1. DPARSF V4.2_161201 released.
1.1. To let the users be more aware what kind of templates they are using, SPM Templates were included under {DPABI}/Templates/ now. If you want to USE YOUR OWN TEMPLATES, please replace the corresponding ones under this directory instead of replacing those under SPM. For example: if you are using normalize by New Segment + DARTEL, please replace {DPABI}/Templates/SPMTemplates/tpm/TPM.nii; If you are using normalize by using EPI template, please replace {DPABI}/Templates/SPMTemplates/toolbox/OldNorm/EPI.nii; If you are using normalize by using T1 image unified segmentation, please replace {DPABI}/Templates/SPMTemplates/toolbox/OldSeg/grey.nii, white.nii, and csf.nii.
1.2. DPARSF Windows version. Previously need to run as administrator to get results both with and without global signal regression (GSR). Now such limitation is removed (change mklink to copyfile).
1.3. DPARSF Advanced Edition Preprocessing for Task fMRI data: For nuisance regression, the option of “Add mean back” is now default. The mean will be added back to the residual after nuisance regression. This is useful for circumstances of ICA or taskbased analysis.
1.4. DPARSFA V4 Parameters (Default Parameters, also for The RfMRI Maps Project). For ROI signals extraction, the global signal (BrainMask_05_91x109x91.img) was added as the 1569th ROI.
2. Statistical Analysis.
2.1. Given the recent concerns regarding multiple comparison correction, especially after Eklund et al. 2016 PNAS paper, We have included permutation test in the Statistical Analysis module. The permutation test was achieved by integrating PALM package, with the kind permission by Dr. Anderson M. Winkler. Please click “Permutation test (PALM)” button on the Statistical Analysis panel to use it. Please read http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/PALM/UserGuide for the details of PALM. Please cite Winkler, A.M., Ridgway, G.R., Douaud, G., Nichols, T.E., Smith, S.M., 2016. Faster permutation inference in brain imaging. Neuroimage 141, 502516 if you used it.
2.2. AlphaSim. For the socalled “bug” — the edge effects within the mask (apply mask and then smooth), DPABI doesn’t have such an issue since DPABI_V1.2_141101. On September 17, 2014, Dr. Katharina Wittfeld reported a bug with AlphaSim for small masks in combination with high smoothness: applying mask before the Gauss filter while the Gauss filter will blur the boundaries of the masked region which will cause problems later (http://rfmri.org/content/alphasimproblemcriticalbug). The code was revised to smooth the whole bounding box first and apply mask later, and the code was distributed with DPABI_V1.2_141101. The estimation of mean and standard deviation is within the whole bounding box as well, which we believe is better than estimating in a small mask (estimation of mean and standard deviation within a small mask might be inaccurate). In addition, Dr. Robert W Cox noted: "Simulations were also repeated with the now infamously "buggy" version of 3dClustSim: the effect of the bug on FPRs was minimal (of order a few percent)." http://biorxiv.org/content/early/2016/07/26/065862.
2.3. AlphaSim. The previous GUI version can only output simulation results for corner connection (Nearest Neighbor 26). Now we output 3 versions: face (NN6), edge (NN18) and corner (NN26) connection.
2.4. Mixed Effect Analysis (withinsubject factor by betweensubject factor) was added to the Statistical Analysis module. The order of the group images should be: Group1Condition1; Group1Condition2; Group2Condition1; Group2Condition2. You will get: *_ConditionEffect_T.nii  the T values of condition differences (corresponding to the first condition minus the second condition) (WithinSubjectFactor); *_Interaction_F.nii  the F values of interaction (BetweenSubjectFactor by WithinSubjectFactor); *_Group_TwoT.nii  the T values of group differences (corresponding to the first group minus the second group), of note, the two conditions will be averaged first for each subject for Group_TwoT analysis. (BetweenSubjectFactor).
3. A new ICC version based on R was included (y_ICC_Image_LMM_CallR.m), as the previous one (y_ICC_Image_LMM.m) fails to converge in many cases. y_ICC_Image_LMM_CallR.m is based on the R code written by Dr. Ting Xu (R_Cal_ICC.R). Of note, as this one needs the users to configure R environment (install.packages("nlme") and install.packages("R.matlab”)), the DPABI_ICC_TOOL (DPABI>Utilities>TestRetest Reliability: ICC) GUI is still using y_ICC_Image_LMM.m. The new version (y_ICC_Image_LMM_CallR.m) should be used in command line.
4. DPABI Results Organizer and Intermediate Files Organizer (under “The RfMRI Maps Project”): revised the parfor loop to prevent errors in case with too many files.
5. y_T1ImgAverager.m (DPABI>Utilities>T1 Images Averager): All the NaN voxels (in some cases) are set to zero now. Previously, the NaN voxels after averaging can induce trouble in bet.
6. MATLAB 2016b compatible.
With the advances of DPABI and DPARSF, we hope researchers/users could join our effort of the RfMRI Maps project (http://rfmri.org/maps). The aim of the RfMRI Maps project is to build a big data of intrinsic brain activity indices, which has the potential to allow us addressing critical questions about the brain. We have shared a broad array of the RfMRI indices of open RfMRI data (through a standard processing pipeline built in DPABI/DPARSF), and encourage researchers share their processed RfMRI indices to public through the RfMRI Maps project. Simply download and utilize the shared data, or share your data after processing with DPABI/DPARSF’s default settings. If you have further questions, please refer to http://rfmri.org/RfMRIMapsDiscussion to discuss.
Best,
ChaoGan YAN

ChaoGan YAN, Ph.D.
Principal Investigator
Deputy Director, Magnetic Resonance Imaging Research Center
Institute of Psychology, Chinese Academy of Sciences
16 Lincui Road, Chaoyang District, Beijing 100101, China
