The NeWBI Strategy

The NeWBI tutorials, the fMRI for Newbies lectures, and a new textbook (Culham J. C., & Goebel, R., in preparation) are all being arranged to cover topics in Neuroimaging of Cognition in a strategic, coordinated order.

Our approach emphasizes “hitting the ground running” with an understanding of Experimental Analysis and Design from the very first lectures. Many neuroimaging courses start with MRI physics, the BOLD signal, data, and preprocessing, which is appropriate for courses in fields like Medical Biophysics. This course is different. The target audience is users of fMRI who want to learn to use the technique to ask interesting questions in fields like Neuroscience and Psychology. For these users a strong understanding of design and analysis are often more foundational than a deep understanding of physics, which can be developed over the longer term.

Foundations

The first six tutorials cover the conceptual foundations of fMRI analysis. Tutorial 1 begins with the data and understanding how it is structured. In Tutorial 2, we dive into fMRI statistics, with an emphasis on the “swiss army knife of statistics”, the General Linear Model (GLM). We begin applying simplest statistical analyses possible — one run of one participant with two conditions in a block design — on “hot-off-the-scanner” raw (unpreprocessed) data. We examine how such data can be analyzed with a simple correlation or with a GLM (with the two approaches being statistically equivalent). We then examine how the GLM can be used to explore more complex designs with multiple conditions. In Tutorial 3, we learn how to correct for violations of statistical assumptions that occur with fMRI data. With a strong understanding of the GLM, in Tutorial 4 and Tutorial 5 we then see how GLM statistics can be improved through different preprocessing steps, motion correction and spatiotemporal filtering, respectively. In Tutorial 6, we learn how data from individual brains of different shapes and sizes can be normalized into a common space and how data from individuals can be combined to derive group statistics.

Advanced Concepts

The last four tutorials cover more advanced topics. In Tutorial 7, we learn how the timing of fMRI designs can be pushed in event-related designs. In Tutorial 8, we learn how to analyze not just the levels of activation, but also spatial patterns of activation using multivoxel pattern analysis. In Tutorial 9, we examine the two major strategies for fMRI data analysis: region-of-interest and voxelwise approaches. We also examine common pitfalls of imaging analyses. Having learned how to do hypothesis-driven analyses in the preceeding tutorials, in Tutorial 10, we examine one approach to data-driven analyses: independent component analysis.