Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
Review
. 2009 Mar;45(1 Suppl):S199-209.
doi: 10.1016/j.neuroimage.2008.11.007. Epub 2008 Nov 21.

Machine learning classifiers and fMRI: a tutorial overview

Francisco Pereira  1 , Tom Mitchell, Matthew Botvinick
Affiliations
Review

Machine learning classifiers and fMRI: a tutorial overview

Francisco Pereira et al. Neuroimage. 2009 Mar.

Abstract

Interpreting brain image experiments requires analysis of complex, multivariate data. In recent years, one analysis approach that has grown in popularity is the use of machine learning algorithms to train classifiers to decode stimuli, mental states, behaviours and other variables of interest from fMRI data and thereby show the data contain information about them. In this tutorial overview we review some of the key choices faced in using this approach as well as how to derive statistically significant results, illustrating each point from a case study. Furthermore, we show how, in addition to answering the question of 'is there information about a variable of interest' (pattern discrimination), classifiers can be used to tackle other classes of question, namely 'where is the information' (pattern localization) and 'how is that information encoded' (pattern characterization).

PubMed Disclaimer

Figures

Figure 1
Figure 1
An example where features are voxels arrayed as a row vector (left) and a dataset is matrix of such row vectors (right).
Figure 2
Figure 2
A classifier is learned from the training set, examples whose labels it can see, and used to predict labels for a test set, examples whose labels it cannot see. The predicted labels are then compared to the true labels and the accuracy of the classifier – the fraction of examples where the prediction was correct – can be computed.
Figure 3
Figure 3
left: Learning a linear classifier is equivalent to learning a line that separates examples in the two classes (vectors in a 2-voxel brain) as well as possible. right: During cross-validation each of 6 groups of examples takes a turn as the test set while the rest serve as the training set.
Figure 4
Figure 4
top: Comparison of accuracy maps obtained with 6-fold cross-validation. Each row contains eight slices from an accuracy map, inferior to superior, top of each slice is posterior. The classifiers used in each row are single voxel GNB and radius 1 neighbourhood searchlight GNB, LDA and linear SVM. On top of each slice is the number of voxels for which accuracy was deemed significant using FDR q = 0.01, which are highlighted in dark red. bottom: Classifier weights for a linear SVM classifier trained on the entire image.
See this image and copyright information in PMC

References

    1. Aguirre GK. Continuous carry-over designs for fmri. NeuroImage. 2007;35:1480–1494. - PMC - PubMed
    1. Botvinick MM, Bylsma LM. Imaging phonological representations and their similarity structure with pattern-analytic fmri. NeuroImage. 2005;26(suppl 1)
    1. Brown Lawrence D, Tony Cai T, Dasgupta Anirban. Interval estimation for a binomial proportion. Statistical Science. 1999;16:101–133.
    1. Calhoun V, Adali T, Hansen L, Larsen J, Pekar J. Ica of functional mri data: an overview. 4th international symposium on Independent Component Analysis and Blind Signal Separation (ICA 2003); 2003.
    1. Carlson TA, Schrater P, He S. Patterns of activity in the categorical representations of objects. Journal of Cognitive Neuroscience. 2003;5(15):704–717. - PubMed

Publication types