Short Course in fMRI Statistics

Faculty: Doug Greve, PhD and Mark Vangel, PhD

This short course will cover the mathematics of the analysis of fMRI, focusing on the analysis of the time series from a single voxel using the general linear model (GLM). This class emphasizes theory -- it is not a how-do-I-analyze-my-functional-data class.

2008 Courses:

No upcoming courses are scheduled

What will be covered:

Day 1, Morning: Linear algebra refresher (optional). Matrix
   operations (addition, multiplication, inversion, etc), fitting data
   to a slope and intercept using matrices.

Day 1, Afternoon: Constructing the design matrix with and without
   assuming a shape to the hemodynamic response, and the trade-offs of
   each. Using basis functions. Nuisance variables. Estimating the
   hemodynamic amplitude.

Day 2, Afternoon: Sources of noise. Correlated/colored temporal noise
  and whitening.  Statistics and testing hypotheses. Creating contrast
  matrices, t-test, F-test, false positives, false negatives.  The
  multiple comparisons problem, and some solutions.  Fixed and random
  effects group analysis (time permitting).

What will not be covered:

Motion correction, spatial smoothing, Talairaching, ROI, experimental design, software analysis packages, ICA/PCA, MRI physics, k-space reconstruction. There will not be a laboratory.

Prerequisites

Prerequisites include familiarity with the conduct and performance of fMRI experiments and some experience in performing fMRI data analysis, including a basic knowledge of matrix algebra and statistics. The course will NOT cover MR physics, but participants are expected to know the fundamental principals of how fMRI signals are generated.

This class will assume that you have a background in linear algebra, and you might not get much out of the class if you don't understand something about matrices. There is a linear algebra refresher, but it is not intended for people who have never worked with matrices before. To help you screen yourself, we have prepared the following quiz. If you have no idea how to answer these questions, then you may want to reconsider taking the class.


What is the difference between a vector and a matrix?

What is the transpose of this 2x2 matrix?
12
34

What is the sum of these two 2x2 matrices?
12
34
+
56
78

What is the product of these two matrices?
12
x
3
4

What is the trace of this matrix?
45
67

What is the inverse of this matrix?
10
02

Registration:

This course is free but limited to 40 attendees. Please use the form below to register.

No upcoming courses are scheduled - contact sladieu@nmr.mgh.harvard.edu with questions.

This educational program is jointly sponsored by the Athinoula A. Martinos Center for Structural and Functional Imaging, the HST MEMP Neuroimaging Training Program and the MGH/MIT GCRC Biomedical Imaging Core.

Contact the Webmaster