Frontiers in Neuroimaging

Resting-state functional network connectivity (rsFNC) has shown utility for identifying characteristic functional brain patterns in individuals with psychiatric and mood disorders, providing a promising avenue for biomarker development. However, several factors have precluded widespread clinical adoption of rsFNC diagnostics, namely a lack of standardized approaches for capturing comparable and reproducible imaging markers across individuals, as well as the disagreement on the amount of data required to robustly detect intrinsic connectivity networks (ICNs) and diagnostically relevant patterns of rsFNC at the individual subject level. Recently, spatially constrained independent component analysis (scICA) has been proposed as an automated method for extracting ICNs standardized to a chosen network template while still preserving individual variation. Leveraging the novel scICA methodology, which solves the former challenge of standardized neuroimaging markers, we investigate the latter challenge of identifying a minimally sufficient data length for clinical applications of resting-state fMRI (rsfMRI). Using a dataset containing individuals with schizophrenia and controls (M = 310) as well as simulated rsfMRI, we evaluate the robustness of ICN and rsFNC estimates at both the subject- and group-level, as well as the performance of diagnostic classification, with respect to the length of the rsfMRI time course. We found individual estimates of ICNs and rsFNC from the full-length (5 minute) reference time course were sufficiently approximated with just 3-3.5 minutes of data (r = 0.85, 0.88, respectively), and significant differences in group-average rsFNC could be sufficiently approximated with even less data, just 2 minutes (r = 0.86). The results from the shorter clinical data were consistent with the results from the longer simulated data, reliably estimating both individual- and group-level metrics from the full-length (30 minute) reference with just 3-4 minutes of data (r = 0.85 - 0.88). Furthermore, we found a model trained on 2 minutes of data retained 97-98% classification accuracy relative to that of the full-length reference model. Our results suggest that clinical rsfMRI scans, when decomposed with scICA, could potentially be shortened to just 2-4 minutes without significant loss of individual rsFNC information or classification performance of longer scan lengths.

Single photon emission computed tomography (SPECT) is a highly specialized imaging modality that enables measurement of regional cerebral perfusion and, in particular, resting cerebral blood flow (rCBF). Recent technological advances have improved SPECT quantification and reliability, making it increasingly useful for studying rCBF abnormalities and perfusion-network alterations in psychiatric and neurological disorders. To characterize large-scale functional organization in SPECT data, data-driven decomposition methods such as independent component analysis (ICA) have been used to extract covarying perfusion patterns that map onto interpretable brain networks. Blind ICA provides a data-driven approach to estimate these networks without strong prior assumptions. More recently, a hybrid approach that leverages spatial priors to guide a spatially constrained ICA (sc-ICA) have been used to fully automate the ICA analysis while also providing participant-specific network estimates. While this has been reliably demonstrated in fMRI with the NeuroMark template, there is currently no comparable SPECT template. A SPECT template would enable automatic estimation of functional SPECT networks with participant-specific expressions that correspond across participants and studies. The current study introduces a new replicable NeuroMark SPECT template for estimating canonical perfusion covariance patterns (networks). We first identify replicable SPECT networks using blind ICA applied to two large sample SPECT datasets. We then demonstrate the use of the resulting template by applying sc-ICA to an independent schizophrenia dataset. In sum, this work presents and shares the first NeuroMark SPECT template and demonstrating its utility in an independent cohort, providing a scalable and robust framework for network-based analyses.

PurposePrior work has shown that whole-band linear regression of nuisance signals can introduce artifactual connectivity in high-frequency resting-state fMRI. Errors of motion regressors and non-stationarity of nuisance signals exacerbate artifacts. Here, we introduce spectral-temporal segmentation of regression vectors to decouple regression in different frequency bands to reduce motion artifacts. MethodsAn alternative approach to whole-band linear nuisance regression is introduced in the present work relying on spectral segmentation of the motion parameters into k-bands using non-causal or FIR filters, with whole-band regression of the filtering residual, and temporal segmentation of regression vectors. The methodology was tested in computer simulations and in-vivo data. Resting-state networks in five healthy controls and two brain tumor patients using high-speed fMRI (TR >= 205 ms) were mapped using the present approach combined with spectrally constrained regression of physiological noise and the results were compared to the conventional whole band regression approach. ResultsComputer simulations showed high tolerance to frequency dependent errors in regression vectors. Motion and physiological noise artifacts in-vivo were substantially reduced without introducing artifactual connectivity. Artifactual connectivity decreased asymptotically with increasing number of frequency bands without decreasing connectivity in major resting-state networks. Connectivity above 0.3 Hz in-vivo was consistent with that in traditional low-frequency networks. ConclusionsSpectral-temporal segmentation of regression vectors is a powerful approach to reduce artifacts from non-stationary high-bandwidth nuisance signals.

The disconnection hypothesis of schizophrenia proposes that symptoms of the disorder arise as a result of aberrant functional integration between segregated areas of the brain. The concept of metastability characterizes the coexistence of competing tendencies for functional integration and functional segregation in the brain and is therefore well suited for the study of schizophrenia. In this study we investigate metastability as a neuromechanistic biomarker of schizophrenia pathology, including a demonstration of reliability and face validity. Group-level discrimination, individual-level classification, and pathophysiological relevance were assessed using two independent case-control studies of schizophrenia, the Human Connectome Project Early Psychosis (HCPEP) study (controls n=82, non-affective psychosis n=53) and the Cobre study (controls n=71, cases n=59). In this work we introduce a new framework that uses Leading Eigenvector Dynamic Analysis (LEiDA) to capture specific features of dynamic functional connectivity and then implements a novel approach to estimate metastability. We used non-parametric testing to evaluate group-level differences and a naive Bayes classifier to discriminate cases from controls. Our results show that our new approach is capable of discriminating cases from controls with elevated effect sizes relative to published literature, reflected in an up to 76% area under the curve (AUC) in out-of-sample classification analyses. Furthermore, our analyses demonstrated that patients with early psychosis exhibit intermittent disconnectivity of subcortical regions with frontal cortex and cerebellar regions, introducing new insights about the mechanistic bases of these conditions. Overall, these findings demonstrate reliability and face validity of metastability as a neuromechanistic biomarker of schizophrenia pathology.

The process of finding reliable treatment for major depression and bipolar disorder can be arduous. The myriad behavioral symptoms presented by patients and resistance to treatment from particular medication classes complicate standard diagnostic and prescription methodologies, often requiring multiple attempted treatments during which symptoms may still be present. Physiological information such as neuroimaging scans may help to alleviate some of the uncertainty surrounding diagnosis and treatment when incorporated into a clinical setting. Changes in functional magnetic resonance imaging show particular promise, as the incorporation of dynamical information may provide insights into physiological changes prior to static, structural changes. In this work, we present a novel method for generating robust and replicable dynamic functional network connectivity (dFNC) features from neuroimaging data using a template of dynamic states derived from a large, non-affected data set. We demonstrate that this template-driven dFNC approach expands on standard dFNC approaches by allowing for the derivation of a continuous state-contribution time series. We demonstrate that the derived biomarkers can support high predictive performance for the identification of medication class and non-responders while also expanding the set of biomarkers available for studying differences in mood disorder medication response.

Psychosis as a symptom manifests in schizophenia and bipolar disorder, two highly heterogeneous psychiatric illnesses with overlapping clinical manifestations. Resting-state functional Magnetic Resonance Imaging (rsfMRI), represents a promising tool for identifying objective biomarkers of functional brain alterations to aid differential diagnosis. In this work, we comparatively evaluate multiple rs-fMRI representations for differentiating schizophrenia and bipolar disorder using intrinsic connectivity network (ICN) temporal profiles and several functional network connectivity (FNC) approaches, including static, dynamic, and high-order connectivity analyses. The study was conducted on a cohort of 371 subjects with psychosis, while evaluation was performed using a separate held-out cohort of 315 subjects. We investigated convolutional neural network architectures applied to ICN temporal profiles, spectrograms, and scalograms, alongside classical machine learning models trained on connectivity-derived features. Across the evaluated approaches, ICN temporal profiles provided the most consistent discriminative performance, with a 1D convolutional neural network achieving the strongest overall results under the benchmark protocol. Among connectivity-based methods, static functional connectivity generally outperformed dynamic and high-order representations, suggesting that increased representational complexity did not necessarily translate into improved generalization. Although the obtained classification performance remained modest, the results highlight the challenges of robust psychosis differentiation using rs-fMRI while emphasizing the relative stability of low-order connectivity representations and temporal ICN features. These findings contribute to ongoing efforts toward reproducible and interpretable neuroimaging biomarkers for psychiatric disorders.

We present a method to map the regional similarity between resting state fMRI activities of different individuals. The similarity was measured using cross-entropy. Group level patterns were displayed based on the Human Connectome Project Youth data. While we only showed the cross-subject brain entropy (BEN) mapping results in this manuscript, the same concept can be directly extended to map the cross-sessional BEN and the cross-regional cross-subject or subject-session BEN.

Mental health disorders affect countless people worldwide and present a major challenge for mental health services, which are struggling with the demand on a global scale. Recent studies have indicated that activity of the brains Default Mode Network (DMN) could prove insightful in monitoring patient recovery from depression and has been used as a therapeutic target itself. An opportunity exists to replicate recent therapeutic protocols targeting DMN connectivity via functional magnetic resonance imaging using the more economically scalable modality of electroencephalogram (EEG). The aim of this work was to validate the accuracy of real-time DMN detection methods applied to EEG data, using a publicly available dataset. Using a Hidden Markov Model to identify a 12-state resting-state network, this work achieved an overall DMN detection accuracy of 95%. Furthermore, the model was able to achieve a correlation of 0.617 between the baseline and calculated DMN fractional occupancy. These results demonstrate the ability of real-time analysis to effectively identify the DMN through EEG data providing an avenue for further applications that monitor and treat mental health disorders.

Magnetic resonance spectroscopy (MRS) is heavily dependent on the investigative team to prescribe, or demarcate, the desired tissue volume-of-interest. Manual prescription, the current standard in the field, requires expertise in neuroanatomy to ensure spatial consistency within and across subjects. Spatial precision of MRS voxel placement thus presents challenges for cross-sectional studies, and even more so for repeated-measure and multi-acquisition designs. Furthermore, voxel prescriptions based-solely on anatomical landmarks may not be ideal in regions with substantial functional and cytoarchitectural variability or to examine individualized/targeted interventions. Here we propose and validate robust and real-time methods to automate MRS voxel placement using functionally defined coordinates within the left dorsolateral prefrontal cortex in clinical cohorts of chronic pain and depression. We hypothesized that increased automation would produce more consistent voxel placement across repeated acquisitions particularly in reference to standard manual prescription. Data were collected and analyzed using two independent prospective transcranial magnetic stimulation studies: 1) a single-day multi-session sandwich design and 2) a longitudinal design. Participants with fibromyalgia syndrome (N=50) and major depressive disorder (N=35) underwent MRI as part of ongoing clinical studies. MEGA-PRESS and Optimized-PRESS MRS acquisitions were acquired at 3-tesla. Evaluation of the reproducibility of spatial location and tissue segmentation was assessed for: 1) manual, 2) semi-automated, and 3) automated voxel prescription approaches. Variability of grey and white matter voxel tissue composition was reduced using automated placement protocols as confirmed by common MRS software processing pipelines (Gannet; SPM-based segmentation) and via Freesurfer-based segmentation. Spatially, post-to pre-voxel center-of-gravity distance was reduced and voxel overlap increased significantly across datasets using automated compared to manual procedures. These results demonstrate the within subject reliability and reproducibility of a method for reducing variability introduced by spatial inconsistencies during MRS acquisitions. The proposed method is a meaningful advance toward improved consistency of MRS data in neuroscience and can be leveraged for multi-session and longitudinal studies that target precisely defined regions-of-interest via a coordinate-based approach.

Development of innovative non-invasive neuroimaging methods and biomarkers are critical for studying brain disease. In this work, we have developed a methodology to characterize the frequency responses and spatial localization of oscillations and movements of cerebrospinal fluid (CSF) flow in the human brain. Using 7 Tesla human MRI and ultrafast echo-planar imaging (EPI), in-vivo images were obtained to capture CSF oscillations and movements. Physiological data was simultaneously collected and correlated with the 7T MR data. The primary components of CSF oscillations were identified using spectral analysis (with frequency bands identified around 0.3Hz, 1.2Hz and 2.4Hz) and were mapped spatially and temporally onto the MR image domain and temporally onto the physiological domain. The developed methodology shows a good consistency and repeatability (standard deviation of 0.052 and 0.078 for 0.3Hz and 1.2Hz bands respectively) in-vivo for potential brain dynamics and CSF flow and clearance studies.

The application of multi-echo functional magnetic resonance imaging (fMRI) studies has considerably increased in the last decade due to its superior BOLD sensitivity compared to single-echo fMRI. Various methods have been developed that combine the fMRI time-series derived at different echo times to improve the data quality. Here we evaluated three multi-echo combination schemes, i.e. optimal combination (T2*-weighted), temporal Signal-to-Noise Ratio (tSNR) weighted, and temporal Contrast-to-Noise Ratio (tCNR) weighted combination. For the first time, the effect of these multi-echo combinations on functional resting-state networks was assessed in the temporal and spatial domain, and compared to networks derived from the second echo (35 ms) functional images. Sixteen healthy volunteers were scanned during a 5 minutes resting-state fMRI session. After obtaining the networks, several temporal and spatial metrics were calculated for their time-series and spatial maps. Our results showed that, compared to the second echo network time-series, the Pearson correlation and root mean square error were the most consistent for the optimal combination time-series and the least with those derived from tSNR-weighted combination. The frequency analysis further suggested that the time-series from the tSNR-weighted combination method reduced hardware- and physiological-related artifacts as reflected by the reduced power for the associated frequencies in almost all networks. Moreover, the spatial stability and extent of the networks significantly increased after multi-echo combination, primarily for the optimal combination, followed by the tSNR-weighted combination. The performance of the tCNR-weighted combination lacked robustness and instead varied remarkedly between resting-state networks in both the temporal and spatial domain. The results highlight the benefits of multi-echo sequences on resting-state networks as well as the importance of adjusting the choice of multi-echo combination method to the research question and domain of interest.

Resting-state fMRI has been used in numerous studies to map networks in the brain that employ spatially disparate regions. However, attempts to map networks with high spatial resolution have been hampered by conflicting technical demands and associated problems. Results from recent fMRI studies have shown that spatial resolution remains around 0.7 x 0.7 x 0.7 mm3, with only partial brain coverage. This work presents a novel fMRI method, TR-external EPI with keyhole (TR-external EPIK), which can provide a nominal spatial resolution of 0.51 x 0.51 x 1.00 mm3 (0.26 mm3 voxel) with whole-brain coverage. TR-external EPIK enabled the identification of various resting-state networks distributed throughout the brain from a single fMRI session, with mapping fidelity onto the grey matter at 7T. The high-resolution functional image further revealed mesoscale anatomical structures, such as small cerebral vessels and the internal granular layer of the cortex within the postcentral gyrus.

The rapid evolution of machine learning (ML) methods has yielded promising results in human brain neuroscience. However, the reproducibility of ML applications in neuroimaging remains limited, challenging the generalizability of inferences to broader populations. In addition to the inherent variability of the brain activity (both in healthy and pathological states), poor reproducibility is further enhanced by inconsistencies in data preprocessing techniques and methods for calculating functional connectivity (FC), which are used as parameters for brain state classification. To systematically assess the impact of abovementioned factors on ML applications to fMRI data, we benchmarked a comprehensive set of FC analysis pipelines for the classification task between fMRI data recorded in two fundamentally different states: eyes open and eyes closed. In contrast to studies involving heterogeneous clinical populations or using complex cognitive tasks, our controlled experimental design - based on two independent datasets of healthy participants collected in different laboratories - minimizes variability related to a task design or pathological brain states. Classification accuracy and reproducibility were compared for 256 distinct FC analysis pipelines, covering common preprocessing approaches, brain parcellation schemes, and connectivity metrics. Notably, we employed two ways of validation: a direct cross-site validation strategy - when a model was trained on one site and tested on another, and few-shot domain adaptation - when a few samples of testing site were added to the train set. Despite the substantial variability in pipeline configurations, we observed consistently high classification accuracy ([~]80%), confirming that FC-based models can robustly discriminate between well-defined brain states (eye conditions) across different acquisition sites. Best results both in terms of classification accuracy and stability were observed using Pearson correlation and tangent space parametrization as FC, Brainnetome as atlas, and confound regression strategies based on the CompCor method. These findings highlight the resilience of rs-fMRI FC-derived characteristics to methodological variation and support their utility in the discovery of biomarkers, particularly in settings that involve stable and reproducible brain states.

Machine learning (ML) methodologies offer significant potential for addressing the intricate challenges inherent in the analysis of neuroimaging data within the realm of neurological research. Nonetheless, the effective application of these techniques is markedly contingent upon the particular task and dataset under examination, and the absence of standardized methodologies poses impediments to cross-study result comparisons. This study contributes substantively to the collective endeavor by conducting a comprehensive evaluation and comparative analysis of ML models in the context of predicting schizophrenia and autism spectrum disorder (ASD) utilizing distinct functional Magnetic Resonance Imaging (fMRI) datasets. In this research, we introduce Canonical Correlation Analysis (CCA) as an innovative modality to augment the classification of these multifaceted neurological conditions. By elucidating the efficacy of CCA in ameliorating classification accuracy within the framework of Support Vector Machines (SVM), our study endeavors to propel the domain of neuroimaging and deepen our understanding of these intricate neurological disorders.

The brainstem is the site of key exchanges between the autonomic and central nervous systems but has historically presented a challenging target for study with BOLD fMRI. A potentially powerful although under-characterized approach to identifying nucleic activation within the brainstem is masked independent component analysis (mICA), which restricts signal decomposition to the brainstem itself, thus aiming to reduce the strong effect of physiological noise in nearby regions such as ventricles and large arteries. In this study, we systematically investigate the use of mICA to uncover signatures of autonomic activation in the brainstem at rest. We apply mICA on 40 subjects in a high-resolution resting state 7T dataset following different strategies for dimensionality selection, denoising, and component classification. We show that among the noise mitigation techniques investigated, cerebrospinal fluid denoising makes the largest impact in terms of mICA outcomes. We further demonstrate that across preprocessing pipelines and previously reported results the majority of components are spatially reproducible, but temporal outcomes differ widely depending on denoising strategy. Evaluating both hand-labelling and whole-brain specificity criteria, we develop an intuitive framework for mICA classifications. Finally, we make a comparison between mICA and atlas-based segmentations of brainstem nuclei, finding little consistency between these two approaches. Based on our evaluation of the effects of methodology on mICA and its relationship to other signals of interest in the brainstem, we provide recommendations for future uses of mICA to identify autonomically-relevant BOLD fluctuations in subcortical structures.

Schizophrenia remains a poorly understood disease, hence the interest in assessing and indirectly characterizing brain activity and connectivity. This paper aims to search for potential biomarkers in schizophrenia with functional magnetic resonance data, between subjects in the resting state. Firstly, we used fMRI from an open database, SchizConnect, of 48 subjects, in which 27 were control subjects, with no apparent disease and the others 21 were patients with schizophrenia. With the SPM tool, we proceeded to manually pre-process the images obtained, at the risk of having influenced the final results. Then, with the AAL atlas as a reference, we divided the brain into 116 areas. Then, brain activity in these areas were analysed, using the LEiDA method, which aims to characterize brain activity at each time point t by phase locking patterns of the BOLD signal. After the application of LEiDA, brain activity was evaluated based on trajectories and bar graphs of functional connectivity states in which the probability of occurrence and their dwell time were calculated for each state. It was also found that the visual cortex was the subsystem that showed significantly more probability of occurrence in schizophrenia patients to be assessed, and may correspond to symptoms of hallucinations by the patients with schizophrenia.

The default mode network (DMN), the salience network (SN), and the central executive network (CEN) could be considered as the core resting-state brain networks (RSN) due to their involvement in a wide range of cognitive tasks. Despite the large body of knowledge relating to their regional spontaneous activity (RSA) and functional connectivity (FC) of these networks, less is known about the influence of task-associated activity on these parameters and on the interaction between these three networks. We have investigated the effects of the visual-oddball paradigm on three fMRI measures (amplitude of low-frequency fluctuations for RSA, regional homogeneity for local FC, and degree centrality for global FC) in these three core RSN networks. A rest-task-rest paradigm was used and the RSNs were identified using independent component analysis (ICA) on the resting-state data. We found that the task-related brain activity induced different patterns of significant changes within the three RS networks. Most changes were strongly associated with the task performance. Furthermore, the task-activity significantly increased the inter-network correlations between the SN and CEN as well as between the DMN and CEN, but not between the DMN and SN. A significant dynamical change in RSA, alongside local and global FC within the three core resting-state networks following a simple cognitive activity may be an expression of the distinct involvement of these networks in the performance of the task and their various outcomes.

This report summarizes experiments on exploring non-linear functional connectivity. Using resting-state functional MRI data from the HCP1200 dataset, we define nodes as 17 functional networks and edges as the joint distribution between times series pairs. Linear dependence is removed before taking the joint distribution. We then employ a test for normality on the joint distribution to find non-normal distribution patterns. However, the result from an experimental run of 10 subjects shows that: only less than 1% of edges is non-normal distributed, and the location of such edges is not consistent across subjects. From this point of view, the non-linear part seems to be governed by random noise.

Proton magnetic resonance spectroscopic imaging (1H-MRSI) provides noninvasive evaluation of brain metabolism. However, there are some limitations of 1H-MRSI preventing its wider use in the clinics, including the spectral quality issues, partial volume effect and chemical shift artifact. Additionally, it is necessary to create metabolite maps for analyzing spectral data along with other MRI modalities. In this study, a MATLAB-based open-source data analysis software for 3D 1H-MRSI, called Oryx-MRSI, which includes modules for visualization of raw 1H-MRSI data and LCModel outputs, chemical shift correction, tissue fraction calculation, metabolite map production, and registration onto standard MNI152 brain atlas while providing automatic spectral quality control, is presented. Oryx-MRSI implements region of interest analysis at brain parcellations defined on MNI152 brain atlas. All generated metabolite maps are stored in NIfTI format. Oryx-MRSI is publicly available at https://github.com/sevimcengiz/Oryx-MRSI along with six example datasets.

Spatial orientation is essential to interacting with a physical environment, and better understanding it could contribute to a better understanding of a variety of diseases and disorders that are characterized by deficits in spatial orientation. Many previous studies have focused on the relationship between spatial orientation and individual brain regions, though in recent years studies have begun to examine spatial orientation from a network perspective. This study analyzes dynamic functional network connectivity (dFNC) values extracted from over 800 resting-state fMRI recordings of healthy young adults (age 22-37 years) and applies unsupervised machine learning methods to identify neural brain states that occur across all subjects. We estimated the occupancy rate (OCR) for each subject, which was proportional to the amount of time that they spent in each state, and investigated the link between the OCR and spatial orientation and the state-specific FNC values and spatial orientation controlling for age and sex. Our findings showed that the amount of time subjects spent in a state characterized by increased connectivity within and between visual, auditory, and sensorimotor networks and within the default mode network while at rest corresponded to their performance on tests of spatial orientation. We also found that increased sensorimotor network connectivity in two of the identified states negatively correlated with decreased spatial orientation, further highlighting the relationship between the sensorimotor network and spatial orientation. This study provides insight into how the temporal properties of the functional brain connectivity within and between key brain networks may influence spatial orientation.