes:
Cumulative patch, bug fixes and improvements
Added new interface for recommended/standard as well as advanced
thresholding options in results explorer voxel-based and ROI-to-ROI
display
Added Threshold Free Cluster Enhancement (TFCE) option to
voxel-based second-level results (Smith & Nichols 2009)
Added Threshold Free Cluster Enhancement (TFCE) option to
ROI-to-ROI second-level results
Added Spatial Pairwise Clustering (SPC) option to ROI-to-ROI
second-level results (Zalesky et al. 2012)
Integrated second-level results explorer interface for voxel-level,
surface-level, and ROI-to-ROI second-level analyses (see "help
conn_display")
Added option to automatically import fMRIPrep-preprocessed datasets
(see Tools.Convert/Import.fromFMRIPREPdataset )
Added option to define and estimate hierarchical multi-level models
(use 'hierarchical model' in Setup.Conditions to implicitly define
a new condition as an arbitrary function/transformation of other
previously defined input conditions, e.g. to estimate connectivity
growth parameters of interest in a growth-curve longitudinal
analyses)
Improved 'Plot effects' option in seed-to-voxel and voxel-to-voxel
results explorer windows to provide effect-size information about
individual effects and conditions in current second-level analysis
within clusters of interest
Improved 'Data preview' and 'Plot effects' options in second-level
Results tab to provide effect-size information about individual
effects and conditions in current second-level analysis at each
target voxel or ROI
(*) Modified computation of Quality Control variables
'QC_MeanMotion' and 'QC_MeanGSchange' to compute average
motion/GSchange measures over valid scans only (disregarding scans
already marked as outliers)
For back-compatibility or mid-analysis updates: change the line in
conn_setup_preproc.m that reads "MEANSOVERVALIDONLY=true" to
"MEANSOVERVALIDONLY=false"
Added internal check for non-estimable contrasts when displaying
second-level results design information (part of '2nd-level model
design information' button in Results tab)
(*) Modifications to second-level analysis multivariate
"F-contrast" two-step procedure to:
1) enforce orthogonal of within-subject contrasts (this aims to
improve the replicability of spm_est_non_sphericity F-threshold
masking during ReML variance estimation under arbitrary
non-rank-reducing linear transformations of the within-subject
contrast matrix; thanks to Martyn McFarquhar for bringing up this
issue; see https://www.nitrc.org/forum/message.php?... for additional details)
For back-compatibility or mid-analysis updates: in conn_process.m
and conn_module_glm.m change the line that reads "FORCEORTH=true;"
to "FORCEORTH=false;"
2) base cluster-mass statistics on aggregated F values (instead of
aggregated sqrt(F) values; this aims to provide exactly the same
cluster-mass values for clusters resulting from two-sided GLM
univariate contrasts as those resulting from GLM multivariate
contrasts when both analyses effectively test the same null
hypothesis)
For back-compatibility or mid-analysis updates: in
conn_displayroi.m conn_randomise.m and conn_vproject.m change the
line that reads "DOSCALEF=true;" to "DOSCALEF=false;"
3) changed default Wilks' Lambda statistics procedure from
Bartlett's Chi-square approximation to Rao's F approximation (this
aims to provide increased accuracy for small sample sizes)
For back-compatibility or mid-analysis updates: in conn_glm.m
change the line that reads "BBDEFAULT='BB_RAO';" to
"BBDEFAULT='BB_BARTLETT';"
(*) Modification to optional CONN outputs generated when selecting
the option 'create confound-corrected timeseries' to: 1) output one
separate NIFTI file per run/session (prepending 'd' letter to
original functional filenames); and 2) add back the average BOLD
timeseries (one value per voxel) in the resulting denoised volumes
(this is intended to maintain anatomical-feature information in the
denoised data; e.g. used by SPM implicit masking procedure)
For back-compatibility or mid-analysis updates: in conn_process.m
change the line that reads "DONEWOUTPUTCONFCORR=1;" to
"DONEWOUTPUTCONFCORR=0;"
Added 'structural_normalize_preservemasks' and
'functional_normalize_indirect_preservemasks' preprocessing steps
to keep externally-generated tissue-class maps (e.g.
FreeSurfer-generated Grey/White/CSF masks) aligned to structural
data during structural or indirect-functional normalization steps
Added 'interleaved (Philips)' slice-order option in Slice Timing
Correction preprocessing step procedure (thanks to Pravesh Parekh
for bringing up this issue and providing the documentation; slice
order in Philips "interleaved" sequences defined as
cell2mat(arrayfun(@(n)n:K:N,1:K,'uni',0)), where N is the number of
slices and K is the interleave number = round(sqrt(N))
Added 'functional_regression' preprocessing step to perform
temporal regression and removal of temporal
components-of-no-interest to functional volumes
Added 'functional_bandpass' preprocessing step to perform temporal
band-pass filtering to functional volumes
Added 'functional_vdm_apply' preprocessing steps to perform
Susceptibility Distortion Correction from VDM maps to functional
volumes (independently of realignment, unlike the existing
functional_realign&unwarp&fieldmap preprocessing step)
Added conn_module('glm',...) options for user-defined ROI-to-ROI
second-level analyses (see "doc conn_module" and "doc
conn_mtx_write")
CUMULATIVE CHANGES
__________________________________________________________________________________________________________________________________________
14 Added surface-based analyses
freesurfer compatibility
surface support for spatial voxel-to-voxel measures
permutation analyses for cluster-level stats
GUI support
(*) Modified BA.img Brodmann areas atlas (Lancaster transform to
improve MNI coregistration)
(note: to replicate results analyzed in previous versions select
original BA.img file)
Improved 3d rendering displays & options
Added option "frequency filter" to 1st-level analysis (estimates
spectral decomposition of seed-to-voxel and ROI-to-ROI
connectivity)
Added option "condition-specific frequency filter" to
Setup.Conditions (to analyze frequency-dependency of fcMRI
measures)
14.f Cumulative patch fixing GUI error conditions
Added ability to run individual spatial-preprocessing steps on
structural/functional files (Tools.SpatialPreprocessing GUI menu)
14.g Cumulative patch fixing GUI error conditions
Added subject-sorting uicontext menu to Tools.Calculator display
(*) Added option to apply despiking before or after regression
step; change default setting to 'after regression' (deprecated
despiking effect in favor of ART-based scrubbing)
(note: to replicate results analyzed in previous versions change
value to 'before regression');
Added ability to run arbitrary sequences of multiple spatial
preprocessing steps to Tools.SpatialPreprocesing
Changes to the default spatial preprocessing pipeline
Added defaultMNI (for MNI-space volume-based analyses) and
defaultSS (for subject-space surface-based analyses) default
preprocessing pipelines to preprocessing wizard and
Tools.SpatialPreprocessing
(*) Change realign to realign&unwarp in the default spatial
preprocessing pipeline (for additional control of motion-related
effects)
(note: to replicate results analyzed in previous versions select
select 'other', enter 'default MNI pipeline', and change
'functional Realignment & unwarp' to 'functional Realignment')
(*) Added ART outlier detection & scrubbing as part of the default
spatial preprocessing pipeline (for additional control of
motion-related and other BOLD signal artifacts)
(note: to replicate results analyzed in previous versions select
select 'other', enter 'default MNI pipeline', and remove
'functional Outlier detection (ART-based scrubbing)' from the
selected list of spatial preprocessing steps)
(*) Removed slice timing correction in the default spatial
preprocessing pipeline (to minimize the impact of potential outlier
scans in conjunction with ART)
(note: to replicate results analyzed in previous versions select
select 'other', enter 'default MNI pipeline', and add 'functional
Slice-timing correction' as the first step in the spatial
preprocessing list)
14.h Improve 'explore clusters' bar display (clearer
labels/ordering)
GUI 'Setup->2nd-level covariates' added ability to define new
covariate as a function of other existing covariates (e.g. easily
defining modulation/product terms)
(*) Added a detrending regressor when selecting 'Regress out
covariates' for PCA decomposition of multi-dimensional ROI signal
(note: to replicate results analyzed in previous versions change
'DETRENDBEFOREPCA=true' to 'DETRENDBEFOREPCA=false' in
conn_process.m)
Additional samples of preprocessing pipelines added to
Spatial.Preprocessing menu (e.g. using phase correction when VDM
files are available)
Added the possibility to define subject-specific TR values
14.i Cumulative patch fixing several error conditions
Spatial preprocessing steps are always prompted for user
confirmation (except batch processes) to increase awareness of the
choice of spatial preprocesing options and to give users a better
chance to adapt these to their needs
Added slice timing correction back to the spatial preprocessing
pipeline templates (reverted change introduced in 14.g to maintain
consistent behavior with previous versions and until clearer
community consensus is reached on this issue; users may consider
removing this step for relatively small TR values; Changwei W. Wu
et al. (2011) Empirical evaluations of slice-timing, smoothing, and
normalization effects in seed-based, resting-state functional
magnetic resonance imaging analyses; Smith et al. (2013)
Resting-state fMRI in the Human Connectome Project)
Added options for polynomial detrending (linear/quadratic/cubic)
during Preprocessing step
Added 'Removal of initial scans' to list of optional spatial
preprocessing steps (this step allows users to easily drop the
first few scans in each session from the analyses)
Added new first-level analysis option (for ROI-to-ROI and
seed-to-voxel analyses) : standard PsychoPhysiological Interaction
(PPI) analyses of task-related effects as well as
PhysioPhysiological Interactions (note: FSL-style implementation,
where task-effect interaction factors are hrf-convolved and linear
interactions are modeled on the resulting BOLD-level signal; c.f.
SPM-style implementation which deconvolves the BOLD timeseries
instead and interactions are modeled on the resulting
"neural"-level signal)
Added option for defining arbitrary temporal-modulation factors in
Setup.Conditions and perform PPI analyses of these factors
(pre-requisite for dynamic connectivity analyses)
Preprocessing step output files have been reorganized to avoid
duplication in the presence of multiple conditions (save storage
space / processing time; note: as a consequence when selecting
'Create confound-corrected time-series' now only one nifti file
will be created per subject -containing the entire timeseries-
instead of one file per condition as before -containing each only
those scans/samples associated with each condition-)
Added option in seed-to-voxel results explorer to specify T-value
voxel-level thresholds (note: limited-precision of associated
voxel-level p-values makes using T/F-value thresholds useful when
you wish to threshold very strong effects; e.g. p<1e-10)
14.j (internal distribution only)
14.k Cumulative patch fixing error conditions
Added option for sliding-window temporal decomposition of FC
measures (in Setup.Conditions.TimeFrequencyDecomposition)
Added ability to re-run Setup/Preprocessing steps for only a subset
of analysis types (ROI-to-ROI/Seed-to-Voxel/Voxel-to-Voxel;
user-selection when pressing 'Done')
Added ability to re-run Preprocessing/FirstLevelAnalysis steps for
only a subset of conditions (user-selection when pressing 'Done')
Added ability to use ROIs defined in fsaverage-space (e.g. from
surface-based result masks, or surface-based parcellations) (only
available when selecting structural volumes that have been
processed in FreeSurfer)
Cumulative GUI updates (less cluttered GUI layout and more detailed
help tips throughout main CONN gui (e.g. right-click on data files
to access data folders; easier access to and higher user-control of
the display options in the "ROI-to-ROI results explorer")
Added "structural/functional manual orientation" to the list of
potential spatial preprocessing steps (centers, rotates, reflects
structural/functional volumes, typically to bring these closer to
templates prior to coregistration/normalization)
Cumulative patch fixing compatibility issues with SPM12b
(*) Added support for new SPM12 normalization and segmentation
spatial preprocessing steps (new methods only available when using
SPM12)
(note: to keep using the older normalization/segmentation
procedures under SPM12 set the variable "PREFERSPM8OVERSPM12=true"
in conn_setup_preproc.m)
Added Dynamic Functional Connectivity analyses
"First-level Dynamic FC" estimates data-driven temporal-modulation
factors that characterize the observed dynamic changes in
functional connectivity (ROI-to-ROI connectivity matrix from
selected ROIs)
The resulting factors are entered into standard PPI analyses to
characterize dynamic changes in FC
Additional Seed-to-voxel or ROI-to-ROI analyses (for the selected
or other ROIs) can be computed by selecting "temporal modulation
effects (Dynamic FC)" in the first-level analysis "analysis type"
option.
Aggregated subject/condition measures are also computed to analyze
between-subject/condition differences in temporal modulation
strength for each factor
Added new sample ROI files to conn/utils/otherrois/ ("large-voxel"
MNI-space ROI parcellations covering grey-matter areas; P###.img;
###=number of ROIs from P304.img -20mm ROIs- to P2186.img -10mm
ROIs-)
14.l Cumulative patch fixing GUI error conditions (includig
window-resizing and java-related-errors in some older Matlab
versions)
Cumulative patch fixing Preprocessing issues (completed batch
commands for flexible preprocessing pipeline; added "interleaved
(Siemens)" and fixed prompt for sliceorder (asking sliceorder for
each subject) in slice-timing preprocessing step; Added prompt for
user-defined ART-thresholds in ART preprocessing step; included
centering prior to structural segmentation or normalization like
previous versions; fixed error-reporting problem in preprocessing
steps (errors occuring inside spm batch processes were not being
appropriately captured/displayed by CONN error-handling routines);
remove art_mask_temporalfiles* after art preprocessing step)
14.m (internal distribution only; CONN course)
14.n Cumulative patch fixing error conditions
(*) Default Denoising steps set to "no despiking" (deprecated
despiking effect in favor of ART-based scrubbing)
(note: to replicate analyses in previous versions change to
"despiking (after regression)")
14.o Cumulative patch, bug fixes and improvements
GUI changes (reduce clutter, more detailed help, early warnings on
incomplete Setup information, more-responsive gui, fix Mac issues,
some Matlab 2014b issues, light- and dark- background support)
Reorganize Dynamic FC analysis output into Dynamic scores
(average/baseline and variability/modulation-strength of observed
temporal modulation factors) and Dynamic loadings (ROI-to-ROI
connectivity matrices characterizing modulations in functional
connectivity)
Added new default atlas (132 ROIs) combining FSL Harvard-Oxford
atlas cortical & subcortical areas and AAL atlas cerebellar areas
(see conn/rois/atlas.info for details), as well as collapsed ROI
file with MPFC/PCC/LLP/RLP Fox ROIs (see conn/rois/dmn.info)
Added to second-level results explorer a new "Slice display" option
(display analysis results on individual slices)
Facilitate post-hoc seed-to-voxel analyses by adding to
"second-level results explorer import values&explore values"
options the ability to use alternative cluster-definition files
and/or to extract connectivity values from alternative second-level
analyses (e.g. estimate current-analysis effects within
alternative-analysis clusters, or alternative-analysis effects
within current-analysis clusters)
Facilitate longitudinal studies by adding the possibility to
specify session-specific anatomical volumes (different anatomical
volumes across sessions) as well as session-specific ROIs
(different ROI files across sessions, e.g. gray/white/CSF masks
generated when using session-specific anatomical files)
More flexible SPM import functionality by adding the possibility to
specify session-specific SPM.mat files when importing experiment
information from SPM
Added red/blue colormaps & colorbars for the display of two-sided
results in "surface display", "volume display", and "slice display"
in second-level results explorer
14.p Cumulative patch, bug fixes and improvements
15.a Cumulative patch, bug fixes and improvements
Added more flexible plot-display options to Calculator
Added selection of colorbar range to 3d-surface results renderings
Facilitate post-hoc ROI-to-ROI analyses by adding to the
"second-level results" tab a barplot display of individual effects
(as well as 'import values & display values' options, equivalent to
those for seed-to-voxel analyses).
15.b Cumulative patch, bug fixes and improvements
Added more flexible inputs for ART-based scrubbing during
preprocessing
Added support for slice-timing correction of multiband sequences
(CONN now supports entering "slice timing" sequence information
instead of "slice order")
Added support for subjects with partial acquisition of
condition-specific data (e.g. if a condition has only been acquired
on a subset of subjects, all second-level analyses involving this
condition will automatically be limited to the subset of subjects
with valid condition-specific data; see
Setup.Conditions.MissingData)
15.c Cumulative patch, bug fixes and improvements
Reorganized and added new options to Dynamic FC analyses: added
display of dynamic factors, display of ROI-to-ROI connectivity
matrix changes across time, display of dynamic vs. static
connectivity measures; added rate-of-change dynamic measure for
each factor
Added support for gPPI models (generalized PsychoPhysiological
Interaction; recommended PPI model to analyze task-related
connectivity in the presence of multiple task conditions) in
addition to sPPI models (single-condition PsychoPhysiological
Interaction)
15.d Cumulative patch, bug fixes and improvements
Added support for .annot ROI files (freesurfer parcellation files)
Added display of Denoising voxel-to-voxel histogram for all
subjects/sessions
Added conn_batch_pct function to facilitate parallel processing of
batch commands (see conn_batch_pct.m for details)
15.e Cumulative patch, bug fixes and improvements
Added native support in SPM12 of cluster-level statistics for
F-contrasts (when using SPM8 and below CONN still uses Keith
Worsley stat_thresh FMRISTATS implementation of cluster-level
statistics for F-contrasts as in previous CONN releases)
Added support for SGE (Sun Grid Engine) parallel batch processing
on computer clusters (see help conn_batch_pct)
Added Second-level covariate import/export-to-file options (GUI
Setup.Covariates tab, import/export buttons)
15.f Cumulative patch, bug fixes and improvements
Added fully integrated Grid Engine, PBS/Torque, and LSF distributed
environment parallelization options for Preprocessing, Setup,
Denoising, and Analysis steps (accessible both through GUI as well
as batch commands; see GUI Help.Documents.GridComputing,
Tools.GridSettings, and conn_batch batch.parallel options)
Facilitate searching through database of user-community
questions/answers from the NITRC Forum by adding search help
functionality to CONN GUI (see GUI Help.Search menu)
Added batch commands for Preprocessing steps (same as previous
batch.New functionality but it can now be run at any time, not only
when the project was first created; see conn_batch
Setup.preprocessing options)
Added option to import all condition information from a spreadsheet
(in Setup.Conditions 'condition tools' menu)
Added batch commands for using non-default Tissue Probability Maps
during segmentation/normalization (SPM8/SPM12) (see conn_batch
tpm_template and tpm_ngaus options)
Added support for .par first-level covariate files (FSL estimated
motion parameters)
Added explicit centering (translation to (0,0,0) coordinates) of
functional and structural data in default MNI preprocessing
pipelines
Added the option to use one-sided (negative) thresholds for
adjacency-matrix generation in graph-theory analyses (compute
graph-measures of anticorrelations networks)
Added option to drop initial scans in ART (through the GUI and in
batch.Setup.preprocessing.art_thresholds)
Added option to modify the default mask thresholding value and
number of erosion steps for CSF & WhiteMatter ROI definitions (see
conn_batch cwthreshold field)
15.g Cumulative patch, bug fixes and improvements
Added ability to compile CONN for use without a Matlab license or
environments
15.h Cumulative patch, bug fixes and improvements
Added option to extract weighted-sum BOLD timeseries from ROI files
(e.g. for probabilistic ROI definitions, or dual regression network
ROIs)
Added the possibility of defining multiple functional sets in
"Setup.Functional.Functional data for ROI-level analyses" (to
simplify using ROIs in different spaces, e.g. some ROIs in
MNI-space while others in subject-space)
Added support for 4d-nifti ROI atlas files (one volume per ROI)
Added group-ICA voxel-to-voxel analyses (spatially independent
components using two-step dimensionality reduction, FastICA for IC
estimation, and dual-regression for subject-level backprojection;
see voxel-to-voxel analysis tab)
Added non-parametric statistics for voxel-level analyses (residual
permutation/randomization tests, univariate and multivariate
models; see Setup.Options "second-level analyses" field, and
"parametric stats/non-parametric stats" option in results explorer
window)
16.a Cumulative patch, bug fixes and improvements
16.b Cumulative patch, bug fixes and improvements
Added option to include higher-order power of confound timeseries
(e.g. Friston24 movement parameters)
Added new voxel-to-voxel r-by-distance scatter plot in Denoising
tab
Added ICA display tools, sorting components (by spatial kurtosis,
skewness, spatial correlation with template, temporal frequency,
and temporal variability), and ICA-based parcellation ('summary
display' tab)
Added analyses of variability and frequency of network timeseries
in ICA ('temporal components' tab)
Added analyses of variability and frequency of modulatory
timeseries in dynamic FC ('temporal components' tab)
Added design-display plot in second-level results tab
Added ability to preview F-contrast and simultaneously display
multiple individual contrasts in seed-to-voxel second-level results
tab
Added multi-slice plots to 'slice display' gui (in results
explorer, and functional/structural/ROI tools)
Added multi-slice display to ROI-to-ROI results tab
17.a Cumulative patch, bug fixes and improvements
Updated manuals/documentation
(*) Changed default ART threshold settings, when not otherwise
defined explicitly they are now set to 'intermediate' (97th
percentile: 0.9mm 5sd); previously set to 'liberal' (99th
percentile: 2mm 9sd)
note: if updating mid-analysis and preprocessing new subjects:
define explicitly ART thresholds to maintain consistency across
subjects
(*) Fixed global scaling parameter in
GlobalCorrelation/IntrinsicConnectivity voxel-to-voxel measures
(prior versions computed average correlation coefficient across all
voxels in the volume -with voxels outside of analysis mask set to
0-; new version computes average correlation coefficients across
analysis-voxels only -disregarding voxels outside of analysis
mask-; difference between old and new behavior is a constant
scaling factor for all measures/subjects/conditions equal to
SizeOfMask/SizeOfVolume = 0.2913 for the default analysis mask;
normalized GC/ICC measures not affected by this change)
note: if updating mid-analysis and still processing new subjects:
GlobalCorrelation/IntrinsicConnectivity voxel-to-voxel analyses
that used raw measures (ie. normalization setting is unchecked)
require rerunning all subjects for consistency
(*) Added ability to define multiple sets of voxel-to-voxel and
dynamic FC analyses within the same conn project
note: if updating mid-analysis and running voxel-to-voxel analyses:
voxel-to-voxel analyses that defined multiple measures (e.g. ICA
and ICC in the same analysis) are now to be broken into separate
analyses (e.g. one analysis for ICA and another for ICC); analyses
already performed that contain multiple measures are still
supported for back-compatibility but no new such analyses can be
created
Added batch functionality for dynamic FC analyses
Added new conn/rois/networks.nii atlas with ROIs characterizing an
extended set of classical networks: Default Mode Network (4 ROIs),
SensoriMotor (2 ROIs), Visual (4 ROIs), Salience /
Cingulo-Opercular (7 ROIs), DorsalAttention (4 ROIs),
FrontoParietal / Central Executive (4 ROIs), Language (4 ROIs),
Cerebellar (2 ROIs) (all ROIs defined from CONN's ICA analyses of
HCP dataset / 497 subjects)
Added labeling of individual networks in ICA analyses, and
suggestions based on match to ICA template
Added new automatically-generated Quality Assurance measures:
during ART preprocessing and Denoising steps the following new
second-level covariates are now created: QA_MaxMotion,
QA_ValidScans, QA_InvalidScans (maximum inter-scan motion, number
of valid/invalid scans per subject); and QA_GCOR (global
correlation index per subject and condition)
Added script-it option to all
Preprocessing/Setup/Denoising/First-level analyses (for any
analysis step, in addition to running immediately this step CONN
can now also automatically create scripts for the same step to be
run at a later time)
Added masking option for group-level ICA, PCA, and MVPA analyses
(e.g. group-level components in masked-ICA are restricted to
within-mask voxels only; subject-level backprojection still
performed across entire brain / Setup.AnalysisMask)
Added 'display single-slice for all subjects' summary displays to
structural-/functional-/ROI- tools (e.g. use to quickly assess
correct normalization/coregistration across all subjects and/or
identify outlier subjects/sessions)
Added 'Motion-Mask estimation' procedure to list of available
preprocessing steps (estimating at each voxel the derivative of the
BOLD signal with respect to the 6 motion parameters; when running
this preprocessing step MotionMask ROIs are automatically added to
the list of ROIs in Setup.ROIs, which can then be used during
Denoising for additional protection against motion-related effects)
Added option to separately filter individual confound effects
before denoising regression (equivalent to 'simult' method but
allowing control over individual effects; e.g. filter
motion/CompCor regressors but not scrubbing regressors)
Added connection bundling, display of reference-slice, user-defined
surfaces, and other usability options to CONN 3d displays
17.b Cumulative patch, bug fixes and improvements
Added ability to save/share/reuse user-defined preprocessing
pipelines
Added ability to enter DICOM structural/functional files
Added Quality Assurance displays to visually inspect the quality of
the structural/functional data coregistration and Denoising step,
and easily identify potential outlier subjects or failed
preprocessing steps (see Tools 'QA plots' option)
Added ALFF/fALFF measures (see Voxel-to-Voxel first-level analysis
tab)
Added ability to bookmark individual second-level results and plots
to help organize complex sets of analyses and results
Added preprocessing options for indirect coregistration, explicit
direct/indirect normalization, and realignment without reslicing
Added higher-resolution slice- and glassbrain displays (see
seed-to-voxel or voxel-to-voxel results explorer)
Added options to display individual-subject voxel- and ROI-level
connectivity measures entered to second-level analyses (see
'display values' in second-level results tab)
Added options to import sets of files from a different CONN project
in Setup tab (see 'AltSelect' option)
Added 'description' fields to second-level covariates (see
Setup.Covariates tab), and to user-defined contrasts (see 'contrast
manager' in second-level results tab)
Added ability to switch between axial/coronal/sagittal views in
Setup tab and ICA results
17.c Cumulative patch, bug fixes and improvements
Added ability to run preprocessing steps for individual
sessions/runs (this facilitates incrementally adding new timepoints
to an existing CONN project in longitudinal analyses)
Added QA plots: QA_artifact movie display of BOLD volumes with ART
global/motion/outlier timeseries, QA_registration plots display
original or eroded white/csf masks
Added option to annotate individual QA plots
Added QA subject-level measures QA_MotionMax/Mean and
QA_GlobalMax/Mean from ART-derived framewise motion and BOLD change
timeseries
(*) Changed default MNI preprocessing pipeline (minor changes):
a) moved ART step to be performed before functional normalization
instead of after it (this allows the functional normalization step
to use a mean-functional image generated by ART which excludes
outlier scans, making coregistration/normalizatoin procedures more
robust to the presence of outliers)
b) changed 'functional_normalize' to
'functional_segment&normalize' (this generates additional
functional gray/white/csf masks to be used in QA plots or as
alternative aCompCor masks)
For back-compatibility or mid-analysis updates: the exact prior
pipeline can be loaded from:
conn/utils/preprocessingpipelines/defaultMNI_legacy.mat
Added compatibility-fix for cluster-computing options in new
SPM+toolboxes standalone ("batteries included" standalone SPM12
prerelease)
Added GUI options to easily select different functional,
structural, ROI, or first-level covariate files using filename
rules (see 'reassign all files' options in
functional/structural/roi/covariate tools menus)
Added movie displays of functional timeseries (see 'display
single-slice for all timepoints' options in functional/first-level
covariate tools menus, and 'display values' option in Denoising
tab)
Added "art_regression_timeseries*.mat" file output to ART and
associated QA_timeseries 1st-level covariate in CONN, containing
(when using default ART settings) scan-to-scan changes in global
BOLD signal (z-scores), and framewise displacement timeseries (mm)
(timeseries that, when thresholded, lead to identified outlier
scans)
Added "carpet" display of BOLD signal timeseries to Denoising tab
and QA plots (similar to Power et al 2017) to help evaluate the
amount of residual whole-brain signals before and after denoising
Added ability to edit through the GUI and batch commands the
binarization/erosion options for grey matter, white matter, and CSF
masks (see Setup.ROIs.Erosion settings; previous
batch.Setup.cwthreshold field)
Added voxel BOLD signal timeseries display to GUI (click on brain
images on Setup.functionals, Denoising, or First-level analysis
tabs)
Added voxel-level "plot effects", "plot values", and "import
values" menus to Second-level results tab (click on brain images on
Second-level results tab)
17.d (internal conn workshop release)
17.e Cumulative patch, bug fixes and improvements
Added ability to import condition-information file using conn_batch
commands (see batch.Setup.conditions.importfile info)
Added ability to load preprocessing-pipeline file using conn_batch
commands (see batch.Setup.preprocessing.steps info)
Added support for BIDS format *_events.tsv condition-information
files (see Setup.Conditions 'import conditions' option)
Added 'roi tools' option to create new spherical-ROI file from
user-defined MNI coordinates (see Setup.ROItools menu)
Additions to parallelization profiles:
User-specific parallelization profiles (user-specific profiles are
now saved in each user folder ~/)
User-defined entries in submitted shell script (see "in-file"
additional submit settings in Tools.Grid/HPC settings)
Users may now force usage of CONN-standalone in cluster nodes (see
"nodes use pre-compiled CONN only" checkbox in Tools.Grid/HPC
settings)
By default the parallelization GUI will not query the cluster
scheduler for status updates unless explicitly requested by the
user (minimize scheduler load; see "check job status automatically"
checkbox in Tools.Grid/HPC settings)
Stopped ART from displaying GUI when run in non-interactive mode
Added ability to specify user-defined activation maps in conn 3d
display (see conn 3d display menu
"Surfaces.BrainSurfaces.BrainSurfacesActivation")
Added ability to automatically copy structural/functional files
into local BIDS folder when importing structural/functional data
into CONN (e.g. for use when importing data from read-only folders)
(see "import selected files | copy first to local BIDS folder"
options in Setup.functional and Setup.structural tabs; see
batch.Setup.localcopy field for batch scripts)
17.f Cumulative patch, bug fixes and improvements
Additions to parallelization profiles:
Added ability to queue sequential parallelization jobs (when
starting a new parallelization job through the GUI while another
one is still running/pending CONN will automatically queue this new
job; when using batch scrips use the profile name 'Null profile' to
queue jobs)
Simplify operation with queued/pending parallelization jobs
(Tools.ViewPendingJobs will now automatically detect finished jobs
and offer to merge the results or submit the next set of queued
batch jobs)
Additions to standalone release
Compiled with Matlab 2017a (MCR 9.2)
Support for user- and system- level cluster configuration settings
Support for Environment Modules installation (see modulefile.txt)
Added alternative Gray/White/CSF binarization/erosion methods
(option to specify binarization threshold in absolute vs.
percentile values; option to specify erosion kernel size in abolute
vs. percentile values; see 'Erosion settings' in Setup.ROIs tab;
see batch.erosion fields for batch scripts)
Added voxelsize_anat and voxelsize_func separate batch fields for
anatomical and functional normalization/reslicing resolutions
during preprocessing (see Preprocessing GUI; see
batch.Setup.preprocessing.voxelsize_* fields)
Added functional/structural_manualspatialdef (user-defined
non-linear spatial transformations from precomupted y_*.nii or
_seg_sn.mat files) to list of available preprocessing steps (see
Preprocessing GUI; see batch.Setup.preprocessing.steps for batch
scripts)
Added ability to specify slice-timing information from BIDS-format
json files (see Preprocessing GUI; see
Setup.preprocessing.sliceorder field for batch scripts)
Added surface-projection ROI display (see ROI
tools.3d-surface-project viewer)
Added links to video tutorials and additional resources hosted on http://www.conn-toolbox.org
18.a Cumulative patch, bug fixes and improvements
Added parallelization profile for Windows 7 and Windows 10 machines
Added 'pending jobs' button to main CONN gui that alerts when there
are running, queued, or pending jobs in your cluster / HPC
environment
Added the ability to parallelize the computation of
permutation/randomization analyses (non-parametric computation of
cluster-level statistics) using Cluster/HPC jobs
Added QC reports (auto-generated html/pdf/xml reports containing a
user-defined selection of Quality Assurance plots; see 'QA plots'
button or conn_qaplotsexplore from Matlab command line)
Added QA plots computing QC-FC associations (association between
Quality Control variables and Functional Connectivity values; see
Ciric et al. 2017, NeuroImage, "Benchmarking of participant-level
confound regression strategies for the control of motion artifact
in studies of functional connectivity")
Added QA plots computing FC vs distance plots (90% CI of FC r
values vs. edge distance (in mm) in a 1000-node network)
Added QA plots computing histogram / quartiles / outliers in QC
variables (QC second-level covariates)
Fixed/Improved randomization non-parametric procedure to include
group-vs-single-subject comparisons (thanks to Stephen Larroque;
see https://www.nitrc.org/forum/forum.php?th... for additional details)
(*) Changed preprocessing voxelsize_anat default value to 1mm
(during normalization anatomical volumes are resampled to 1mm
isotropic voxelsize for improved resolution of anatomical images)
For back-compatibility or mid-analysis updates please change
anatomical resolution to 2mm
Simplified GUI for secondary functional datasets (secondary
dataset-1 and above can now also be defined, displayed, and
preprocessed from the Setup.functionals tab)
Added 'plot subjects' button to S2V and V2V results tabs, as well
as to voxel-level results explorer window (slice/surface/volume
display of individual subject connectivity maps)
Added 'export' button to ROI-to-ROI results explorer window
(exports suprathreshold connections in ROI-to-ROI connectivity
matrix)
Added interactive thresholding options to slice display
18.b Cumulative patch, bug fixes and improvements
Added conn_dcmdir (to display organization of DICOM files/folders)
and conn_dcmconvert (to organize DICOM files/folders into DICOM
series and convert each series to a nifti file)
Added richer JSON information when importing DICOM files (including
reading RepetitionTime and Siemens SliceTiming information from
DICOM headers)
Added ability to use data with different TR's across sessions/runs
(if TR value for a given subject is set to NaN -e.g. in Setup.Basic
tab-, then the TR information will be read from the
'RepetitionTime' field in a sidecar JSON file associated with the
functional volume, separately for each subject&session)
Added 'Import from BIDS dataset' GUI to import CONN Setup
information including functional/structural volumes and condition
information from BIDS-compatible datasets
Added 'Import from DICOM data files' GUI to identify and import
subjects functional/structural volumes from DICOM files
Added 'Import from SPM design files' GUI to import CONN Setup
information including functional/structural volumes and condition
information from SPM first-level design files (SPM.mat)
Added ability to run QA plots in parallel (when using the GUI see
Tools.QAplots; when using batch commands see "help conn_batch")
Added ability to label secondary functional datasets (from GUI and
during preprocessing steps using new 'functional_label' step)
Added 'functional_smooth_masked' preprocessing step (spatial
smoothing of functional data aggregating across grey matter voxels
only)
Added 'functional_surface_resample' as an explicit preprocessing
step (converts volume-level data to surface-level data, by
resampling functional volume at the location of
FreeSurfer-generated cortical surface)
Added 'functional_surface_smooth' as an explicit preprocessing step
(smooths surface-level data)
Added 'functional_vdm_create' preprocessing step (creates vdm
voxel-displacement map from double-echo fieldmap acquisition files;
magnitude1+phasediff or real1+imag1+real2+imag2 or fieldmap (Hz)
volumes)
(*) Added new 'functional_vdm_create' step by default in
'Indirect-Normalization pipeline' and 'Surface-level pipelines', so
these pipelines do not require any longer that users manually
create their vdm files (e.g. using SPM FieldMap toolbox)
For back-compatibility, or if you have already created vdm files
from your fieldmap acquisition volumes, simply remove the
'functional_vdm_create' step from your pipeline (same pipelines
without these step are also saved as *_legacy2018 in
conn/utils/preprocessingpipelines)
Fixed ability to extract 1000-node network sample timeseries (used
for "QA plots" FC histogram displays) from voxels within the
MNI-space standard gray matter mask (when the analysis mask is set
to 'explicit masking' and the mask is the default MNI-space mask),
or from voxels within the grey-matter ROI (when the analysis mask
is set to a different 'explicit' file, 'implicit' or 'none')
Added rtm option to realignment preprocesing step (choose
realign-to-mean vs. realign-to-first during preprocessing
realignment step)
Gui file 'Find' button in Setup tab now uses natural sorting
(non-numeric substrings sorted alphabetically, numeric substrings
sorted numerically)
Added Gui ctrl-F functionality (search keywords) to all listbox
menus
Added options to compute alternative/additional first-level
covariates, including alternative Framewise Displacement measures
(Power et al. 2012, Jenkinson 2002) as well as options to
re-compute the 'scrubbing' covariate using different thresholds
without needing to re-run ART preprocessing step (see 'compute
secondary covariates' option in 'covariate tools' menu; see help
conn_convertl12l1covariate for scripting options)
Added GUI options to more easily update the Denoising tab info
after changes/edits in the Setup tab (e.g. changes in condition
definitions/values, first-level covariate definitions/values, or
ROI information; see 'propagate changes' in 'condition tools',
'covariate tools', and 'roi tools', respectively)
Updates in conn_updatefilepaths for faster filename translations
when accessing CONN projects within mounted drives (where name of
mounted drive varies depending on the machine used to access the
project)
Added percentage match value displayed in QC-FC correlation plot
(comparing robserved and expected distributions of QC-FC
correlations)
Added in group-ICA analyses the ability to choose between different
FastICA algorithms G1 (tanh; default), G2 (gauss), or G3 (pow3), as
well as between different Group-ICA back-projection algorithms
GICA1, or GICA3 (default)
(*) Default group-ICA backprojection algorithm switched from GICA1
to GICA3
For back-compatibility or mid-analysis updates: go to first-level
analysis ICA tab, select an existing ICA analysis and change the
option 'G1 FastICA + GICA3 Back-projection' to 'G1 FastICA + GICA1
Back-projection' if you prefer to continue using the previous GICA1
method
Added log history functionality to save and review all screen
output associated with each CONN project (also increased the detail
of information output in several preprocessing and analysis steps
to make this log more informative)
Additions to standalone release
Compiled with Matlab 2018b (MCR 9.5) and SPM12 (7487)
Included new PC (Windows10) and Mac (macOS) releases
19.a (internal distribution only)
19.b Cumulative patch, bug fixes and improvements
Added new interface for recommended/standard as well as advanced
thresholding options in results explorer voxel-based and ROI-to-ROI
display
Unified results explorer interface and options for surface-based
and volume-based analyses
Added Threshold Free Cluster Enhancement (TFCE) option to
voxel-based second-level results (Smith & Nichols 2009)
Added Threshold Free Cluster Enhancement (TFCE) option to
ROI-to-ROI second-level results
Added Spatial Pairwise Clustering (SPC) option to ROI-to-ROI
second-level results (Zalesky et al. 2012)
Integrated second-level results explorer interface for voxel-level,
surface-level, and ROI-to-ROI second-level analyses (see "help
conn_display")
Added option to automatically import fMRIPrep-preprocessed datasets
(see Tools.Convert/Import.fromFMRIPREPdataset )
Added option to define and estimate hierarchical multi-level models
(use 'hierarchical model' in Setup.Conditions to implicitly define
a new condition as an arbitrary function/transformation of other
previously defined input conditions, e.g. to estimate connectivity
growth parameters of interest in a growth-curve longitudinal
analyses)
Improved 'Plot effects' option in seed-to-voxel and voxel-to-voxel
results explorer windows to provide effect-size information about
individual effects and conditions in current second-level analysis
within clusters of interest
Improved 'Data preview' and 'Plot effects' options in second-level
Results tab to provide effect-size information about individual
effects and conditions in current second-level analysis at each
target voxel or ROI
(*) Modified computation of Quality Control variables
'QC_MeanMotion' and 'QC_MeanGSchange' to compute average
motion/GSchange measures over valid scans only (disregarding scans
already marked as outliers)
For back-compatibility or mid-analysis updates: change the line in
conn_setup_preproc.m that reads "MEANSOVERVALIDONLY=true" to
"MEANSOVERVALIDONLY=false"
Added internal check for non-estimable contrasts when displaying
second-level results design information (part of '2nd-level model
design information' button in Results tab)
(*) Modifications to second-level analysis multivariate
"F-contrast" two-step procedure to:
1) enforce orthogonal of within-subject contrasts (this aims to
improve the replicability of spm_est_non_sphericity F-threshold
masking during ReML variance estimation under arbitrary
non-rank-reducing linear transformations of the within-subject
contrast matrix; thanks to Martyn McFarquhar for bringing up this
issue; see https://www.nitrc.org/forum/message.php?... for additional details)
For back-compatibility or mid-analysis updates: in conn_process.m
and conn_module_glm.m change the line that reads "FORCEORTH=true;"
to "FORCEORTH=false;"
2) base cluster-mass statistics on aggregated F values (instead of
aggregated sqrt(F) values; this aims to provide exactly the same
cluster-mass values for clusters resulting from two-sided GLM
univariate contrasts as those resulting from GLM multivariate
contrasts when both analyses effectively test the same null
hypothesis)
For back-compatibility or mid-analysis updates: in
conn_displayroi.m conn_randomise.m and conn_vproject.m change the
line that reads "DOSCALEF=true;" to "DOSCALEF=false;"
3) changed default Wilks' Lambda statistics procedure from
Bartlett's Chi-square approximation to Rao's F approximation (this
aims to provide increased accuracy for small sample sizes)
For back-compatibility or mid-analysis updates: in conn_glm.m
change the line that reads "BBDEFAULT='BB_RAO';" to
"BBDEFAULT='BB_BARTLETT';"
(*) Modification to optional CONN outputs generated when selecting
the option 'create confound-corrected timeseries' to: 1) output one
separate NIFTI file per run/session (prepending 'd' letter to
original functional filenames); and 2) add back the average BOLD
timeseries (one value per voxel) in the resulting denoised volumes
(this is intended to maintain anatomical-feature information in the
denoised data; e.g. used by SPM implicit masking procedure)
For back-compatibility or mid-analysis updates: in conn_process.m
change the line that reads "DONEWOUTPUTCONFCORR=1;" to
"DONEWOUTPUTCONFCORR=0;"
Added 'structural_normalize_preservemasks' and
'functional_normalize_indirect_preservemasks' preprocessing steps
to keep externally-generated tissue-class maps (e.g.
FreeSurfer-generated Grey/White/CSF masks) aligned to structural
data during structural or indirect-functional normalization steps
Added 'interleaved (Philips)' slice-order option in Slice Timing
Correction preprocessing step procedure (thanks to Pravesh Parekh
for bringing up this issue and providing the documentation; slice
order in Philips "interleaved" sequences defined as
cell2mat(arrayfun(@(n)n:K:N,1:K,'uni',0)), where N is the number of
slices and K is the interleave number = round(sqrt(N))
Added 'functional_regression' preprocessing step to perform
temporal regression and removal of temporal
components-of-no-interest to functional volumes
Added 'functional_bandpass' preprocessing step to perform temporal
band-pass filtering to functional volumes
Added 'functional_vdm_apply' preprocessing steps to perform
Susceptibility Distortion Correction from VDM maps to functional
volumes (independently of realignment, unlike the existing
functional_realign&unwarp&fieldmap preprocessing step)
Added conn_module('glm',...) options for user-defined ROI-to-ROI
second-level analyses (see "doc conn_module" and "doc
conn_mtx_write")
|
