BRAIN project (Cox)

National Institute Of Mental Health

Investigators

Linked publications & trials

Abstract

Over the last year, we have developed AFNI's tests extensively to prevent regression in our codebase and provide a better guarantee of AFNI's functionality. This test infrastructure, implemented in a containerized- and cloud-environment has undergone iteration with feedback from members of the neuroimaging community and the AFNI team in order to be a robust solution, convenient for use by team members, that shares any common aspects of testing with the community via the testkraken software package. We continue to develop the way in which we package AFNI to make it easier to install across all platforms, including container and cloud contexts. One significant part of this effort has been to use the Cmake (cross-platform make) build tool and to specify the projects dependencies using the conda package manager. We are continuing to work on increasing the flexibility of statistical model specification for neuroimaging analysis in R. In order to make this a more tractable feat we have begun to refactor the code to make the software easier to use, install, maintain, and develop. The output of these efforts will be the improvements in model specification detailed in the grant proposal and an R package hosted on the CRAN repository. We have made substantial contributions toward the development of an implementation of the BIDS Statsmodels specification. Rather than extend our own interface to accommodate this, we worked directly in the Fitlins codebase (also partly funded by the BRAIN Initiative). This route provides a more integrated implementation of AFNI's unique single subject analysis for models specified in the standards JSON model notation. One added benefit is to reduce redundancy in efforts. While fitlins creates a general interface for fitting statistical models in neuroimaging, our work under the hood creates a bridge to AFNI's 3dREMLfit tool, providing all the careful decisions in neuroimaging statistical modelling and improvements to algorithmic efficiency that come along with it. An added benefit to this collaboration is that the battle-hardened workflow manager, nipype, used by fitlins allows us to use AFNI code to fit the statistical models in a manner that is reproducible, tracks provenance, and scalable. Regarding web visualization we have developed a HTML output for afni_proc.py that allows a user to assess many of the quality control details that need to be considered all in the convenient interface of a single webpage.