Thought experiment: Decoding cognitive processes from the fMRI data of one individual

Abstract

Cognitive processes, such as the generation of language, can be mapped onto the brain using fMRI. These maps can in turn be used for decoding the respective processes from the brain activation patterns. Given individual variations in brain anatomy and organization, analyzes on the level of the single person are important to improve our understanding of how cognitive processes correspond to patterns of brain activity. They also allow to advance clinical applications of fMRI, because in the clinical setting making diagnoses for single cases is imperative. In the present study, we used mental imagery tasks to investigate language production, motor functions, visuo-spatial memory, face processing, and resting-state activity in a single person. Analysis methods were based on similarity metrics, including correlations between training and test data, as well as correlations with maps from the NeuroSynth meta-analysis. The goal was to make accurate predictions regarding the cognitive domain (e.g. language) and the specific content (e.g. animal names) of single 30-second blocks. Four teams used the dataset, each blinded regarding the true labels of the test data. Results showed that the similarity metrics allowed to reach the highest degrees of accuracy when predicting the cognitive domain of a block. Overall, 23 of the 25 test blocks could be correctly predicted by three of the four teams. Excluding the unspecific rest condition, up to 10 out of 20 blocks could be successfully decoded regarding their specific content. The study shows how the information contained in a single fMRI session and in each of its single blocks can allow to draw inferences about the cognitive processes an individual engaged in. Simple methods like correlations between blocks of fMRI data can serve as highly reliable approaches for cognitive decoding. We discuss the implications of our results in the context of clinical fMRI applications, with a focus on how decoding can support functional localization.

Volume-wise analysis of fMRI time series allows accurate prediction of language lateralization

“…Also, it is important to note that the presented approach, while using volume-wise information, cannot work when a patient provides only a single volume of data. The z-scoring of the voxels requires a full time series (the longer, the better) and a control condition against which the relative increase in language activity can be compared (Wegrzyn et al, 2018). As the z-scored values of each volume depend on the values of all the other volumes, it is necessary that the patient alternates between the language and comparison condition in accordance with the design.…”

“…This would mean that each volume of a patient's fMRI dataset would have to be diagnosed separately and the overall agreement between volumes could serve as a measure of uncertainty. Previous work indicates that this could be feasible, given that language processing can be distinguished from other tasks with near-perfect accuracies using short blocks of an individual person's fMRI (Wegrzyn et al, 2018). In the present work, we further explore how shifting from a "voxelwise" to a "volume-wise" analysis could complement established methods of language lateralization.…”

“…To account for the delay in the hemodynamic response function (HRF), the onsets of each block were shifted by two volumes (6 seconds) with respect of the actual timing of events. Such a simple boxcar function does not take into account the shape of the HRF, but has been shown to be effective (Wegrzyn et al, 2018). Not weighting each volume by its expected response was also important because the subsequent steps of training and testing the classifier were designed to treat all 200 volumes as independent observations of equal importance.…”

An Event Based Topic Learning Pipeline for Neuroimaging Literature Mining

“…Neuroimaging text mining is to extract knowledge from neuroimaging text [1] and has received widespread attention. In neuroinformatics and computational neuroscience, Shardlow et al [2] combined active learning and deep learning to recognize various neuroscience entities for curating research information in computational neuroscience.…”