Robert W. Cox

The paper introduces FMRI quality control tools in AFNI, providing systematic, in-depth, and interactive QC features for preprocessing pipelines.

The paper identifies information loss sources in neuroimaging and proposes a Bayesian framework to improve inference and reproducibility.

The paper examines how balancing trial and subject sample sizes affects statistical efficiency in neuroimaging, finding trial sample size has a comparable impact to subject sample size.

This paper introduces a hierarchical modeling framework to assess test-retest reliability, showing higher reliability estimates compared to conventional methods and highlighting the need for large trial numbers.

This paper introduces multilevel smoothing splines for nonlinear analysis in neuroimaging, demonstrating their application to longitudinal structural MRI data.

This paper introduces a Bayesian multilevel model framework to account for cross-trial variability in fMRI data, showing improved statistical sensitivity and accuracy compared to traditional methods.

The paper develops and validates age-specific Indian brain templates and atlases for spatial normalization of MRI data across ages 6-60 years.

The paper shows that different analysis workflows by 70 teams on the same fMRI dataset led to varied results, but meta-analysis found consensus in activated regions.

This paper introduces a Bayesian multilevel framework for inter-subject correlation analysis in naturalistic scanning data, improving inference efficiency and reproducibility.

The paper proposes a Bayesian multilevel approach to address multiple testing in neuroimaging by embracing multiplicity through global calibration, aiming to improve model efficiency and reproducibility.

The paper introduces a Bayesian multilevel modeling approach for neuroimaging connectomics that integrates region, region pairs, and subject analyses to improve statistical inference and avoid multiple testing issues.

ETAC improves FMRI clustering by balancing false positive rates across sub-tests, reducing reliance on arbitrary parameters and improving detection power.

This paper introduces Bayesian multilevel modeling to address inefficiency and over-penalization in neuroimaging by improving error control and inference efficiency.

The paper shows that using two one-sided t-tests without correction doubles the false positive rate compared to a single two-sided test.

The paper comments on and corrects a previous study comparing FMRI analysis software, focusing on AFNI's processing pipeline and its implementation.

The paper improves ICC modeling for neuroimaging by introducing four new approaches and recommends their use based on data characteristics.