AUTHOR=Lamballais Sander , Muetzel Ryan L. TITLE=QDECR: A Flexible, Extensible Vertex-Wise Analysis Framework in R JOURNAL=Frontiers in Neuroinformatics VOLUME=15 YEAR=2021 URL=https://www.frontiersin.org/journals/neuroinformatics/articles/10.3389/fninf.2021.561689 DOI=10.3389/fninf.2021.561689 ISSN=1662-5196 ABSTRACT=

The cerebral cortex is fundamental to the functioning of the mind and body. In vivo cortical morphology can be studied through magnetic resonance imaging in several ways, including reconstructing surface-based models of the cortex. However, existing software for surface-based statistical analyses cannot accommodate “big data” or commonly used statistical methods such as the imputation of missing data, extensive bias correction, and non-linear modeling. To address these shortcomings, we developed the QDECR package, a flexible and extensible R package for group-level statistical analysis of cortical morphology. QDECR was written with large population-based epidemiological studies in mind and was designed to fully utilize the extensive modeling options in R. QDECR currently supports vertex-wise linear regression. Design matrix generation can be done through simple, familiar R formula specification, and includes user-friendly extensions for R options such as polynomials, splines, interactions and other terms. QDECR can handle unimputed and imputed datasets with thousands of participants. QDECR has a modular design, and new statistical models can be implemented which utilize several aspects from other generic modules which comprise QDECR. In summary, QDECR provides a framework for vertex-wise surface-based analyses that enables flexible statistical modeling and features commonly used in population-based and clinical studies, which have until now been largely absent from neuroimaging research.