Meeting Banner
Abstract #2293

Multi-parametric brain morphometry using a big data approach

Farshid Sepehrband1, Clio Gonzalez-Zacarias2, Lu Zhao2, Arthur W Toga2, and Kristi A Clark2

1Laboratory of Neuro Imaging, Keck School of Medicine of USC, Los Angeles, CA, United States, 2Laboratory of Neuro Imaging

Many studies have explored the relationship between neuroanatomical measures (such as cortical thickness or surface area) and cognition or health. Conventionally, generalized linear models are used to identify between-group differences within single measurements (e.g., regional cortical thickness), which ignores the possible interaction between neuroanatomical features. Incorporating a large number of regressors is not recommended in regression analyses, mainly due to the curse of dimensionality [1]. Multi-parametric classification approaches can be used to ameliorate the latter issue and to capture brain complexity [2]. The down side is that these approaches have less interpretability compared to regression techniques –because these techniques primarily focus on prediction accuracy rather than building an interpretable model. Here we present an approach that enables multi-parametric regression analysis by employing big data routine.

This abstract and the presentation materials are available to members only; a login is required.

Join Here