ABSTRACT
Data analysis workflows in many scientific domains have become increasingly complex and flexible. Here we assess the effect of this flexibility on the results of functional magnetic resonance imaging by asking 70 independent teams to analyse the same dataset, testing the same 9 ex-ante hypotheses1. The flexibility of analytical approaches is exemplified by the fact that no two teams chose identical workflows to analyse the data. This flexibility resulted in sizeable variation in the results of hypothesis tests, even for teams whose statistical maps were highly correlated at intermediate stages of the analysis pipeline. Variation in reported results was related to several aspects of analysis methodology. Notably, a meta-analytical approach that aggregated information across teams yielded a significant consensus in activated regions. Furthermore, prediction markets of researchers in the field revealed an overestimation of the likelihood of significant findings, even by researchers with direct knowledge of the dataset2-5. Our findings show that analytical flexibility can have substantial effects on scientific conclusions, and identify factors that may be related to variability in the analysis of functional magnetic resonance imaging. The results emphasize the importance of validating and sharing complex analysis workflows, and demonstrate the need for performing and reporting multiple analyses of the same data. Potential approaches that could be used to mitigate issues related to analytical variability are discussed.
Subject(s)
Data Analysis , Data Science/methods , Data Science/standards , Datasets as Topic , Functional Neuroimaging , Magnetic Resonance Imaging , Research Personnel/organization & administration , Brain/diagnostic imaging , Brain/physiology , Datasets as Topic/statistics & numerical data , Female , Humans , Logistic Models , Male , Meta-Analysis as Topic , Models, Neurological , Reproducibility of Results , Research Personnel/standards , SoftwareABSTRACT
Cerebral Amyloid Angiopathy (CAA) is characterized by cerebrovascular amyloid-ß accumulation leading to hallmark cortical MRI markers, such as vascular reactivity, but white matter is also affected. By studying the relationship in different disease stages of Dutch-type CAA (D-CAA), we tested the relation between vascular reactivity and microstructural white matter integrity loss. In a cross-sectional study in D-CAA, 3 T MRI was performed with Blood-Oxygen-Level-Dependent (BOLD) fMRI upon visual activation to assess vascular reactivity and diffusion tensor imaging to assess microstructural white matter integrity through Peak Width of Skeletonized Mean Diffusivity (PSMD). We assessed the relationship between BOLD parameters - amplitude, time-to-peak (TTP), and time-to-baseline (TTB) - and PSMD, with linear and quadratic regression modeling. In total, 25 participants were included (15/10 pre-symptomatic/symptomatic; mean age 36/59 y). A lowered BOLD amplitude (unstandardized ß = 0.64, 95%CI [0.10, 1.18], p = 0.02, Adjusted R2 = 0.48), was quadratically associated with increased PSMD levels. A delayed BOLD response, with prolonged TTP (ß = 8.34 × 10-6, 95%CI [1.84 × 10-6, 1.48 × 10-5], p = 0.02, Adj. R2 = 0.25) and TTB (ß = 6.57 × 10-6, 95%CI [1.92 × 10-6, 1.12 × 10-5], p = 0.008, Adj. R2 = 0.29), was linearly associated with increased PSMD. In D-CAA subjects, predominantly in the symptomatic stage, impaired cerebrovascular reactivity is related to microstructural white matter integrity loss. Future longitudinal studies are needed to investigate whether this relation is causal.