Dear colleagues,
We are pleased to announce the release of DPABI V5.0, DPABISurf V1.4, DPARSF V5.1 and BrainImageNet V1.0.
New features of DPABI_V5.0_201001 (download at http://rfmri.org/dpabi, please also update the docker file by: docker pull cgyan/dpabi):
1. New Module: BrainImageNet included! With BrainImageNet, the model can classify the sex of a participant with brain structural imaging data from anybody and any scanner with about 95% accuracy. The model fine-tuned to Alzheimer’s Disease (AD) achieved 88.4% accuracy in leave-sites-out five-fold cross-validation on the ADNI dataset and 86.1% accuracy for a direct test on an unseen independent dataset (OASIS). When directly testing this AD classifier on brain images of unseen mild cognitive impairment (MCI) patients, 63.2% who finally converted into AD were predicted as AD, versus 22.1% who did not convert into AD were predicted as AD. Predicted scores of the AD classifier showed significant correlations with severity of illness. Users can utilize our models as bases for transfer learning. Please read more details at https://www.biorxiv.org/content/10.1101/2020.08.18.256594v2.
2. DPABISurf V1.4.
2.1. fMRIPrep backend updated to version 20.2.0 (Long-Term Support).
2.2. For surface based multiple comparison correction, the method of FDR correction was added to DPABISurf. See DPABISurf_VIEW->Cluster...->Apply FDR Correction.
2.3. For surface based multiple comparison correction, the method of Monte Carlo simulation was added to DPABISurf. See DPABISurf_VIEW->Cluster...->Apply FWE (Monte Carlo Simulation) Correction.
2.4. Pre-calculated Monte Carlo Simulation tables were added for most used situations. Please see {DPABI}/StatisticalAnalysis/DPABISurf_MonteCarlo/MonteCarloTable/. Users can index the minimum cluster area for an estimated smoothness and set at DPABISurf_VIEW->Cluster...->Set Cluster Size.
3. DPARSF V5.1.
3.1. Fixed a bug in applying slice timing information from DICOM files to DPARSF settings. This bug only affected DPARSF V5.0 while not setting slice timing information (using the default) in slice timing preprocessing. Automatically applying slice timing information from DICOM files missed the unit change for SPM.
4. Scrubbing default setting changed to FD_Jenkinson, 0.2mm, no time points before or after "bad" time points.
5. PALM updated to version alpha115.
6. dcm2niix updated to version 31-March-2020 (v1.0.20200331)
7. DICOM Sorter. Now DICOM Sorter can handle files with DICOM files mixed with other kind of files when setting suffix to none.
8. Fixed a bug of "Add Image" in ROI Signals Extractor.
9. y_ANCOVA1_Multcompare_Image now is compatible for .gii.
10. Docker File: MCR files extracted for easy use in Singularity.
Best,
Chao-Gan
Chao-Gan YAN, Ph.D.
Professor, Principal Investigator
Director, International Big-Data Center for Depression Research
Director, Magnetic Resonance Imaging Research Center
Institute of Psychology, Chinese Academy of Sciences
16 Lincui Road, Chaoyang District, Beijing 100101, China
Institute of Psychology, Chinese Academy of Sciences
16 Lincui Road, Chaoyang District, Beijing 100101, China
-
Initiator
Forums