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
. 2008 Sep 1;2008(11):97-104.
doi: 10.1901/jaba.2008.2008-97.

Prediction of Successful Memory Encoding from fMRI Data

S K Balci  1 , M R Sabuncu, J Yoo, S S Ghosh, S Whitfield-Gabrieli, J D E Gabrieli, P Golland
Affiliations

Prediction of Successful Memory Encoding from fMRI Data

S K Balci et al. Med Image Comput Comput Assist Interv. .

Abstract

In this work, we explore the use of classification algorithms in predicting mental states from functional neuroimaging data. We train a linear support vector machine classifier to characterize spatial fMRI activation patterns. We employ a general linear model based feature extraction method and use the t-test for feature selection. We evaluate our method on a memory encoding task, using participants' subjective prediction about learning as a benchmark for our classifier. We show that the classifier achieves better than random predictions and the average accuracy is close to subject's own prediction performance. In addition, we validate our tool on a simple motor task where we demonstrate an average prediction accuracy of over 90%. Our experiments demonstrate that the classifier performance depends significantly on the complexity of the experimental design and the mental process of interest.

PubMed Disclaimer

Figures

Fig. 1
Fig. 1
Left: ROC curves for the motor task for 10 subjects for classification with feature selection. Circles show the operating points corresponding to min-error classification accuracy. Right: Min-error classification accuracy for classification without feature selection (light-gray) and with feature selection (dark-gray).
Fig. 2
Fig. 2
(a) ROC curves for memory encoding experiment for 10 subjects. Crosses represent subject's prediction accuracy. Blue curves correspond to strategy 1, using the training set for feature selection. Red curves correspond to training the classifier only on correctly predicted samples (strategy 2). Green curves correspond to strategy 3, including test set in feature selection. Circles show the operating points corresponding to min-error classification accuracy. (b) Min-error classification accuracy.
Fig. 3
Fig. 3
Feature overlap maps for the best(left) and the worst(right) performing subjects for the memory encoding task. For all five functional runs feature selection is performed on each run. The color indicates the number of runs in which a voxel was selected. Dark red color shows the voxels selected only in one run and white color displays voxels selected in all runs.
See this image and copyright information in PMC

References

    1. King J, et al. Judgements of knowing: the influence of retrieval practice. Am. J. Psychol. 1980;93(2):329–343.
    1. Kao Y, Davix E, Gabrieli J. Neural correlates of actual and predicted memory formation. Nature Neuroscience. 2005;8(12):1776–1783. - PubMed
    1. Friston K, et al. Statistical parametric maps in functional imaging: A general linear approach. Human Brain Mapping. 1995;2(4):189–210.
    1. O'Toole A, et al. Theoretical, statistical, and practical perspectives on pattern-based classification approaches to the analysis of functional neuroimaging data. Journal of Cognitive Neuroscience. 2007;19(11):1735–1752. - PubMed
    1. Spiers H, Maguire E. Decoding human brain activity during real-world experiences. Trends in Cognitive Sciences. 2007;11(8):356–365. - PubMed

LinkOut - more resources