Embedding Anatomical or Functional Knowledge in Whole-Brain Multiple Kernel Learning Models

Academic-related stressors predict depressive symptoms in graduate students: A machine learning study

BACKGROUND

Graduate students face higher depression rates worldwide, which were further exacerbated during the COVID-19 pandemic. This study employed a machine learning approach to predict depressive symptoms using academic-related stressors.

METHODS

We surveyed students across four graduate programs at a Federal University in Brazil between October 15, 2021, and March 26, 2022, when most activities were restricted to taking place online due to the pandemic. Through an online self-reported screening, participants rated ten academic stressors and completed the Patient Health Questionnaire (PHQ-9). Machine learning analysis tested whether the stressors would predict depressive symptoms. Gender, age, and race and ethnicity were used as covariates in the predictive model.

RESULTS

Participants (n=172), 67.4 % women, mean age: 28.0 (SD: 4.53) fully completed the online questionnaires. The machine learning approach, employing an epsilon-insensitive support vector regression (Ɛ-SVR) with a k-fold (k=5) cross-validation strategy, effectively predicted depressive symptoms (r=0.51; R2=0.26; NMSE=0.79; all p=0.001). Among the academic stressors, those that made the greatest contribution to the predictive model were "fear and worry about academic performance", "financial difficulties", "fear and worry about academic progress and plans", and "fear and worry about academic deadlines".

CONCLUSIONS

This study highlights the vulnerability of graduate students to depressive symptoms caused by academic-related stressors during the COVID-19 pandemic through an artificial intelligence methodology. These findings have the potential to guide policy development to create intervention programs and public health initiatives targeted towards graduate students.

Direction and velocity kinematic features of point-light displays grasping actions are differentially coded within the action observation network

The processing of kinematic information embedded in observed actions is an essential ability for understanding others' behavior. Previous research showed that the action observation network (AON) may encode some action kinematic features. However, our understanding of how direction and velocity are encoded within the AON is still limited. In this study, we employed event-related fMRI to investigate the neural substrates specifically activated during observation of hand grasping actions presented as point-light displays, performed with different directions (right, left) and velocities (fast, slow). Twenty-three healthy adult participants took part in the study. To identify brain regions differentially recruited by grasping direction and velocity, univariate and multivariate pattern analysis (MVPA) were performed. The results of univariate analysis demonstrate that direction is encoded in occipito-temporal and posterior visual areas, while velocity recruits lateral occipito-temporal, superior parietal and intraparietal areas. Results of MVPA further show: a) a significant decoding accuracy of both velocity and direction at the network level; b) the possibility to decode within lateral occipito-temporal and parietal areas both direction and velocity; c) a contribution of bilateral premotor areas to velocity decoding models. These results indicate that posterior parietal nodes of the AON are mainly involved in coding grasping direction and that premotor regions are crucial for coding grasping velocity, while lateral occipito-temporal cortices play a key role in encoding both parameters. The current findings could have implications for observational-based rehabilitation treatments of patients with motor disorders and artificial intelligence-based hand action recognition models.

Migraine aura discrimination using machine learning: an fMRI study during ictal and interictal periods

Functional magnetic resonance imaging (fMRI) studies on migraine with aura are challenging due to the rarity of patients with triggered cases. This study optimized methodologies to explore differences in ictal and interictal spatiotemporal activation patterns based on visual stimuli using fMRI in two patients with unique aura triggers. Both patients underwent separate fMRI sessions during the ictal and interictal periods. The Gaussian Process Classifier (GPC) was used to differentiate these periods by employing a machine learning temporal embedding approach and spatiotemporal activation patterns based on visual stimuli. When restricted to visual and occipital regions, GPC had an improved performance, with accuracy rates for patients A and B of roughly 86-90% and 77-81%, respectively (p < 0.01). The algorithm effectively differentiated visual stimulation and rest periods and identified times when aura symptoms manifested, as evident from the varying predicted probabilities in the GPC models. These findings contribute to our understanding of the role of visual processing and brain activity patterns in migraine with aura and the significance of temporal embedding techniques in examining aura phenomena. This finding has implications for diagnostic tools and therapeutic techniques, especially for patients suffering from aura symptoms.

Abnormalities of regional brain activity and executive function in patients with temporal lobe epilepsy: A cross-sectional and longitudinal resting-state functional MRI study

PURPOSE

We decided to track changes in regional brain activity and executive function in temporal lobe epilepsy (TLE) patients based on cross-sectional and longitudinal designs and sought potential imaging features for follow-up observation.

METHODS

Thirty-two TLE patients and thirty-three healthy controls (HCs) were recruited to detect changes in fractional amplitude of low-frequency fluctuation (fALFF) and regional homogeneity (ReHo) and to evaluate executive function both at baseline and at two-year (23.3 ± 8.3 months) follow-up. Moreover, multivariate pattern analysis (MVPA) was used for follow-up observation.

RESULTS

TLE patients displayed lower fALFF values in the right superior frontal gyrus (SFG) and higher ReHo values in the left putamen (PUT) relative to the HCs. Longitudinal analysis revealed that TLE patients at follow-up exhibited higher fALFF values in the left postcentral gyrus (PoCG), higher ReHo values in the left PoCG and the right middle frontal gyrus (MFG), lower ReHo values in the bilateral PUT and the right fusiform gyrus (FFG) compared with these patients at baseline. The executive function was impaired in TLE patients but didn't deteriorate over time. No correlations were discovered between regional brain activity and executive function. The MVPA based on ReHo performed well in differentiating the follow-up group from the baseline group.

CONCLUSION

We revealed the abnormalities in regional brain activity and executive function as well as their longitudinal trends in TLE patients. The ReHo might be a good imaging feature for follow-up observation.

Exploring intricate connectivity patterns for cognitive functioning and neurological disorders: incorporating frequency-domain NC method into fMRI analysis

This study extends the application of the frequency-domain new causality method to functional magnetic resonance imaging analysis. Strong causality, weak causality, balanced causality, cyclic causality, and transitivity causality were constructed to simulate varying degrees of causal associations among multivariate functional-magnetic-resonance-imaging blood-oxygen-level-dependent signals. Data from 1,252 groups of individuals with different degrees of cognitive impairment were collected. The frequency-domain new causality method was employed to construct directed efficient connectivity networks of the brain, analyze the statistical characteristics of topological variations in brain regions related to cognitive impairment, and utilize these characteristics as features for training a deep learning model. The results demonstrated that the frequency-domain new causality method accurately detected causal associations among simulated signals of different degrees. The deep learning tests also confirmed the superior performance of new causality, surpassing the other three methods in terms of accuracy, precision, and recall rates. Furthermore, consistent significant differences were observed in the brain efficiency networks, where several subregions defined by the multimodal parcellation method of Human Connectome Project simultaneously appeared in the topological statistical results of different patient groups. This suggests a significant association between these fine-grained cortical subregions, driven by multimodal data segmentation, and human cognitive function, making them potential biomarkers for further analysis of Alzheimer's disease.

Individual differences in the evaluation of ambiguous visual and auditory threat-related expressions

This study investigated the neural correlates of the judgement of auditory and visual ambiguous threat-related information, and the influence of state anxiety on this process. Healthy subjects were scanned using a fast, high-resolution functional magnetic resonance imaging (fMRI) multiband sequence while they performed a two-alternative forced-choice emotion judgement task on faces and vocal utterances conveying explicit anger or fear, as well as ambiguous ones. Critically, the latter was specific to each subject, obtained through a morphing procedure and selected prior to scanning following a perceptual decision-making task. Behavioural results confirmed a greater task-difficulty for subject-specific ambiguous stimuli and also revealed a judgement bias for visual fear, and, to a lesser extent, for auditory anger. Imaging results showed increased activity in regions of the salience and frontoparietal control networks (FPCNs) and deactivation in areas of the default mode network for ambiguous, relative to explicit, expressions. In contrast, the right amygdala (AMG) responded more strongly to explicit stimuli. Interestingly, its response to the same ambiguous stimulus depended on the subjective judgement of the expression. Finally, we found that behavioural and neural differences between ambiguous and explicit expressions decreased as a function of state anxiety scores. Taken together, our results show that behavioural and brain responses to emotional expressions are determined not only by emotional clarity but also modality and the subjects' subjective perception of the emotion expressed, and that some of these responses are modulated by state anxiety levels.

Characterizing social and cognitive EEG-ERP through multiple kernel learning

EEG-ERP social-cognitive studies with healthy populations commonly fail to provide significant evidence due to low-quality data and the inherent similarity between groups. We propose a multiple kernel learning-based approach to enhance classification accuracy while keeping the traceability of the features (frequency bands or regions of interest) as a linear combination of kernels. These weights determine the relevance of each source of information, which is crucial for specialists. As a case study, we classify healthy ex-combatants of the Colombian armed conflict and civilians through a cognitive valence recognition task. Although previous works have shown accuracies below 80% with these groups, our proposal achieved an F1 score of 98%, revealing the most relevant bands and brain regions, which are the base for socio-cognitive trainings. With this methodology, we aim to contribute to standardizing EEG analyses and enhancing their statistics.

Enhanced Mirror Neuron Network Activity and Effective Connectivity during Live Interaction Among Female Subjects

Facial expressions are indispensable in daily human communication. Previous neuroimaging studies investigating facial expression processing have presented pre-recorded stimuli and lacked live face-to-face interaction. Our paradigm alternated between presentations of real-time model performance and pre-recorded videos of dynamic facial expressions to participants. Simultaneous functional magnetic resonance imaging (fMRI) and facial electromyography activity recordings, as well as post-scan valence and arousal ratings were acquired from 44 female participants. Live facial expressions enhanced the subjective valence and arousal ratings as well as facial muscular responses. Live performances showed greater engagement of the right posterior superior temporal sulcus (pSTS), right inferior frontal gyrus (IFG), right amygdala and right fusiform gyrus, and modulated the effective connectivity within the right mirror neuron system (IFG, pSTS, and right inferior parietal lobule). A support vector machine algorithm could classify multivoxel activation patterns in brain regions involved in dynamic facial expression processing in the mentalizing networks (anterior and posterior cingulate cortex). These results indicate that live social interaction modulates the activity and connectivity of the right mirror neuron system and enhances spontaneous mimicry, further facilitating emotional contagion.

Treatment Response Prediction for Major Depressive Disorder Patients via Multivariate Pattern Analysis of Thalamic Features

Antidepressant treatment, as an important method in clinical practice, is not suitable for all major depressive disorder (MDD) patients. Although magnetic resonance imaging (MRI) studies have found thalamic abnormalities in MDD patients, it is not clear whether the features of the thalamus are suitable to serve as predictive aids for treatment responses at the individual level. Here, we tested the predictive value of gray matter density (GMD), gray matter volume (GMV), amplitude of low-frequency fluctuations (ALFF), and fractional ALFF (fALFF) of the thalamus using multivariate pattern analysis (MVPA). A total of 74 MDD patients and 44 healthy control (HC) subjects were recruited. Thirty-nine MDD patients and 35 HC subjects underwent scanning twice. Between the two scanning sessions, patients in the MDD group received selective serotonin reuptake inhibitor (SSRI) treatment for 3-month, and HC group did not receive any treatment. Gaussian process regression (GPR) was trained to predict the percentage decrease in the Hamilton Depression Scale (HAMD) score after treatment. The percentage decrease in HAMD score after SSRI treatment was predicted by building GPRs trained with baseline thalamic data. The results showed significant correlations between the true percentage of HAMD score decreases and predictions (p < 0.01, r² = 0.11) in GPRs trained with GMD. We did not find significant correlations between the true percentage of HAMD score decreases and predictions in GMV (p = 0.16, r² = 0.00), ALFF (p = 0.125, r² = 0.00), and fALFF (p = 0.485, r² = 0.10). Our results suggest that GMD of the thalamus has good potential as an aid in individualized treatment response predictions of MDD patients.

Structural Features Related to Affective Instability Correctly Classify Patients With Borderline Personality Disorder. A Supervised Machine Learning Approach

Previous morphometric studies of Borderline Personality Disorder (BPD) reported inconsistent alterations in cortical and subcortical areas. However, these studies have investigated the brain at the voxel level using mass univariate methods or region of interest approaches, which are subject to several artifacts and do not enable detection of more complex patterns of structural alterations that may separate BPD from other clinical populations and healthy controls (HC). Multiple Kernel Learning (MKL) is a whole-brain multivariate supervised machine learning method able to classify individuals and predict an objective diagnosis based on structural features. As such, this method can help identifying objective biomarkers related to BPD pathophysiology and predict new cases. To this aim, we applied MKL to structural images of patients with BPD and matched HCs. Moreover, to ensure that results are specific for BPD and not for general psychological disorders, we also applied MKL to BPD against a group of patients with bipolar disorder, for their similarities in affective instability. Results showed that a circuit, including basal ganglia, amygdala, and portions of the temporal lobes and of the orbitofrontal cortex, correctly classified BPD against HC (80%). Notably, this circuit positively correlates with the affective sector of the Zanarini questionnaire, thus indicating an involvement of this circuit with affective disturbances. Moreover, by contrasting BPD with BD, the spurious regions were excluded, and a specific circuit for BPD was outlined. These results support that BPD is characterized by anomalies in a cortico-subcortical circuit related to affective instability and that this circuit discriminates BPD from controls and from other clinical populations.

Structural Features Predict Sexual Trauma and Interpersonal Problems in Borderline Personality Disorder but Not in Controls: A Multi-Voxel Pattern Analysis

Child trauma plays an important role in the etiology of Bordeline Personality Disorder (BPD). Of all traumas, sexual trauma is the most common, severe and most associated with receiving a BPD diagnosis when adult. Etiologic models posit sexual abuse as a prognostic factor in BPD. Here we apply machine learning using Multiple Kernel Regression to the Magnetic Resonance Structural Images of 20 BPD and 13 healthy control (HC) to see whether their brain predicts five sources of traumas: sex abuse, emotion neglect, emotional abuse, physical neglect, physical abuse (Child Trauma Questionnaire; CTQ). We also applied the same analysis to predict symptom severity in five domains: affective, cognitive, impulsivity, interpersonal (Zanarini Rating Scale for Borderline Personality Disorder; Zan-BPD) for BPD patients only. Results indicate that CTQ sexual trauma is predicted by a set of areas including the amygdala, the Heschl area, the Caudate, the Putamen, and portions of the Cerebellum in BPD patients only. Importantly, interpersonal problems only in BPD patients were predicted by a set of areas including temporal lobe and cerebellar regions. Notably, sexual trauma and interpersonal problems were not predicted by structural features in matched healthy controls. This finding may help elucidate the brain circuit affected by traumatic experiences and connected with interpersonal problems BPD suffer from.

Prognosis of a second clinical event from baseline MRI in patients with a CIS: a multicenter study using a machine learning approach

PURPOSE

To predict the occurrence of a second clinical event in patients with a CIS suggestive of MS, from baseline magnetic resonance imaging (MRI), by means of a pattern recognition approach.

METHODS

Two hundred sixty-six patients with a CIS were recruited from four participating centers. Over a follow-up of 3 years, 130 patients had a second clinical episode and 136 did not. Grey matter and white matter T1-hypointensities masks segmented from 3D T1-weighted images acquired on 3 T scanners were used as features for the classification approach. Differences between CIS that remained CIS and those that developed a second event were assessed at a global level and at a regional level, arranging the regions according to their contribution to the classification model.

RESULTS

All classification metrics were around or even below 50% for both global and regional approaches. Accuracies did not change when T1-hypointensity maps were added to the model; just the specificity was increased up to 80%. Among the 30 regions with the largest contribution, 26 were grey matter and 4 were white matter regions. For grey matter, regions contributing showed either a larger or a smaller volume in the group of patients that remained CIS, compared to those with a second event. The volume of T1-hypointensities was always larger for the group that presented a second event.

CONCLUSIONS

Prediction of a second clinical event in CIS patients from baseline MRI seems to present a highly heterogeneous pattern, leading to very low classification accuracies. Adding the T1-hypointensity maps does not seem to improve the accuracy of the classification model.

Differential mechanisms of posterior cingulate cortex downregulation and symptom decreases in posttraumatic stress disorder and healthy individuals using real-time fMRI neurofeedback

BACKGROUND

Intrinsic connectivity networks, including the default mode network (DMN), are frequently disrupted in individuals with posttraumatic stress disorder (PTSD). The posterior cingulate cortex (PCC) is the main hub of the posterior DMN, where the therapeutic regulation of this region with real-time fMRI neurofeedback (NFB) has yet to be explored.

METHODS

We investigated PCC downregulation while processing trauma/stressful words over 3 NFB training runs and a transfer run without NFB (total n = 29, PTSD n = 14, healthy controls n = 15). We also examined the predictive accuracy of machine learning models in classifying PTSD versus healthy controls during NFB training.

RESULTS

Both the PTSD and healthy control groups demonstrated reduced reliving symptoms in response to trauma/stressful stimuli, where the PTSD group additionally showed reduced symptoms of distress. We found that both groups were able to downregulate the PCC with similar success over NFB training and in the transfer run, although downregulation was associated with unique within-group decreases in activation within the bilateral dmPFC, bilateral postcentral gyrus, right amygdala/hippocampus, cingulate cortex, and bilateral temporal pole/gyri. By contrast, downregulation was associated with increased activation in the right dlPFC among healthy controls as compared to PTSD. During PCC downregulation, right dlPFC activation was negatively correlated to PTSD symptom severity scores and difficulties in emotion regulation. Finally, machine learning algorithms were able to classify PTSD versus healthy participants based on brain activation during NFB training with 80% accuracy.

CONCLUSIONS

This is the first study to investigate PCC downregulation with real-time fMRI NFB in both PTSD and healthy controls. Our results reveal acute decreases in symptoms over training and provide converging evidence for EEG-NFB targeting brain networks linked to the PCC.

An empirical evaluation of functional alignment using inter-subject decoding

Inter-individual variability in the functional organization of the brain presents a major obstacle to identifying generalizable neural coding principles. Functional alignment-a class of methods that matches subjects' neural signals based on their functional similarity-is a promising strategy for addressing this variability. To date, however, a range of functional alignment methods have been proposed and their relative performance is still unclear. In this work, we benchmark five functional alignment methods for inter-subject decoding on four publicly available datasets. Specifically, we consider three existing methods: piecewise Procrustes, searchlight Procrustes, and piecewise Optimal Transport. We also introduce and benchmark two new extensions of functional alignment methods: piecewise Shared Response Modelling (SRM), and intra-subject alignment. We find that functional alignment generally improves inter-subject decoding accuracy though the best performing method depends on the research context. Specifically, SRM and Optimal Transport perform well at both the region-of-interest level of analysis as well as at the whole-brain scale when aggregated through a piecewise scheme. We also benchmark the computational efficiency of each of the surveyed methods, providing insight into their usability and scalability. Taking inter-subject decoding accuracy as a quantification of inter-subject similarity, our results support the use of functional alignment to improve inter-subject comparisons in the face of variable structure-function organization. We provide open implementations of all methods used.

Decoding Color Visual Working Memory from EEG Signals Using Graph Convolutional Neural Networks

Color has an important role in object recognition and visual working memory (VWM). Decoding color VWM in the human brain is helpful to understand the mechanism of visual cognitive process and evaluate memory ability. Recently, several studies showed that color could be decoded from scalp electroencephalogram (EEG) signals during the encoding stage of VWM, which process visible information with strong neural coding. Whether color could be decoded from other VWM processing stages, especially the maintaining stage which processes invisible information, is still unknown. Here, we constructed an EEG color graph convolutional network model (ECo-GCN) to decode colors during different VWM stages. Based on graph convolutional networks, ECo-GCN considers the graph structure of EEG signals and may be more efficient in color decoding. We found that (1) decoding accuracies for colors during the encoding, early, and late maintaining stages were 81.58%, 79.36%, and 77.06%, respectively, exceeding those during the pre-stimuli stage (67.34%), and (2) the decoding accuracy during maintaining stage could predict participants' memory performance. The results suggest that EEG signals during the maintaining stage may be more sensitive than behavioral measurement to predict the VWM performance of human, and ECo-GCN provides an effective approach to explore human cognitive function.

Multimodal MR Images-Based Diagnosis of Early Adolescent Attention-Deficit/Hyperactivity Disorder Using Multiple Kernel Learning

Attention-deficit/hyperactivity disorder (ADHD) is one of the most common brain diseases among children. The current criteria of ADHD diagnosis mainly depend on behavior analysis, which is subjective and inconsistent, especially for children. The development of neuroimaging technologies, such as magnetic resonance imaging (MRI), drives the discovery of brain abnormalities in structure and function by analyzing multimodal neuroimages for computer-aided diagnosis of brain diseases. This paper proposes a multimodal machine learning framework that combines the Boruta based feature selection and Multiple Kernel Learning (MKL) to integrate the multimodal features of structural and functional MRIs and Diffusion Tensor Images (DTI) for the diagnosis of early adolescent ADHD. The rich and complementary information of the macrostructural features, microstructural properties, and functional connectivities are integrated at the kernel level, followed by a support vector machine classifier for discriminating ADHD from healthy children. Our experiments were conducted on the comorbidity-free ADHD subjects and covariable-matched healthy children aged 9-10 chosen from the Adolescent Brain and Cognitive Development (ABCD) study. This paper is the first work to combine structural and functional MRIs with DTI for early adolescents of the ABCD study. The results indicate that the kernel-level fusion of multimodal features achieves 0.698 of AUC (area under the receiver operating characteristic curves) and 64.3% of classification accuracy for ADHD diagnosis, showing a significant improvement over the early feature fusion and unimodal features. The abnormal functional connectivity predictors, involving default mode network, attention network, auditory network, and sensorimotor mouth network, thalamus, and cerebellum, as well as the anatomical regions in basal ganglia, are found to encode the most discriminative information, which collaborates with macrostructure and diffusion alterations to boost the performances of disorder diagnosis.

Individualized diagnosis of major depressive disorder via multivariate pattern analysis of thalamic sMRI features

BACKGROUND

Magnetic resonance imaging (MRI) studies have found thalamic abnormalities in major depressive disorder (MDD). Although there are significant differences in the structure and function of the thalamus between MDD patients and healthy controls (HCs) at the group level, it is not clear whether the structural and functional features of the thalamus are suitable for use as diagnostic prediction aids at the individual level. Here, we were to test the predictive value of gray matter density (GMD), gray matter volume (GMV), amplitude of low-frequency fluctuations (ALFF), and fractional amplitude of low-frequency fluctuations (fALFF) in the thalamus using multivariate pattern analysis (MVPA).

METHODS

Seventy-four MDD patients and 44 HC subjects were recruited. The Gaussian process classifier (GPC) was trained to separate MDD patients from HCs, Gaussian process regression (GPR) was trained to predict depression scores, and Multiple Kernel Learning (MKL) was applied to explore the contribution of each subregion of the thalamus.

RESULTS

The primary findings were as follows: [1] The balanced accuracy of the GPC trained with thalamic GMD was 96.59% (P < 0.001). The accuracy of the GPC trained with thalamic GMV was 93.18% (P < 0.001). The correlation between Hamilton Depression Scale (HAMD) score targets and predictions in the GPR trained with GMD was 0.90 (P < 0.001, r2 = 0.82), and in the GPR trained with GMV, the correlation between HAMD score targets and predictions was 0.89 (P < 0.001, r2 = 0.79). [2] The models trained with ALFF and fALFF in the thalamus failed to discriminate MDD patients from HC participants. [3] The MKL model showed that the left lateral prefrontal thalamus, the right caudal temporal thalamus, and the right sensory thalamus contribute more to the diagnostic classification.

CONCLUSIONS

The results suggested that GMD and GMV, but not functional indicators of the thalamus, have good potential for the individualized diagnosis of MDD. Furthermore, the thalamus shows the heterogeneity in the structural features of thalamic subregions for predicting MDD. To our knowledge, this is the first study to focus on the thalamus for the prediction of MDD using machine learning methods at the individual level.

Predicting subclinical psychotic-like experiences on a continuum using machine learning

Previous studies applying machine learning methods to psychosis have primarily been concerned with the binary classification of chronic schizophrenia patients and healthy controls. The aim of this study was to use electroencephalographic (EEG) data and pattern recognition to predict subclinical psychotic-like experiences on a continuum between these two extremes in otherwise healthy people. We applied two different approaches to an auditory oddball regularity learning task obtained from N = 73 participants: A feature extraction and selection routine incorporating behavioural measures, event-related potential components and effective connectivity parameters; Regularisation of spatiotemporal maps of event-related potentials. Using the latter approach, optimal performance was achieved using the response to frequent, predictable sounds. Features within the P50 and P200 time windows had the greatest contribution toward lower Prodromal Questionnaire (PQ) scores and the N100 time window contributed most to higher PQ scores. As a proof-of-concept, these findings demonstrate that EEG data alone are predictive of individual psychotic-like experiences in healthy people. Our findings are in keeping with the mounting evidence for altered sensory responses in schizophrenia, as well as the notion that psychosis may exist on a continuum expanding into the non-clinical population.

Brain activation during the observation of real soccer game situations predicts creative goal scoring

Creativity is an important source of success in soccer players. In order to be effective in soccer, unpredictable, sudden and at the same time creative (i.e. unique, original and effective) ideas are required in situations with high time pressure. Accordingly, creative task performance in soccer should be primarily driven by rapid and automatic cognitive processes. This study investigated if functional patterns of brain activation during the observation/encoding of real soccer game situations can predict creative soccer task performance. A machine learning approach (multivariate pattern recognition) was applied in a sample of 35 experienced male soccer players. The results revealed that brain activation during the observation of the soccer scenes significantly predicted creative soccer task performance, while brain activation during the subsequent ideation/elaboration period did not. The identified brain network included areas such as the angular gyrus, the supramarginal gyrus, the occipital cortex, parts of the cerebellum and (left) supplementary motor areas, which are important for semantic information processing, memory retrieval, integration of sensory information and motor control. This finding suggests that early and presumably automatized neurocognitive processes, such as (implicit) knowledge about motor movements, and the rapid integration of information from different sources are important for creative task performance in soccer.

Regional brain volume predicts response to methylphenidate treatment in individuals with ADHD

BACKGROUND

Despite the effectiveness of methylphenidate for treating ADHD, up to 30% of individuals with ADHD show poor responses to methylphenidate. Neuroimaging biomarkers to predict medication responses remain elusive. This study characterized neuroanatomical features that differentiated between clinically good and poor methylphenidate responders with ADHD.

METHODS

Using a naturalistic observation design selected from a larger cohort, we included 79 drug-naive individuals (aged 6-42 years) with ADHD without major psychiatric comorbidity, who had acceptable baseline structural MRI data quality. Based on a retrospective chart review, we defined responders by individuals' responses to at least one-month treatment with methylphenidate. A nonparametric mass-univariate voxel-based morphometric analysis was used to compare regional gray matter volume differences between good and poor responders. A multivariate pattern recognition based on the support vector machine was further implemented to identify neuroanatomical indicators to predict an individual's response.

RESULTS

63 and 16 individuals were classified in the good and poor responder group, respectively. Using the small-volume correction procedure based on the hypothesis-driven striatal and default-mode network masks, poor responders had smaller regional volumes of the left putamen as well as larger precuneus volumes compared to good responders at baseline. The machine learning approach identified that volumetric information among these two regions alongside the left frontoparietal regions, occipital lobes, and posterior/inferior cerebellum could predict clinical responses to methylphenidate in individuals with ADHD.

CONCLUSION

Our results suggest regional striatal and precuneus gray matter volumes play a critical role in mediating treatment responses in individuals with ADHD.

Vulnerability and Protective Factors for PTSD and Depression Symptoms Among Healthcare Workers During COVID-19: A Machine Learning Approach

Background: Healthcare workers are at high risk for developing mental health problems during the COVID-19 pandemic. There is an urgent need to identify vulnerability and protective factors related to the severity of psychiatric symptoms among healthcare workers to implement targeted prevention and intervention programs to reduce the mental health burden worldwide during COVID-19. Objective: The present study aimed to apply a machine learning approach to predict depression and PTSD symptoms based on psychometric questions that assessed: (1) the level of stress due to being isolated from one's family; (2) professional recognition before and during the pandemic; and (3) altruistic acceptance of risk during the COVID-19 pandemic among healthcare workers. Methods: A total of 437 healthcare workers who experienced some level of isolation at the time of the pandemic participated in the study. Data were collected using a web survey conducted between June 12, 2020, and September 19, 2020. We trained two regression models to predict PTSD and depression symptoms. Pattern regression analyses consisted of a linear epsilon-insensitive support vector machine (ε-SVM). Predicted and actual clinical scores were compared using Pearson's correlation coefficient (r), the coefficient of determination (r²), and the normalized mean squared error (NMSE) to evaluate the model performance. A permutation test was applied to estimate significance levels. Results: Results were significant using two different cross-validation strategies to significantly decode both PTSD and depression symptoms. For all of the models, the stress due to social isolation and professional recognition were the variables with the greatest contributions to the predictive function. Interestingly, professional recognition had a negative predictive value, indicating an inverse relationship with PTSD and depression symptoms. Conclusions: Our findings emphasize the protective role of professional recognition and the vulnerability role of the level of stress due to social isolation in the severity of posttraumatic stress and depression symptoms. The insights gleaned from the current study will advance efforts in terms of intervention programs and public health messaging.

Linking Brain Biology to Intellectual Endowment: A Review on the Associations of Human Intelligence With Neuroimaging Data

Human intelligence has always been a fascinating subject for scientists. Since the inception of Spearman's general intelligence in the early 1900s, there has been significant progress towards characterizing different aspects of intelligence and its relationship with structural and functional features of the brain. In recent years, the invention of sophisticated brain imaging devices using Diffusion-Weighted Imaging (DWI) and functional Magnetic Resonance Imaging (fMRI) has allowed researchers to test hypotheses about neural correlates of intelligence in humans.This review summarizes recent findings on the associations of human intelligence with neuroimaging data. To this end, first, we review the literature that has related brain morphometry to intelligence. Next, we elaborate on the applications of DWI and restingstate fMRI on the investigation of intelligence. Then, we provide a survey of literature that has used multimodal DWI-fMRI to shed light on intelligence. Finally, we discuss the state-of-the-art of individualized prediction of intelligence from neuroimaging data and point out future strategies. Future studies hold promising outcomes for machine learning-based predictive frameworks using neuroimaging features to estimate human intelligence.

A machine learning-based classification approach on Parkinson's disease diffusion tensor imaging datasets

Introduction

The presence of motor signs and symptoms in Parkinson’s disease (PD) is the result of a long-lasting prodromal phase with an advancing neurodegenerative process. The identification of PD patients in an early phase is, however, crucial for developing disease-modifying drugs. The objective of our study is to investigate whether Diffusion Tensor Imaging (DTI) of the Substantia nigra (SN) analyzed by machine learning algorithms (ML) can be used to identify PD patients.

Methods

Our study proposes the use of computer-aided algorithms and a highly reproducible approach (in contrast to manually SN segmentation) to increase the reliability and accuracy of DTI metrics used for classification.

Results

The results of our study do not confirm the feasibility of the DTI approach, neither on a whole-brain level, ROI-labelled analyses, nor when focusing on the SN only.

Conclusions

Our study did not provide any evidence to support the hypothesis that DTI-based analysis, in particular of the SN, could be used to identify PD patients correctly.

Multimodal magnetic resonance imaging correlates of motor outcome after stroke using machine learning

This study applied machine learning regression to predict motor function after stroke based on multimodal magnetic resonance imaging. Fifty-four stroke patients, who underwent T1 weighted, diffusion tensor, and resting state functional magnetic resonance imaging were retrospectively included. The kernel rigid regression machine algorithm was applied to gray and white matter maps in T1 weighted, fractional anisotropy and mean diffusivity maps in diffusion tensor, and two motor-related independent component analysis maps in resting state functional magnetic resonance imaging to predict Fugl-Meyer motor assessment scores with the covariate as the onset duration after stroke. The results were validated using the leave-one-subject-out cross-validation method. This study is the first to apply machine learning in this area using multimodal magnetic resonance imaging data, which constitutes the main novelty. Multimodal magnetic resonance imaging correctly predicted the Fugl-Meyer motor assessment score in 72 % of cases with a normalized mean squared error of 5.93 (p value = 0.0020). The ipsilesional premotor, periventricular, and contralesional cerebellar areas were shown to be of relatively high importance in the prediction. Machine learning using multimodal magnetic resonance imaging data after a stroke may predict motor outcome.

All roads lead to the default-mode network-global source of DMN abnormalities in major depressive disorder

Major depressive disorder (MDD) is a psychiatric disorder characterized by abnormal resting state functional connectivity (rsFC) in various neural networks and especially in default-mode network (DMN). However, inconsistent findings, i.e., increased and decreased DMN rsFC, have been reported, which raise the question for the source of DMN changes in MDD. Testing whether the DMN abnormalities in MDD can be traced to either a local, i.e., intra-network, or a global, i.e., inter-network, source, we conducted a novel sequence of rsFC analyses, i.e., global FC, intra-network FC, and inter-network FC. Moreover, all analyses were conducted without global signal regression (non-GSR) and with GSR in order to identify the impact of specifically the global component of functional connectivity on within-network functional connectivity within specifically the DMN. In MDD our findings demonstrate (i) increased representation of global signal correlation (GSCORR) in DMN regions, as confirmed independently by degree of centrality (DC) and by an independent DMN template, (ii) increased within-network DMN rsFC, (iii) highly increased inter-network rsFC of both lower- and higher order non-DMN networks with DMN, (iv) high accuracy in classifying MDD vs. healthy subjects by using GSCORR as predictor. Further supporting the global, i.e., non-DMN source of within-network rsFC of the DMN, all results were obtained only when including the global signal, i.e., non-GSR, but not when conducting GSR. Together, we show for the first time increased global signal representation within rsFC of DMN as stemming from inter-network sources as distinguished from local sources, i.e., within- or intra-DMN.

MVPANI: A Toolkit With Friendly Graphical User Interface for Multivariate Pattern Analysis of Neuroimaging Data

With the rapid development of machine learning techniques, multivariate pattern analysis (MVPA) is becoming increasingly popular in the field of neuroimaging data analysis. Several software packages have been developed to facilitate its application in neuroimaging studies. As most of these software packages are based on command lines, researchers are required to learn how to program, which has greatly limited the use of MVPA for researchers without programming skills. Moreover, lacking a graphical user interface (GUI) also hinders the standardization of the application of MVPA in neuroimaging studies and, consequently, the replication of previous studies or comparisons of results between different studies. Therefore, we developed a GUI-based toolkit for MVPA of neuroimaging data: MVPANI (MVPA for Neuroimaging). Compared with other existing software packages, MVPANI has several advantages. First, MVPANI has a GUI and is, thus, more friendly for non-programmers. Second, MVPANI offers a variety of machine learning algorithms with the flexibility of parameter modification so that researchers can test different algorithms and tune parameters to identify the most suitable algorithms and parameters for their own data. Third, MVPANI also offers the function of data fusion at two levels (feature level or decision level) to utilize complementary information contained in different measures obtained from multimodal neuroimaging techniques. In this paper, we introduce this toolkit and provide four examples to demonstrate its usage, including (1) classification between patients and controls, (2) identification of brain areas containing discriminating information, (3) prediction of clinical scores, and (4) multimodal data fusion.

Predicting differential diagnosis between bipolar and unipolar depression with multiple kernel learning on multimodal structural neuroimaging

One of the greatest challenges in providing early effective treatment in mood disorders is the early differential diagnosis between major depression (MDD) and bipolar disorder (BD). A remarkable need exists to identify reliable biomarkers for these disorders. We integrate structural neuroimaging techniques (i.e. Tract-based Spatial Statistics, TBSS, and Voxel-based morphometry) in a multiple kernel learning procedure in order to define a predictive function of BD against MDD diagnosis in a sample of 148 patients. We achieved a balanced accuracy of 73.65% with a sensitivity for BD of 74.32% and specificity for MDD of 72.97%. Mass-univariates analyses showed reduced grey matter volume in right hippocampus, amygdala, parahippocampal, fusiform gyrus, insula, rolandic and frontal operculum and cerebellum, in BD compared to MDD. Volumes in these regions and in anterior cingulate cortex were also reduced in BD compared to healthy controls (n = 74). TBSS analyses revealed widespread significant effects of diagnosis on fractional anisotropy, axial, radial, and mean diffusivity in several white matter tracts, suggesting disruption of white matter microstructure in depressed patients compared to healthy controls, with worse pattern for MDD. To best of our knowledge, this is the first study combining grey matter and diffusion tensor imaging in predicting BD and MDD diagnosis. Our results prompt brain quantitative biomarkers and multiple kernel learning as promising tool for personalized treatment in mood disorders.

Classifying heterogeneous presentations of PTSD via the default mode, central executive, and salience networks with machine learning

Intrinsic connectivity networks (ICNs), including the default mode network (DMN), the central executive network (CEN), and the salience network (SN) have been shown to be aberrant in patients with posttraumatic stress disorder (PTSD). The purpose of the current study was to a) compare ICN functional connectivity between PTSD, dissociative subtype PTSD (PTSD+DS) and healthy individuals; and b) to examine the use of multivariate machine learning algorithms in classifying PTSD, PTSD+DS, and healthy individuals based on ICN functional activation. Our neuroimaging dataset consisted of resting-state fMRI scans from 186 participants [PTSD (n = 81); PTSD + DS (n = 49); and healthy controls (n = 56)]. We performed group-level independent component analyses to evaluate functional connectivity differences within each ICN. Multiclass Gaussian Process Classification algorithms within PRoNTo software were then used to predict the diagnosis of PTSD, PTSD+DS, and healthy individuals based on ICN functional activation. When comparing the functional connectivity of ICNs between PTSD, PTSD+DS and healthy controls, we found differential patterns of connectivity to brain regions involved in emotion regulation, in addition to limbic structures and areas involved in self-referential processing, interoception, bodily self-consciousness, and depersonalization/derealization. Machine learning algorithms were able to predict with high accuracy the classification of PTSD, PTSD+DS, and healthy individuals based on ICN functional activation. Our results suggest that alterations within intrinsic connectivity networks may underlie unique psychopathology and symptom presentation among PTSD subtypes. Furthermore, the current findings substantiate the use of machine learning algorithms for classifying subtypes of PTSD illness based on ICNs.

Investigating the effect of changing parameters when building prediction models for post-stroke aphasia

Neuroimaging has radically improved our understanding of how speech and language abilities map to the brain in normal and impaired participants, including the diverse, graded variations observed in post-stroke aphasia. A handful of studies have begun to explore the reverse inference: creating brain-to-behaviour prediction models. In this study, we explored the effect of three critical parameters on model performance: (1) brain partitions as predictive features, (2) combination of multimodal neuroimaging and (3) type of machine learning algorithms. We explored the influence of these factors while predicting four principal dimensions of language and cognition variation in post-stroke aphasia. Across all four behavioural dimensions, we consistently found that prediction models derived from diffusion-weighted data did not improve performance over models using structural measures extracted from T1 scans. Our results provide a set of principles to guide future work aiming to predict outcomes in neurological patients from brain imaging data.

Intrinsic Cortico-Subcortical Functional Connectivity in Developmental Dyslexia and Developmental Coordination Disorder

Developmental dyslexia (DD) and developmental coordination disorder (DCD) are distinct diagnostic disorders. However, they also frequently co-occur and may share a common etiology. It was proposed conceptually a neural network framework that explains differences and commonalities between DD and DCD through impairments of distinct or intertwined cortico-subcortical connectivity pathways. The present study addressed this issue by exploring intrinsic cortico-striatal and cortico-cerebellar functional connectivity in a large (n = 136) resting-state fMRI cohort study of 8–12-year-old children with typical development and with DD and/or DCD. We delineated a set of cortico-subcortical functional circuits believed to be associated with the brain’s main functions (visual, somatomotor, dorsal attention, ventral attention, limbic, frontoparietal control, and default-mode). Next, we assessed, using general linear and multiple kernel models, whether and which circuits distinguished between the groups. Findings revealed that somatomotor cortico-cerebellar and frontoparietal cortico-striatal circuits are affected in the presence of DCD, including abnormalities in cortico-cerebellar connections targeting motor-related regions and cortico-striatal connections mapping onto posterior parietal cortex. Thus, DCD but not DD may be considered as an impairment of cortico-subcortical functional circuits.

Fast temporal dynamics and causal relevance of face processing in the human temporal cortex

We measured the fast temporal dynamics of face processing simultaneously across the human temporal cortex (TC) using intracranial recordings in eight participants. We found sites with selective responses to faces clustered in the ventral TC, which responded increasingly strongly to marine animal, bird, mammal, and human faces. Both face-selective and face-active but non-selective sites showed a posterior to anterior gradient in response time and selectivity. A sparse model focusing on information from the human face-selective sites performed as well as, or better than, anatomically distributed models when discriminating faces from non-faces stimuli. Additionally, we identified the posterior fusiform site (pFUS) as causally the most relevant node for inducing distortion of conscious face processing by direct electrical stimulation. These findings support anatomically discrete but temporally distributed response profiles in the human brain and provide a new common ground for unifying the seemingly contradictory modular and distributed modes of face processing.

Early identification of bipolar from unipolar depression before manic episode: Evidence from dynamic rfMRI

OBJECTIVE

Misdiagnosis of bipolar disorder (BD) as unipolar disorder (UD) may cause improper treatment strategy to be chosen, especially in the early stages of disease. The aim of this study was to characterize alterations in specific brain networks for depressed patients who transformed into BD (tBD) from UD.

METHOD

The module allegiance from resting-fMRI by applying a multilayer modular method was estimated in 99 patients (33 tBD, 33 BD, 33 UD) and 33 healthy controls (HC). A classification model was trained on tBD and UD patients. HC was used to explore the functional declination patterns of BD, tBD, and UD.

RESULTS

Based on our classification model, difference mainly reflected in default-mode network (DMN). Compared with HC, both BD and tBD focused on the difference of somatomotor network (SMN), while UD on the abnormity of DMN. The patterns of brain network between patients with BD and tBD were well-overlapped, except for cognitive control network (CCN).

CONCLUSION

The functional declination of internal interaction in DMN was suggested to be useful for the identification of BD from UD in the early stage. The higher recruitment of DMN may predispose patients to depressive states, while higher recruitment of SMN makes them more sensitive to external stimuli and prone to mania. Furthermore, CCN may be a critical network for identifying different stages of BD, suggesting that the onset of mania in depressed patients is accompanied by CCN related cognitive impairments.

How do you perceive threat? It's all in your pattern of brain activity

Whether subtle differences in the emotional context during threat perception can be detected by multi-voxel pattern analysis (MVPA) remains a topic of debate. To investigate this question, we compared the ability of pattern recognition analysis to discriminate between patterns of brain activity to a threatening versus a physically paired neutral stimulus in two different emotional contexts (the stimulus being directed towards or away from the viewer). The directionality of the stimuli is known to be an important factor in activating different defensive responses. Using multiple kernel learning (MKL) classification models, we accurately discriminated patterns of brain activation to threat versus neutral stimuli in the directed towards context but not during the directed away context. Furthermore, we investigated whether it was possible to decode an individual’s subjective threat perception from patterns of whole-brain activity to threatening stimuli in the different emotional contexts using MKL regression models. Interestingly, we were able to accurately predict the subjective threat perception index from the pattern of brain activation to threat only during the directed away context. These results show that subtle differences in the emotional context during threat perception can be detected by MVPA. In the directed towards context, the threat perception was more intense, potentially producing more homogeneous patterns of brain activation across individuals. In the directed away context, the threat perception was relatively less intense and more variable across individuals, enabling the regression model to successfully capture the individual differences and predict the subjective threat perception.

Atlas-Based Classification Algorithms for Identification of Informative Brain Regions in fMRI Data

Multi-voxel pattern analysis (MVPA) has been successfully applied to neuroimaging data due to its larger sensitivity compared to univariate traditional techniques. Searchlight is the most widely employed approach to assign functional value to different regions of the brain. However, its performance depends on the size of the sphere, which can overestimate the region of activation when a large sphere size is employed. In the current study, we examined the validity of two different alternatives to Searchlight: an atlas-based local averaging method (ABLA, Schrouff et al. Neuroinformatics 16, 117-143, 2013a) and a Multi-Kernel Learning (MKL, Rakotomamonjy et al. Journal of Machine Learning 9, 2491-2521, 2008) approach, in a scenario where the goal is to find the informative brain regions that support certain mental operations. These methods employ weights to measure the informativeness of a brain region and highly reduce the large computational cost that Searchlight entails. We evaluated their performance in two different scenarios where the differential BOLD activation between experimental conditions was large vs. small, and employed nine different atlases to assess the influence of diverse brain parcellations. Results show that both methods were able to localize informative regions when differences between conditions were large, demonstrating a large sensitivity and stability in the identification of regions across atlases. Moreover, the sign of the weights reported by these methods provided the directionality of univariate approaches. However, when differences were small, only ABLA localized informative regions. Thus, our results show that atlas-based methods are useful alternatives to Searchlight, but that the nature of the classification to perform should be taken into account when choosing the specific method to implement.

Combining heterogeneous data sources for neuroimaging based diagnosis: re-weighting and selecting what is important

Combining neuroimaging and clinical information for diagnosis, as for example behavioral tasks and genetics characteristics, is potentially beneficial but presents challenges in terms of finding the best data representation for the different sources of information. Their simple combination usually does not provide an improvement if compared with using the best source alone. In this paper, we proposed a framework based on a recent multiple kernel learning algorithm called EasyMKL and we investigated the benefits of this approach for diagnosing two different mental health diseases. The well known Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset tackling the Alzheimer Disease (AD) patients versus healthy controls classification task, and a second dataset tackling the task of classifying an heterogeneous group of depressed patients versus healthy controls. We used EasyMKL to combine a huge amount of basic kernels alongside a feature selection methodology, pursuing an optimal and sparse solution to facilitate interpretability. Our results show that the proposed approach, called EasyMKLFS, outperforms baselines (e.g. SVM and SimpleMKL), state-of-the-art random forests (RF) and feature selection (FS) methods.

Quantifying performance of machine learning methods for neuroimaging data

Machine learning is increasingly being applied to neuroimaging data. However, most machine learning algorithms have not been designed to accommodate neuroimaging data, which typically has many more data points than subjects, in addition to multicollinearity and low signal-to-noise. Consequently, the relative efficacy of different machine learning regression algorithms for different types of neuroimaging data are not known. Here, we sought to quantify the performance of a variety of machine learning algorithms for use with neuroimaging data with various sample sizes, feature set sizes, and predictor effect sizes. The contribution of additional machine learning techniques - embedded feature selection and bootstrap aggregation (bagging) - to model performance was also quantified. Five machine learning regression methods - Gaussian Process Regression, Multiple Kernel Learning, Kernel Ridge Regression, the Elastic Net and Random Forest, were examined with both real and simulated MRI data, and in comparison to standard multiple regression. The different machine learning regression algorithms produced varying results, which depended on sample size, feature set size, and predictor effect size. When the effect size was large, the Elastic Net, Kernel Ridge Regression and Gaussian Process Regression performed well at most sample sizes and feature set sizes. However, when the effect size was small, only the Elastic Net made accurate predictions, but this was limited to analyses with sample sizes greater than 400. Random Forest also produced a moderate performance for small effect sizes, but could do so across all sample sizes. Machine learning techniques also improved prediction accuracy for multiple regression. These data provide empirical evidence for the differential performance of various machines on neuroimaging data, which are dependent on number of sample size, features and effect size.

Predicting Bipolar Disorder Risk Factors in Distressed Young Adults From Patterns of Brain Activation to Reward: A Machine Learning Approach

Background

The aim of this study was to apply multivariate pattern recognition to predict the severity of behavioral traits and symptoms associated with risk for bipolar spectrum disorder from patterns of whole-brain activation during reward expectancy to facilitate the identification of individual-level neural biomarkers of bipolar disorder risk.

Methods

We acquired functional neuroimaging data from two independent samples of transdiagnostically recruited adults (18–25 years of age; n = 56, mean age 21.9 ± 2.2 years, 42 women; n = 36, mean age 21.2 ± 2.2 years, 24 women) during reward expectancy task performance. Pattern recognition model performance in each sample was measured using correlation and mean squared error between actual and whole-brain activation–predicted scores on behavioral traits and symptoms.

Results

In the first sample, the model significantly predicted severity of a specific hypo/mania-related symptom, heightened energy, measured by the energy manic subdomain of the Mood Spectrum Structured Interviews (r = .42, p = .001; mean squared error = 9.93, p = .001). The region with the highest contribution to the model was the left ventrolateral prefrontal cortex. Results were confirmed in the second sample (r = .33, p = .01; mean squared error = 8.61, p = .01), in which the severity of this symptom was predicted using a bilateral ventrolateral prefrontal cortical mask (r = .33, p = .009, mean squared error = 9.37, p = .04).

Conclusions

The severity of a specific hypo/mania-related symptom was predicted from patterns of whole-brain activation in two independent samples. Given that emerging manic symptoms predispose to bipolar disorders, these findings could provide neural biomarkers to aid early identification of individual-level bipolar disorder risk in young adults.

Predicting anxiety from wholebrain activity patterns to emotional faces in young adults: a machine learning approach

BACKGROUND

It is becoming increasingly clear that pathophysiological processes underlying psychiatric disorders categories are heterogeneous on many levels, including symptoms, disease course, comorbidity and biological underpinnings. This heterogeneity poses challenges for identifying biological markers associated with dimensions of symptoms and behaviour that could provide targets to guide treatment choice and novel treatment. In response, the research domain criteria (RDoC) (Insel et al., 2010) was developed to advocate a dimensional approach which omits any disease definitions, disorder thresholds, or cut-points for various levels of psychopathology to understanding the pathophysiological processes underlying psychiatry disorders. In the present study we aimed to apply pattern regression analysis to identify brain signatures during dynamic emotional face processing that are predictive of anxiety and depression symptoms in a continuum that ranges from normal to pathological levels, cutting across categorically-defined diagnoses.

METHODS

The sample was composed of one-hundred and fifty-four young adults (mean age=21.6 and s.d.=2.0, 103 females) consisting of eighty-two young adults seeking treatment for psychological distress that cut across categorically-defined diagnoses and 72 matched healthy young adults. Participants performed a dynamic face task involving fearful, angry and happy faces (and geometric shapes) while undergoing functional Magnetic Resonance Imaging (fMRI). Pattern regression analyses consisted of Gaussian Process Regression (GPR) implemented in the Pattern Recognition for Neuroimaging toolbox (PRoNTo). Predicted and actual clinical scores were compared using Pearson's correlation coefficient (r) and normalized mean squared error (MSE) to evaluate the models' performance. Permutation test was applied to estimate significance levels.

RESULTS

GPR identified patterns of neural activity to dynamic emotional face processing predictive of self-report anxiety in the whole sample, which covered a continuum that ranged from healthy to different levels of distress, including subthreshold to fully-syndromal psychiatric diagnoses. Results were significant using two different cross validation strategies (two-fold: r=0.28 (p-value=0.001), MSE=4.47 (p-value=0.001) and five fold r=0.28 (p-value=0.002), MSE=4.62 (p-value=0.003). The contributions of individual regions to the predictive model were very small, demonstrating that predictions were based on the overall pattern rather than on a small combination of regions.

CONCLUSIONS

These findings represent early evidence that neuroimaging techniques may inform clinical assessment of young adults irrespective of diagnoses by allowing accurate and objective quantitative estimation of psychopathology.

Pain-Related Fear-Dissociable Neural Sources of Different Fear Constructs

Fear of pain demonstrates significant prognostic value regarding the development of persistent musculoskeletal pain and disability. Its assessment often relies on self-report measures of pain-related fear by a variety of questionnaires. However, based either on “fear of movement/(re)injury/kinesiophobia,” “fear avoidance beliefs,” or “pain anxiety,” pain-related fear constructs plausibly differ while it is unclear how specific the questionnaires are in assessing these different constructs. Furthermore, the relationship of pain-related fear to other anxiety measures such as state or trait anxiety remains ambiguous. Advances in neuroimaging such as machine learning on brain activity patterns recorded by functional magnetic resonance imaging might help to dissect commonalities or differences across pain-related fear constructs. We applied a pattern regression approach in 20 human patients with nonspecific chronic low back pain to reveal predictive relationships between fear-related neural pattern information and different pain-related fear questionnaires. More specifically, the applied multiple kernel learning approach allowed the generation of models to predict the questionnaire scores based on a hierarchical ranking of fear-related neural patterns induced by viewing videos of activities potentially harmful for the back. We sought to find evidence for or against overlapping pain-related fear constructs by comparing the questionnaire prediction models according to their predictive abilities and associated neural contributors. By demonstrating evidence of nonoverlapping neural predictors within fear-processing regions, the results underpin the diversity of pain-related fear constructs. This neuroscientific approach might ultimately help to further understand and dissect psychological pain-related fear constructs.

Random Forests Based Group Importance Scores and Their Statistical Interpretation: Application for Alzheimer's Disease

Machine learning approaches have been increasingly used in the neuroimaging field for the design of computer-aided diagnosis systems. In this paper, we focus on the ability of these methods to provide interpretable information about the brain regions that are the most informative about the disease or condition of interest. In particular, we investigate the benefit of group-based, instead of voxel-based, analyses in the context of Random Forests. Assuming a prior division of the voxels into non overlapping groups (defined by an atlas), we propose several procedures to derive group importances from individual voxel importances derived from Random Forests models. We then adapt several permutation schemes to turn group importance scores into more interpretable statistical scores that allow to determine the truly relevant groups in the importance rankings. The good behaviour of these methods is first assessed on artificial datasets. Then, they are applied on our own dataset of FDG-PET scans to identify the brain regions involved in the prognosis of Alzheimer's disease.