Abstract
Recently, several deep learning methods have been applied to decoding in task-related fMRI, and their advantages have been exploited in a variety of ways. However, this paradigm is sometimes problematic, due to the difficulty of applying deep learning to high-dimensional data and small sample size conditions. The difficulties in gathering a large amount of data to develop predictive machine learning models with multiple layers from fMRI experiments with complicated designs and tasks are well-recognized. Group-level, multi-voxel pattern analysis with small sample sizes results in low statistical power and large accuracy evaluation errors; failure in such instances is ascribed to the individual variability that risks information leakage, a particular issue when dealing with a limited number of subjects. In this study, using a small-size fMRI dataset evaluating bilingual language switch in a property generation task, we evaluated the relative fit of different deep learning models, incorporating moderate split methods to control the amount of information leakage. Our results indicated that using the session shuffle split as the data folding method, along with the multichannel 2D convolutional neural network (M2DCNN) classifier, recorded the best authentic classification accuracy, which outperformed the efficiency of 3D convolutional neural network (3DCNN). In this manuscript, we discuss the tolerability of within-subject or within-session information leakage, of which the impact is generally considered small but complex and essentially unknown; this requires clarification in future studies.
Highlights
In cognitive neuroscience, the framework for predicting the stimuli given to subjects or the tasks they perform based on their neural activity is called “decoding.” From a modeling perspective, we can evaluate predictive power and identify the brain regions that are the most informative for specific stimuli or tasks
The target of the group-level Multi Voxel Pattern Analysis (MVPA) was focused on the discrimination of the conceptual categories (“mammal” versus “tool”), the language difference could result in a small degree of interference
The accuracy and p values for the three cross-validations and the four classifiers are shown in Figure 3 and Table 1
Summary
The framework for predicting the stimuli given to subjects or the tasks they perform based on their neural activity is called “decoding.” From a modeling perspective, we can evaluate predictive power and identify the brain regions that are the most informative for specific stimuli or tasks. The most widely used decoding strategy is a pattern classification method called Multi Voxel Pattern Analysis (MVPA; Cohen et al, 2017). Haxby et al (2001) showed that visual categories of stimuli can be classified based on neural activity, distributed and not clustered in small areas of the ventral temporal lobe. The feasibility of decoding has been explored using a variety of machine learning methods. These include various types of classifiers such as the logistic regressions, the Support Vector Machine, and the Gaussian Naive Bayes
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
