NEWS.md

6.0

5.0

?

4.0

estimate_template add custom smoothing add parallel over subjects

3.0

Fixes and cleaning up

2.0

Data pre-processing, for both template estimation and TemplateICA norm_BOLD replaces scale_BOLD, since now it detrends the data too. Do not scale across space by default. Instead, scale across space only where necessary (one of the regressions in dual regression, and PCA-based operations). Option to detrend. Compute scale after centering and detrending.

Dual regression * Option to normalize A matrix

Template estimation Merge estimate_template.cifti and estimate_template.nifti into estimate_template Option to denoise each scan before computing DR If using pseudo retest data, split each subject's scan after denoising and detrending, not before. Option for different variance template (non-negative) Option to obtain DR results Return parameters, including data normalization choices. * print and summary for templates

TemplateICA Merge templateICA.cifti and templateICA.nifti into templateICA Option to return DR results in estimate_template Option to skip dimension reduction Option to use parallel computing for iterating over voxels. Option to provide "xifti" or "nifti" objects directly Option to provide multiple scans For denoising: replace maxQ with Q2_max to simplify logic Fix cases where time_inds and resamp_res are provided. Return parameters, including data normalization choices. Replace subject IC variance with the standard error * summary and plot for activations

Misc Implement sign matching Refine group_ICA Refine make_mesh Remove INLA required version (bring back?)

For developers Laid some groundwork for estimating templates using >2 scans per subject: var_decomp can handle >2 scans per subject (but need to test it!) New code template calculation

1.2

Update functions that work with "xifti" objects to conserve time and memory.

1.0



Try the templateICAr package in your browser

Any scripts or data that you put into this service are public.

templateICAr documentation built on Feb. 16, 2023, 8:14 p.m.