Comparison of feature selection methods based on discrimination and reliability for fMRI decoding analysis

Investigating the relationship between CSAT scores and prefrontal fNIRS signals during cognitive tasks using a quantum annealing algorithm

Academic achievement is a critical measure of intellectual ability, prompting extensive research into cognitive tasks as potential predictors. Neuroimaging technologies, such as functional near-infrared spectroscopy (fNIRS), offer insights into brain hemodynamics, allowing understanding of the link between cognitive performance and academic achievement. Herein, we explored the association between cognitive tasks and academic achievement by analyzing prefrontal fNIRS signals. A novel quantum annealer (QA) feature selection algorithm was applied to fNIRS data to identify cognitive tasks correlated with CSAT scores. Twelve features (signal mean, median, variance, peak, number of peaks, sum of peaks, range, minimum, kurtosis, skewness, standard deviation, and root mean square) were extracted from fNIRS signals at two time windows (10- and 60-s) to compare results from various feature variable conditions. The feature selection results from the QA-based and XGBoost regressor algorithms were compared to validate the former's performance. In a two-step validation process using multiple linear regression models, model fitness (adjusted R2) and model prediction error (RMSE) values were calculated. The quantum annealer demonstrated comparable performance to classical machine learning models, and specific cognitive tasks, including verbal fluency, recognition, and the Corsi block tapping task, were correlated with academic achievement. Group analyses revealed stronger associations between Tower of London and N-back tasks with higher CSAT scores. Quantum annealing algorithms have significant potential in feature selection using fNIRS data, and represents a novel research approach. Future studies should explore predictors of academic achievement and cognitive ability.

Task sub-type states decoding via group deep bidirectional recurrent neural network

Decoding brain states under different cognitive tasks from functional magnetic resonance imaging (fMRI) data has attracted great attention in the neuroimaging filed. However, the well-known temporal dependency in fMRI sequences has not been fully exploited in existing studies, due to the limited temporal-modeling capacity of the backbone machine learning algorithms and rigid training sample organization strategies upon which the brain decoding methods are built. To address these limitations, we propose a novel method for fine-grain brain state decoding, namely, group deep bidirectional recurrent neural network (Group-DBRNN) model. We first propose a training sample organization strategy that consists of a group-task sample generation module and a multiple-scale random fragment strategy (MRFS) module to collect training samples that contain rich task-relevant brain activity contrast (i.e., the comparison of neural activity patterns between different tasks) and maintain the temporal dependency. We then develop a novel decoding model by replacing the unidirectional RNNs that are widely used in existing brain state decoding studies with bidirectional stacked RNNs to better capture the temporal dependency, and by introducing a multi-task interaction layer (MTIL) module to effectively model the task-relevant brain activity contrast. Our experimental results on the Human Connectome Project task fMRI dataset (7 tasks consisting of 23 task sub-type states) show that the proposed model achieves an average decoding accuracy of 94.7% over the 23 fine-grain sub-type states. Meanwhile, our extensive interpretations of the intermediate features learned in the proposed model via visualizations and quantitative assessments of their discriminability and inter-subject alignment evidence that the proposed model can effectively capture the temporal dependency and task-relevant contrast.

ConnSearch: A framework for functional connectivity analysis designed for interpretability and effectiveness at limited sample sizes

Functional connectivity studies increasingly turn to machine learning methods, which typically involve fitting a connectome-wide classifier, then conducting post hoc interpretation analyses to identify the neural correlates that best predict a dependent variable. However, this traditional analytic paradigm suffers from two main limitations. First, even if classifiers are perfectly accurate, interpretation analyses may not identify all the patterns expressed by a dependent variable. Second, even if classifiers are generalizable, the patterns implicated via interpretation analyses may not replicate. In other words, this traditional approach can yield effective classifiers while falling short of most neuroscientists' goals: pinpointing the neural correlates of dependent variables. We propose a new framework for multivariate analysis, ConnSearch, which involves dividing the connectome into components (e.g., groups of highly connected regions) and fitting an independent model for each component (e.g., a support vector machine or a correlation-based model). Conclusions about the link between a dependent variable and the brain are based on which components yield predictive models rather than on interpretation analysis. We used working memory data from the Human Connectome Project (N = 50-250) to compare ConnSearch with four existing connectome-wide classification/interpretation methods. For each approach, the models attempted to classify examples as being from the high-load or low-load conditions (binary labels). Relative to traditional methods, ConnSearch identified neural correlates that were more comprehensive, had greater consistency with the WM literature, and better replicated across datasets. Hence, ConnSearch is well-positioned to be an effective tool for functional connectivity research.

Brain Decoding Using fMRI Images for Multiple Subjects through Deep Learning

Substantial information related to human cerebral conditions can be decoded through various noninvasive evaluating techniques like fMRI. Exploration of the neuronal activity of the human brain can divulge the thoughts of a person like what the subject is perceiving, thinking, or visualizing. Furthermore, deep learning techniques can be used to decode the multifaceted patterns of the brain in response to external stimuli. Existing techniques are capable of exploring and classifying the thoughts of the human subject acquired by the fMRI imaging data. fMRI images are the volumetric imaging scans which are highly dimensional as well as require a lot of time for training when fed as an input in the deep learning network. However, the hassle for more efficient learning of highly dimensional high-level features in less training time and accurate interpretation of the brain voxels with less misclassification error is needed. In this research, we propose an improved CNN technique where features will be functionally aligned. The optimal features will be selected after dimensionality reduction. The highly dimensional feature vector will be transformed into low dimensional space for dimensionality reduction through autoadjusted weights and combination of best activation functions. Furthermore, we solve the problem of increased training time by using Swish activation function, making it denser and increasing efficiency of the model in less training time. Finally, the experimental results are evaluated and compared with other classifiers which demonstrated the supremacy of the proposed model in terms of accuracy.

Spatial Information of Somatosensory Stimuli in the Brain: Multivariate Pattern Analysis of Functional Magnetic Resonance Imaging Data

Background

Multivoxel pattern analysis has provided new evidence on somatotopic representation in the human brain. However, the effects of stimulus modality (e.g., penetrating needle versus non-penetrating touch) and level of classification (e.g., multiclass versus binary classification) on patterns of brain activity encoding spatial information of body parts have not yet been studied. We hypothesized that performance of brain-based prediction models may vary across the types of stimuli, and neural patterns of voxels in the SI and parietal cortex would significantly contribute to the prediction of stimulated locations.

Objective

We aimed to (1) test whether brain responses to tactile stimuli could distinguish among stimulated locations on the body surface, (2) investigate whether the stimulus modality and number of classes affect classification performance, and (3) localize brain regions encoding the spatial information of somatosensory stimuli.

Methods

Fifteen healthy participants completed two functional magnetic resonance imaging (MRI) scans and were stimulated via the insertion of acupuncture needles or by non-invasive touch stimuli (5.46-sized von Frey filament). Participants received the stimuli at four different locations on the upper and lower limbs (two sites each) for 5 min while blood-oxygen-level-dependent activity (BOLD) was measured using 3-Tesla MRI. We performed multivariate pattern analysis (MVPA) using parameter estimate images of each trial for each participant and the support vector classifier (SVC) function, and the prediction accuracy and other MVPA outcomes were evaluated using stratified five-fold cross validation. We estimated the significance of the classification accuracy using a permutation test with randomly labeled training data (n = 10,000). Searchlight analysis was conducted to identify brain regions associated with significantly higher accuracy compared to predictions based on chance as obtained from a random classifier.

Results

For the four-class classification (classifying four stimulated points on the body), SVC analysis of whole-brain beta values in response to acupuncture stimulation was able to discriminate among stimulated locations (mean accuracy, 0.31; q < 0.01). The searchlight analysis found that values related to the right primary somatosensory cortex (SI) and intraparietal sulcus were significantly more accurate than those due to chance (p < 0.01). On the other hand, the same classifier did not predict stimulated locations accurately for touch stimulation (mean accuracy, 0.25; q = 0.66). For binary classification (discriminating between two stimulated body parts, i.e., the arm or leg), the SVC algorithm successfully predicted the stimulated body parts for both acupuncture (mean accuracy, 0.63; q < 0.001) and touch stimulation (mean accuracy, 0.60; q < 0.01). Searchlight analysis revealed that predictions based on the right SI, primary motor cortex (MI), paracentral gyrus, and superior frontal gyrus were significantly more accurate compared to predictions based on chance (p < 0.05).

Conclusion

Our findings suggest that the SI, as well as the MI, intraparietal sulcus, paracentral gyrus, and superior frontal gyrus, is responsible for the somatotopic representation of body parts stimulated by tactile stimuli. The MVPA approach for identifying neural patterns encoding spatial information of somatosensory stimuli may be affected by the stimulus type (penetrating needle versus non-invasive touch) and the number of classes (classification of four small points on the body versus two large body parts). Future studies with larger samples will identify stimulus-specific neural patterns representing stimulated locations, independent of subjective tactile perception and emotional responses. Identification of distinct neural patterns of body surfaces will help in improving neural biomarkers for pain and other sensory percepts in the future.