Surface-based Information Detection from Cortical Activity

Soon, CS; Namburi, Praneeth; Goh, Csf; Chee, Mwl; Haynes, J. M.

doi:10.1016/s1053-8119(09)70551-8

Cited by 3 publications

(5 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…While a few researchers have reported the use of surface-based information mapping approaches (Wiestler et al, 2009;Soon et al, 2009;Oosterhof et al, 2009), to our knowledge there are no reports that show that for real data this approach has advantages compared to volume-based approaches. In the present paper, we compare the volume-based and surface-based approaches in an example dataset.…”

Section: Introductionmentioning

confidence: 95%

A comparison of volume-based and surface-based multi-voxel pattern analysis

et al. 2011

View full text Add to dashboard Cite

Section: Introductionmentioning

confidence: 95%

A comparison of volume-based and surface-based multi-voxel pattern analysis

et al. 2011

View full text Add to dashboard Cite

“…Popular strategies include: preselecting voxels based on an anatomical mask [18] , or a separate functional localizer [20] , [21] ; spatial subsampling [22] ; finding informative voxels using univariate models [3] , [11] , [12] or locally multivariate searchlight methods [23] , [24] ; and unsupervised dimensionality reduction [4] , [25] . Other recently proposed strategies attempt to account for the inherent spatial structure of the feature space [23] , [26] , [27] or use voxel-wise models to infer a particular stimulus identity [28] – [30] . Finally, those submissions that performed best in the Pittsburgh Brain Activity Interpretation Competition (PBAIC 2007) highlighted the utility of kernel ridge regression [31] and relevance vector regression [31] , [32] .…”

Section: Introductionmentioning

confidence: 99%

“…The second challenge for classification methods concerns the interpretation of their results. Most classification studies to date draw conclusions from overall prediction accuracies [33] , [11] , the spatial deployment of informative voxels [19] , [34] , [18] , [35] – [39] , the temporal evolution of discriminative information [40] , [37] , [41] , [42] , [26] , or patterns of undirected regional correlations [43] . These approaches may support discriminative decisions, but they are blind to the neuronal mechanisms (such as effective connectivity or synaptic plasticity) that underlie discriminability of brain or disease states.…”

Section: Introductionmentioning

confidence: 99%

Generative Embedding for Model-Based Classification of fMRI Data

et al. 2011

View full text Add to dashboard Cite

Decoding models, such as those underlying multivariate classification algorithms, have been increasingly used to infer cognitive or clinical brain states from measures of brain activity obtained by functional magnetic resonance imaging (fMRI). The practicality of current classifiers, however, is restricted by two major challenges. First, due to the high data dimensionality and low sample size, algorithms struggle to separate informative from uninformative features, resulting in poor generalization performance. Second, popular discriminative methods such as support vector machines (SVMs) rarely afford mechanistic interpretability. In this paper, we address these issues by proposing a novel generative-embedding approach that incorporates neurobiologically interpretable generative models into discriminative classifiers. Our approach extends previous work on trial-by-trial classification for electrophysiological recordings to subject-by-subject classification for fMRI and offers two key advantages over conventional methods: it may provide more accurate predictions by exploiting discriminative information encoded in ‘hidden’ physiological quantities such as synaptic connection strengths; and it affords mechanistic interpretability of clinical classifications. Here, we introduce generative embedding for fMRI using a combination of dynamic causal models (DCMs) and SVMs. We propose a general procedure of DCM-based generative embedding for subject-wise classification, provide a concrete implementation, and suggest good-practice guidelines for unbiased application of generative embedding in the context of fMRI. We illustrate the utility of our approach by a clinical example in which we classify moderately aphasic patients and healthy controls using a DCM of thalamo-temporal regions during speech processing. Generative embedding achieves a near-perfect balanced classification accuracy of 98% and significantly outperforms conventional activation-based and correlation-based methods. This example demonstrates how disease states can be detected with very high accuracy and, at the same time, be interpreted mechanistically in terms of abnormalities in connectivity. We envisage that future applications of generative embedding may provide crucial advances in dissecting spectrum disorders into physiologically more well-defined subgroups.

show abstract

“…However, most of these methods are only loosely constrained by rules of biological plausibility. Notable exceptions are approaches that attempt to account for the inherent spatial structure of the feature space (Kriegeskorte et al, 2006; Soon, Namburi, Goh, Chee, & Haynes, 2009; Grosenick, Klingenberg, Greer, Taylor, & Knutson, 2009) or that use a model to identify a particular stimulus identity (e.g., Kay, Naselaris, Prenger, & Gallant, 2008; Mitchell et al, 2008; Formisano, De Martino, Bonte, & Goebel, 2009). However, conventional methods for feature selection may easily lead to rather arbitrary subsets of selected voxels—deemed informative by the classifier, yet not trivial to interpret physiologically.…”

Section: Introductionmentioning

confidence: 99%

“…(e.g., Kamitani & Tong, 2005, 2006; Haynes & Rees, 2005; Hampton & O’Doherty, 2007; Kriegeskorte, Formisano, Sorger, & Goebel, 2007; Grosenick, Greer, & Knutson, 2008; Hassabis et al, 2009; Howard, Plailly, Grueschow, Haynes, & Gottfried, 2009); and (iii) temporal pattern localization : when does specific information become available to a brain region? (e.g., Polyn, Natu, Cohen, & Norman, 2005; Grosenick et al, 2008; Bode & Haynes, 2009; Harrison & Tong, 2009; Soon et al, 2009). Yet, mechanistic conclusions that relate to biologically meaningful entities such as brain connectivity or synaptic plasticity are hard to draw.…”

Section: Introductionmentioning

confidence: 99%

Model-based feature construction for multivariate decoding

et al. 2011

View full text Add to dashboard Cite

Conventional decoding methods in neuroscience aim to predict discrete brain states from multivariate correlates of neural activity. This approach faces two important challenges. First, a small number of examples are typically represented by a much larger number of features, making it hard to select the few informative features that allow for accurate predictions. Second, accuracy estimates and information maps often remain descriptive and can be hard to interpret. In this paper, we propose a model-based decoding approach that addresses both challenges from a new angle. Our method involves (i) inverting a dynamic causal model of neurophysiological data in a trial-by-trial fashion; (ii) training and testing a discriminative classifier on a strongly reduced feature space derived from trial-wise estimates of the model parameters; and (iii) reconstructing the separating hyperplane. Since the approach is model-based, it provides a principled dimensionality reduction of the feature space; in addition, if the model is neurobiologically plausible, decoding results may offer a mechanistically meaningful interpretation. The proposed method can be used in conjunction with a variety of modelling approaches and brain data, and supports decoding of either trial or subject labels. Moreover, it can supplement evidence-based approaches for model-based decoding and enable structural model selection in cases where Bayesian model selection cannot be applied. Here, we illustrate its application using dynamic causal modelling (DCM) of electrophysiological recordings in rodents. We demonstrate that the approach achieves significant above-chance performance and, at the same time, allows for a neurobiological interpretation of the results.

show abstract

Surface-based Information Detection from Cortical Activity

Cited by 3 publications

References 0 publications

A comparison of volume-based and surface-based multi-voxel pattern analysis

A comparison of volume-based and surface-based multi-voxel pattern analysis

Generative Embedding for Model-Based Classification of fMRI Data

Model-based feature construction for multivariate decoding

Contact Info

Product

Resources

About