Smoothing fMRI data prior to analysis is a fundamental and widely used technique to increase sensitivity. Unconstrained smoothing can also reduce the spatial specificity of the analysis by introducing artifacts in the data. This study tested the effects of smoothing on the reliability and accuracy of both task fMRI and resting state data. The effects of unconstrained smoothing were compared to those of an anatomically constrained smoothing method, which prevents smoothing across the white and gray matter surfaces of the cortex. Unconstrained Gaussian smoothing and anatomically constrained smoothing were applied to simulated data, a sensory task fMRI dataset, a precision fMRI motor task mapping dataset, and a resting state fMRI dataset. Smoothing-related artifacts were tested for and compared between the smoothing methods, and the effects of the smoothing methods on the reliability and accuracy were measured. In the experiments with simulated data, unconstrained Gaussian smoothing demonstrated decreased accuracy and increased white matter activation compared to constrained smoothing. In the sensory task activation analysis, both Gaussian and constrained smoothing increased the reliability of the sensory task fMRI activations, but Gaussian smoothing increased the percentage of active voxels in the white matter relative to constrained smoothing (p < 0.001). Relative to constrained smoothing, Gaussian smoothing with FWHM > 3 mm also decreased the accuracy of motor mapping results from individual sessions to the precision maps (p < 0.001). With cluster significance thresholding, mean false positive voxel percentages remained below 5% for both methods across the tested kernel widths. Both Gaussian and constrained smoothing demonstrated a biasing effect on the resting state connectivity of nearby regions and on the graph theory metrics of the functional connectomes. This study showed that unconstrained Gaussian smoothing spreads activation across cortical boundaries, increases white matter activation, and biases graph theory connectivity metrics. Anatomically constrained smoothing reduced some of these smoothing artifacts while still increasing reliability and may be a reasonable alternative to unconstrained Gaussian smoothing.
Predicting the trajectory of clinical decline in aging individuals is a pressing challenge, especially for people with mild cognitive impairment, Alzheimer's disease, Parkinson's disease, or vascular dementia. Accurate predictions can guide treatment decisions, identify risk factors, and optimize clinical trials. In this study, we compared two deep learning approaches for forecasting changes, over a 2-year interval, in the Clinical Dementia Rating scale 'sum of boxes' score (sobCDR), as a continuous outcome (regression). This is a key metric in dementia research and clinical trials, and scores range from 0 (no impairment) to 18 (severe impairment). To predict decline, we trained a hybrid convolutional neural network (CNN) that integrates 3D T1-weighted brain MRI scans with tabular clinical and demographic features (including age, sex, body mass index (BMI), and baseline sobCDR). We benchmarked its performance against AutoGluon, an automated multimodal machine learning framework that selects an appropriate neural network architecture (an 'autoML' approach). We evaluated the models using data from 2,319 unique participants drawn from three independent cohorts-ADNI, OASIS-3, and NACC. For each participant, we used one T1-weighted brain MRI scan along with corresponding clinical and demographic information. Our results demonstrate the importance of combining image and tabular data in predictive modeling for this clinical application. Deep learning algorithms can fuse information from image-based brain signatures and tabular clinical data, with potential for personalized prognostics in aging and dementia. Rather than concluding that multimodal fusion uniformly improves performance, our results show that deep learning applied to volumetric MRI data may struggle to add predictive value, particularly when clinical covariates explain substantial variance and provide a strong baseline. In other conditions and tasks, it may help to have a hybrid system that can learn from both data types, and their relative value may be different. Conversely, AutoML-based multimodal fusion provides a robust baseline when tabular data already provide strong predictive value for the task. These insights clarify how different multimodal strategies could be selected in clinical prognostic applications.
Neuroimaging plays a key role in the diagnostic workup of patients with idiopathic normal pressure hydrocephalus (iNPH). A flow void in the aqueduct - indicating increased cerebrospinal fluid (CSF) velocity - is a common, but unspecific finding. Aim of this study was to investigate CSF-flow characteristics in iNPH patients before and after spinal tap test (STT) using novel, real-time phase-contrast magnetic resonance imaging (RT-PC MRI). We included consecutive patients with clinical signs of iNPH being electively admitted for diagnostic workup, including neurological examination, conventional MRI and STT. RT-PC MRI and clinical examination were performed before and within 24 h after STT. CSF-flow volumes were determined at five regions in the inner and outer CSF spaces. Fifteen patients with suspected iNPH and five age-matched healthy controls (HC) were included. Baseline RT-PC MRI revealed elevated CSF-flow volumes in the inner ventricular system of iNPH patients compared to healthy controls, being detectable predominantly in the third ventricle (iNPH vs. HC: 15.93 ± 7.01 mL vs. 6.58 ± 2.99 mL, p = 0.020). There was a positive correlation between the Evans Index and CSF-flow in the third ventricle (r = 0.586, p = 0.017), cerebral aqueduct (r = 0.639, p = 0.006) and the fourth ventricle (r = 0.649, p = 0.007). There was no statistically significant change of CSF-flow volumes before and after STT in the iNPH-group. RT-PC MRI provides a promising, non-invasive approach for evaluating CSF-flow in iNPH. Baseline CSF-flow volumes were elevated in the inner ventricular system, particularly in the third ventricle, and correlated with ventricular enlargement, suggesting that increased CSF-flow may reflect disease progression rather than therapeutic response. However, in contrast to clinical tests, the lack of change of CSF-flow after STT limits its utility for patient selection for ventriculo-peritoneal-shunt implantation.
Glioblastoma (GBM) is the most common malignant brain tumor with an abysmal prognosis. Since complete tumor cell removal is impossible due to the infiltrative nature of GBM, accurate measurement is paramount for GBM assessment. Preoperative magnetic resonance images (MRIs) are crucial for initial diagnosis and surgical planning, while follow-up MRIs are vital for evaluating treatment response. The structural changes in the brain caused by surgical and therapeutic measures create significant differences between preoperative and follow-up MRIs. In clinical research, advanced deep learning models trained on preoperative MRIs are often applied to assess follow-up scans, but their effectiveness in this context remains underexplored. Our study evaluates the performance of these models on follow-up MRIs, revealing suboptimal results. To overcome this limitation, we developed a Bayesian deep segmentation model specifically designed for follow-up MRIs. This model is capable of accurately segmenting various GBM tumor sub-regions, including FLAIR hyperintensity regions, enhancing tumor areas, and non-enhancing central necrosis regions. By integrating uncertainty information, our model can identify and correct misclassifications, significantly improving segmentation accuracy. Therefore, the goal of this study is to provide an effective deep segmentation model for accurately segmenting GBM tumor sub-regions in follow-up MRIs, ultimately enhancing clinical decision-making and treatment evaluation. A novel deep segmentation model was developed utilizing 311 follow-up MRIs to segment tumor subregions. This model integrates Bayesian learning to assess the uncertainty of its predictions and employs transfer learning techniques to effectively recognize and interpret textures and spatial details of regions that are typically underrepresented in follow-up MRI data. The proposed model significantly outperformed existing models, achieving DSC scores of 0.833, 0.901, and 0.931 for fluid attenuation inversion recovery hyperintensity, enhancing tumoral and non-enhancing central necrosis, respectively. Our proposed model incorporates brain structural changes following surgical and therapeutic interventions and leverages uncertainty metrics to refine estimates of tumor, demonstrating the potential for improved patient management.
Over the past decade, functional magnetic resonance imaging (fMRI) has emerged as a widely adopted in vivo imaging technique for examining neural activity in the brain. A common preprocessing step in fMRI analysis is spatial smoothing, which helps in detecting cluster-like active regions. The use of a heuristically selected Gaussian filter for spatial smoothing is frequently preferred due to its simplicity and computational efficiency. Neurons in the cerebral cortex are located within a thin sheet of gray matter at the surface of the brain, and the human brain's gyrification results in a complex gray matter anatomy. For task-based fMRI activation analysis, isotropic Gaussian smoothing can reduce spatial specificity, introducing spatial blurring artifacts where inactive voxels near active regions are mistakenly identified as active. This blurring is beneficial for group-level analysis as it helps mitigate anatomical variability across subjects and inaccuracies in spatial normalization. However, it poses challenges in subject-level analysis, particularly in clinical applications such as presurgical planning and fMRI fingerprinting, which demand high spatial specificity. Previous studies have proposed several adaptive spatial smoothing techniques to address these issues. In this study, we introduce a versatile deep neural network (DNN) that builds on the strengths of previous approaches while overcoming their limitations. This method can incorporate additional neighboring voxels for estimating optimal spatial smoothing without significantly increasing computational costs, making it suitable for ultrahigh-resolution (sub-millimeter) task fMRI data. Furthermore, the proposed neural network incorporates brain tissue properties, enabling more accurate characterization of brain activation at the individual level.
Randomized controlled trials (RCTs) are essential for evaluating treatment efficacy, typically comparing active interventions to control conditions. In situations where blinding is impractical-such as in psychological therapies or physical rehabilitation-waitlist controls are often used to account for natural symptom progression and test-retest variability. This study examines the biases introduced by post-hoc analyses under conditions of low statistical power, particularly in neuroimaging research. Through large-scale simulations involving 100 million datasets with varying sample sizes, treatment effects, and test-retest variability, the study demonstrates that the common practice of conducting post-hoc tests only on brain regions showing significant interaction effects can substantially increase the false positive rate in the control condition. These findings underscore the relevance of Berkson's paradox in interpreting unexpected control group outcomes and caution against overinterpreting such results. A complementary neuroimaging simulation reinforces these conclusions, emphasizing the need for critical scrutiny when evaluating significant effects in control groups. Overall, this work challenges conventional post-hoc testing strategies and advocates for a more nuanced and statistically informed interpretation of results, especially in studies with limited power.
Schizophrenia is extremely heterogenous, and the underlying brain mechanisms are not fully understood. Many attempts have been made to substantiate and delineate the relationship between schizophrenia and the brain through unbiased exploratory investigations of resting-state functional magnetic resonance imaging (rs-fMRI). The results of numerous data-driven rs-fMRI studies have converged in support of the disconnection hypothesis framework, reporting aberrant connectivity in cortical-subcortical-cerebellar circuitry. However, this model is vague and underspecified, encompassing a vast array of findings across studies. It is necessary to further refine this model to identify consistent patterns and establish stable imaging markers of schizophrenia and psychosis. The organizational structure of the NeuroMark atlas is especially well-equipped for describing functional units derived through independent component analysis (ICA) and uniting findings across studies utilizing data-driven whole-brain functional connectivity (FC) to characterize schizophrenia and psychosis. Toward this goal, a systematic literature review was conducted on primary empirical articles published in English in peer-reviewed journals between January 2019-February 2025 which utilized cortical-subcortical-cerebellar terminology to describe schizophrenia-control comparisons of whole-brain FC in human rs-fMRI. The electronic databases utilized included Google scholar, PubMed, and APA PsycInfo, and search terms included ("schizophrenia" OR "psychosis") AND "resting-state fMRI" AND ("cortical-subcortical-cerebellar" OR "cerebello-thalamo-cortical"). Ten studies were identified and NeuroMark nomenclature was utilized to describe findings within a common reference space. The most consistent patterns included cerebellar-thalamic hypoconnectivity, cerebellar-cortical (sensorimotor & insular-temporal) hyperconnectivity, subcortical (basal ganglia and thalamic)-cortical (sensorimotor, temporoparietal, insular-temporal, occipitotemporal, and occipital) hyperconnectivity, and cortical-cortical (insular-temporal and occipitotemporal) hypoconnectivity. Patterns implicating prefrontal cortex are largely inconsistent across studies and may not be effective targets for establishing stable imaging markers based on static FC in rs-fMRI. Instead, adapting new analytical strategies, or focusing on nodes in the cerebellum, thalamus, and primary motor and sensory cortex may prove to be a more effective approach.
Tractography is the only available technique for visualizing whitematter pathways within the living brain. Avoiding these pathways during surgical interventions for brain tumors and epilepsy is key to reducing postoperative neurological deficits whilst achieving maximum safe resection. Despite this, the use of intraoperative tractography is not widely adopted in clinical practice, with time required to run analyses often cited as a limitation. This systematic review and meta-analysis aimed to assess the impact of intraoperative tractography on neurosurgical outcomes in both tumor and epilepsy surgeries. Conducted in accordance with PRISMA guidelines, five major databases were searched using neurosurgery, tractography, brain tumor, and epilepsy terms. Original primary research studies in English were included. A risk of bias analysis was conducted using the MINORS tool. The search strategy identified 2,611 papers. Following de-duplication and screening, 26 papers were included in the final analysis. Risk of bias was found to be moderate. Findings suggest that the use of intraoperative tractography has the potential to improve surgical outcomes for patients undergoing tumor and epilepsy surgery. Meta-analysis indicated a good rate of gross total resection, 79%, and only three studies of brain tumors and one study of epilepsy reported worsening of neurological deficits. Though the evidence supporting its use remains limited, results indicate that intraoperative tractography can be a valuable tool in improving neurosurgical outcomes and reducing the risk of postoperative deficits. Further research is required to determine optimal use in clinical practice. https://www.crd.york.ac.uk/PROSPERO/view/CRD42023427427, Identifier: CRD42023427427.
White matter (WM) has traditionally been considered structurally important but functionally inert in fMRI research. However, growing evidence indicates that WM exhibits meaningful BOLD fluctuations and participates in functional connectivity. Here, we investigate alterations in WM functional network connectivity (FNC) across the Alzheimer's disease (AD) spectrum using resting-state fMRI data from the Alzheimer's Disease Neuroimaging Initiative (ADNI; 415 cognitively normal (CN), 283 mild cognitive impairment (MCI), 91 AD). We applied a guided independent component analysis (ICA) approach based on a combined multiscale template including 202 intrinsic connectivity networks [ICNs; 97 WM, 105 gray matter (GM)] to estimate subject-specific timecourses and compute FNC. Group differences in WM-WM, GM-GM, and WM-GM functional network connectivity (AD-CN, AD-MCI, MCI-CN) were evaluated using two-sample t-tests on residual FNC values for age, sex, and mean framewise displacement. Multiple comparisons across edges were controlled using false discovery rate correction (q < 0.05), and effect sizes were quantified using Hedges' g. Results showed robust alterations in WM-WM and WM-GM connectivity in AD, particularly involving WM subcortical, frontal, sensorimotor, and occipitotemporal networks. Several WM-GM interactions with cerebellar and hippocampal GM networks were also disrupted, including reduced GM-cerebellar: WM-frontal coupling and increased GM-hippocampal: WM-frontal connectivity. Notably, MCI already showed WM-GM dysconnectivity relative to CN, suggesting that functional disruption of WM circuits emerges prior to overt dementia. These findings provide converging evidence that WM functional connectivity is both measurable and selectively altered across the AD continuum. Our findings support WM FNC as a candidate biomarker to GM-based measures for staging and monitoring AD. Together, these results position WM-GM dysconnectivity as an important systems-level signature of the AD continuum and support WM functional network connectivity as a promising complement to established GM-based measures for understanding disease progression.
Schizophrenia (Sz) and autism spectrum disorder (ASD) are associated with reduced accuracy offace emotion recognition (FER). Nevertheless, the underlying pathophysiological mechanisms may diverge, potentially related to differential processing patterns within the early visual system. Here, we investigated physiological-level responses to emotional faces. We hypothesized that Sz and ASD would be associated with convergent behavioral performance, but divergent pathophysiological mechanisms. Simultaneous eye-tracking and continuous EEG data were obtained from 23 adults diagnosed with schizophrenia (Sz), 21 autistic adults, and 24 neurotypical controls (NC) in response to intact and chimeric emotion faces. Event-related potentials (ERP) were calculated from the ongoing EEG data using time- and time-frequency (TF) domain approaches. Symptoms were rated using the Positive and Negative Symptom Scale (PANSS) and the Autism Diagnostic Observation Schedule, Second Edition (ADOS-2) in Sz and ASD, respectively. As predicted, Sz and ASD were associated with similar levels of reduced FER accuracy relative to NC, but differential patterns of eye tracking and EEG-related activity. Rates of eye- vs. mouth-fixations were reduced across groups but did not correlate with FER. Nevertheless, the ability to utilize eye-information diverged across groups. Thus, when viewing chimeric faces, Sz was associated with reduced tendency to utilize eye information and increased tendency to utilize mouth information even when fixation location was considered. In TF analyses, reduced FER accuracy was associated with reduced initial sensory responses in Sz, as reflected in the theta-band time-frequency response. In contrast, in ASD, reduced FER accuracy was associated with increased alpha-frequency event-related desynchronization (alpha-ERD) consistent with hyper-engagement of secondary visual regions (V2). A combination of physiological and eye-tracking measures differentiated schizophrenia and ASD with >90% accuracy. V2 hyper-engagement in ASD correlated with both reduced FER accuracy and ADOS Social Interaction domain scores. Schizophrenia and ASD are associated with divergent physiological-level alterations within the early visual system during emotional face processing, supporting models of magnocellular visual hypoactivity in schizophrenia but retinotectal visual hyperactivity leading to hyper-engagement of non-face regions (V2) by face stimuli in ASD. These alterations, in turn, may serve as targets for future intervention studies related to social cognition.
The characteristic brain function and network activity patterns in adolescents with first-episode depression (FED) remain systematically underexplored. This study aims to investigate abnormalities in cerebral function and networks in adolescent FED patients through analyses of the amplitude of low-frequency fluctuations (ALFF), fractional amplitude of low-frequency fluctuations (fALFF), and independent component analysis (ICA). A cohort of 36 adolescents with first-episode depression (patient group, PT) and 34 healthy controls (HC group) were enrolled. Depressive symptoms were assessed using the Hamilton Depression Rating Scale (HAMD) and Children's Depression Inventory (CDI). All participants underwent resting-state functional magnetic resonance imaging (rs-fMRI). Neuronal activity and functional network alterations were analyzed via ALFF, fALFF, and ICA methodologies. Compared to the HC group, the PT group exhibited increased ALFF values in the left fusiform gyrus (Fusiform_L), left middle temporal gyrus (Temporal_Mid_L), right middle occipital gyrus (Occipital_Mid_R), right middle temporal gyrus (Temporal_Mid_R), right calcarine cortex (Calcarine_R), right angular gyrus (Angular_R), and left calcarine cortex (Calcarine_L). Elevated fALFF values were observed in the right calcarine cortex (Calcarine_R) and left superior temporal gyrus (Temporal_Sup_L), while decreased fALFF values were detected in the left superior temporal pole (Temporal_Pole_Sup_L), right medial superior frontal gyrus (Frontal_Sup_Medial_R), left superior frontal gyrus (Frontal_Sup_L), and left precuneus (Precuneus_L). Connectivity differences within the visual network (VIN) were identified between groups, with a peak difference in the right inferior temporal gyrus (Temporal_Inf_R), where the PT group demonstrated hyperconnectivity. In summary, neurofunctional abnormalities in adolescent FED patients involve the temporal lobe emotion-processing network, prefrontal executive control system, and default mode network (DMN). Aberrant low-frequency activity in the temporal pole and superior frontal gyrus may exacerbate emotion dysregulation, whereas hyperactivation of the precuneus and visual cortex could potentiate negative self-referential processing. Notably, the right middle occipital gyrus may represent a distinctive biomarker of adolescent depression. These findings provide novel insights into the early neural mechanisms underlying adolescent depression and suggest that non-invasive neuromodulation techniques targeting specific brain regions (e.g., transcranial magnetic stimulation, TMS) hold therapeutic potential.
Cerebral infarction is the most common neurological complication in patients with hypereosinophilic syndromes (HES), typically occurring in border-zone regions. However, intracranial artery stenosis is rarely observed in HES, and the underlying mechanisms of cerebral infarction remain largely unknown. Here, we report a case of HES complicated by acute ischemic stroke secondary to severe stenosis of left middle cerebral artery (MCA). A diagnosis of idiopathic HES was established based on eosinophilia (14.08%) in bone marrow aspiration and negative genetic testing. Without contraindications, intravenous thrombolysis with alteplase was administered, resulting in a decrease of the National Institutes of Health Stroke Scale score from 13 to 2. High-resolution magnetic resonance imaging (HR-MRI) showed homogeneous, concentric wall thickening and enhancement in the terminal segments of the left internal carotid artery and at the origin of the MCA, indicating an inflammatory process. Follow-up HR-MRI at 17 months demonstrated a reduction in vessel wall enhancement after immunosuppressive therapy. Over the two-year follow-up period, the eosinophil count remained within the range of 0.22-1.09 × 109/L, and no stroke recurrence was observed. In the literature review, only three cases of stroke associated with HES reported intracranial stenosis, all located in the M1 segment of the MCA. Their clinical outcomes improved following immunosuppressive therapy. Thus, intracranial large artery stenosis is a rare etiology of stroke in patients with HES. Homogeneous vessel wall enhancement on HR-MRI suggests an underlying vasculitis, which appears responsive to immunosuppressive therapy.
Diffusion MRI is increasingly used to study white-matter architecture, but tractography and diffusion metrics can be biased by different sampling schemes. We assessed systematic differences across four common protocols-single-shell high-angular resolution diffusion imaging (HARDI), Siemens clinical multi-shell (Sms), diffusion spectrum imaging (DSI), and Human Connectome Project multi-shell (HCPms)-in healthy adults and individuals with corpus callosum dysgenesis (CCD). All data were acquired on a single 3 T scanner and processed uniformly to extract fractional anisotropy (FA), mean diffusivity (MD), effective contrast-to-noise ratio (eCNR), and orientation dispersion within the corpus callosum (CC), corona radiata (CR), and centrum semiovale (CSO). In controls, we measured tract volumes for CC, bilateral CR, anterior commissure (AC) and posterior commissure (PC), and streamline counts for AC and PC; in CCD, we quantified volumes of the Probst and sigmoid bundles. Across participants, FA and MD showed moderate cross-scheme correlations for most ROIs, but matched means were rare (only Sms-HARDI in CC). eCNR and dispersion exhibited few cross-scheme correlations; however, means were similar for eCNR between Sms and HCPms and for dispersion among HARDI, DSI, and HCPms. Tract-based volumes correlated across Sms, DSI, and HCPms for CC in controls and for the right sigmoid and both Probst bundles in CCD. DSI and HCPms yielded similar volumes in all ROIs (controls and CCD). In controls, Sms volumes agreed with DSI/HCPms in CR, but were lower in CC and in all CCD ROIs. HARDI produced higher volumes in CC and bilateral CR in controls and in all CCD ROIs. For AC and PC in controls, tract-based means (volumes, streamlines, streamlines/volume) were consistent across schemes; nonetheless, correlations were limited-streamlines and streamlines/volume correlated for Sms, DSI, and HARDI in AC, and for DSI and HCPms in PC. These findings demonstrate systematic differences in voxel-wise metrics and tractography outcomes from four diffusion-sampling schemes. In addition to qualitatively informing attempts to consolidate or contrast data across schemes, future work could explore regression-based harmonization-and other methods-to reduce residual bias and enable pooled analyses across diverse protocols.
Voxel-based meta-analyses-also known as coordinate-based meta-analyses (CBMAs)-are powerful tools for synthesizing evidence from neuroimaging studies in human neuroscience, including investigations of psychological functions and differences in brain disorders. To achieve their full potential in accurately assessing the evidence, CBMAs should adhere to established best-practicmpe guidelines, such as the "Ten Simple Rules" published in 2018. Yet, even when studies report following these recommendations, the degree to which individual items are applicable or fully addressed is often unclear. To better support the evaluation of methodological rigor-which the 10 rules already promote but are not always consistently applied-, the developers of the most used CBMA methods followed a Delphi-style iterative process to create a reporting checklist focused on the methodological quality of CBMAs (Qual-CBMA). Qual-CBMA comprises criteria (e.g., preregistration, systematic search, homogeneous study characteristics, etc.) that authors should verify and comment on explicitly in the checklist (and, when unmet, also in the manuscript). The checklist encourages rigor and transparency by prompting authors to identify potential methodological limitations and to discuss their relevance-or irrelevance-in the context of their specific study. The checklist is designed as an aid to make reporting clearer and more transparent, not as a tool for evaluating whether authors have done something incorrectly. In this context, a high-quality CBMA is not defined by meeting every criterion, but by clearly commenting on the criteria-and explaining when unmet criteria are appropriately not applicable given the study's objectives. We encourage authors to submit the Qual-CBMA checklist, together with their accompanying comments, when publishing new CBMAs, thereby reinforcing transparency and rigorous methodology and advancing understanding in cognitive neuroscience and clinical conditions.
Fourier base fitting for masked or incomplete structured data holds significant importance, for example in biomedical image data processing. However, data incompleteness destroys the simple unitary form of the Fourier transformation, necessitating the construction and solving of a linear system-a task that can suffer from poor conditioning and be computationally expensive. Despite its importance, suitable methodology addressing this challenge is not readily available. In this study, we propose an efficient and fast Fourier base fitting method suitable for handling masked or incomplete structured data. The developed method can be used for processing multi-dimensional data, including smoothing and intra-/extrapolation, even when confronted with missing data. The developed method was verified using 1D, 2D, and 3D benchmarks. Its application is demonstrated in the reconstruction of noisy and partially unreliable brain pulsation data in the context of the development of a biomarker for non-invasive craniospinal compliance monitoring and neurological disease diagnostics. The study investigated the impact of different analytical and numerical performance improvement measures (e.g., term rearrangement, precomputation of recurring functions, vectorization) on computational complexity and speed. Quantitative evaluations on these benchmarks demonstrated that peak reconstruction errors in masked regions remained acceptable (i.e., below 10 % of the data range for all investigated benchmarks), while the proposed computational optimizations reduced matrix assembly time from 843 s to 11 s in 3D cases, demonstrating a 75-fold speed-up compared to unoptimized implementations. Singular value decomposition (SVD) can optionally be employed as part of the solving-step to provide regularization when needed. However, SVD quickly becomes the performance limiting in terms of computational complexity and resource cost, as the number of considered Fourier modes increases.
While risk factors have been identified for numerous psychiatric disorders, many individuals exposed to these risk factors do not develop psychopathology. A growing neuroimaging literature has sought to find structural and functional brain features that confer psychological resilience against developing psychiatric disorders. We conducted a systematic review and meta-analysis of neuroimaging studies associated with psychological resilience. Searches of Pubmed, Embase, Web of Science and PsychInfo yielded 2,658 potentially relevant articles published 2000-2021. Of these, we identified 154 human neuroimaging articles which provided anatomical coordinates of regions promoting resilience against psychiatric disorders including PTSD (44% of articles), schizophrenia (18%), major depressive disorder (14%) and bipolar disorder (12%). Meta-analysis conducted in GingerALE identified three regions as promoting psychological resilience across disorders (cluster-level FWE p < 0.05): left amygdala, right amygdala, and anterior cingulate. We additionally introduce a novel framework for conducting systematic reviews and meta-analyses that is compliant with best practices of Open Science: our publicly viewable systematic review was curated and annotated using the open-source reference manager Zotero, with customizable Python scripts for extracting curated data for meta-analyses. Our methodological pipeline not only permits independent replication of our findings but also supports customization for future neuroimaging meta-analyses.
Electroencephalography (EEG) source localization (SL) has shown potential for various applications, from epilepsy and seizure focus localization to psychiatric disorder evaluation. However, questions remain about its neurophysiological plausibility in real-world settings where only EEG signals are available without subject-specific anatomical information. This study investigates whether established pre-processing and source localization methods can produce neurophysiologically plausible activation patterns when applied to naturalistic EEG data without structural magnetic resonance imaging (MRI) or digitized electrode positions. Proven methods are aggregated into an end-to-end pipeline that includes automatic pre-processing, eLORETA for source estimation, and a shared forward model derived from the ICBM 2009c Nonlinear Symmetric template and its corresponding CerebrA atlas. The pipeline is validated using two distinct datasets: the Healthy Brain Network (HBN) dataset comparing resting and naturalistic video-watching states and the multi-session and multi-task EEG cognitive dataset (COGBCI) comparing different cognitive workload levels. The validation approach focuses on whether the reconstructed source activations exhibit expected neurophysiological patterns via permutation testing. Findings revealed significant differences between resting state and video-watching tasks, with greater activation in posterior regions during video-watching, consistent with known visual processing pathways. The cognitive workload analysis similarly showed progressive activation increases with task difficulty, mapping to regions associated with executive function. These results prove that established source localization methods can produce neurophysiologically plausible activation patterns without subject-specific information, highlighting the strengths and limitations of applying these methods to mid-length naturalistic EEG data. This research demonstrates the viability of template-based source analysis for research settings where individual structural imaging is unavailable or impractical.
Alzheimer's disease (AD) is a degenerative neurological disorder marked by cognitive decline and functional disability. Despite the extensive use of magnetic resonance imaging (MRI) in machine learning (ML)-based AD studies, the relative and combined contributions of MRI-derived morphometric (MO), microstructural (MS), and graph-theoretical (GT) features are still not well explored in a unified, comparative framework. It remains unclear whether adding multimodal MRI-derived features consistently improves the predictive performance of ML-based approaches for AD diagnosis and cognitive decline. Addressing this gap, this study systematically analyzed the individual (MO, MS, GT) and combined (MO+MS, MO+GT, MS+GT, MO+MS+GT) utility of MRI-based feature sets. We developed an ensemble-based ML framework with a nested cross-validation module for two key tasks: (i) Alzheimer's disease cognitive stage classification (DSC) and (ii) longitudinal cognitive decline prediction (LCDP) in terms of mini-mental state examination (MMSE) score. In this study, we conducted feature ablation and statistical analysis to evaluate performance improvements resulting from the incremental addition of feature sets. The results of the study indicated that the proposed ensemble-based ML approach achieved the best predictive performance (balanced accuracy [BACC]: 0.898 ± 0.051) using a combination of MO and MS feature sets for cognitively normal (CN) vs. AD dementia (CN-ADD). In contrast, the best results for mild cognitive impairment (MCI) vs. ADD (MCI-ADD) and CN-MCI were achieved using the MO feature set alone, with BACC of 0.769 ± 0.116 and 0.652 ± 0.044, respectively. Likewise, for the LCDP task, the MO-based ensemble learner achieved an R2 of 0.212 ± 0.177. These results demonstrate that MO features capture the most robust disease-related information, while multimodal integration offers task-specific and limited benefits. In addition, these findings demonstrate the potential of integrated MRI-derived features in ML frameworks for enhancing ADD diagnosis and cognitive decline prediction and underscore the importance of feature selection based on task complexity.
High-dose Methotrexate (HDMTX) can induce neurotoxicity, yet its impact on brain metabolism remains underexplored. This study aimed to assess short- and long-term brain metabolic changes post-HDMTX on 18F-FDG PET/MRI relative to baseline (pre-HDMTX) scans. In this IRB approved, retrospective study, we included 19 children and young adults (3 females and 16 males; age 17.9 ± 4.3 years), with lymphoma (n = 13) or osteosarcoma (n = 6). All patients underwent 18F-FDG PET/MRI before (baseline) and after HDMTX (>1000 mg/m2). Post-treatment scans were conducted ≤3 months (short-term group, n = 11) or >3 months (long-term group, n = 8) after completion of HDMTX and were compared with baseline scans. SUVmean and SUVmax of the whole brain cortex and six subregions were measured with PMOD software. A generalized linear regression model was used to evaluate post-pre-HDMTX SUV values differences in whole cortex with p < 0.05 and for with of different brain subregions, with p < 0.008 after Bonferroni correction. In the short-term group, compared with baseline, both SUVmean (pre-HDMTX vs. post-HDMTX: 5.06 ± 1.62 vs. 6.31 ± 1.71, p < 0.001) and SUVmax (9.16 ± 3.33 vs. 13.25 ± 3.35, p < 0.001) significantly increased in the whole cortex following HDMTX. In contrast, the long-term group showed no significant changes in SUVmean (6.31 ± 1.71 vs. 6.30 ± 1.54, p = 0.1) or SUVmax (12.01 ± 3.53 vs. 11.58 ± 3.07, p = 0.1) after HDMTX. 18F-FDG PET/MRI revealed short-term increases in brain metabolism post-HDMTX compared with baseline, possibly reflecting neuroinflammation. Long-term follow up scans revealed normalization of brain metabolism or decreased brain metabolism compared to baseline, the latter possibly indicating neurotoxicity.
Superficial Siderosis of the Central Nervous System is an infrequent neurological disorder resulting from hemosiderin deposition due to chronic and recurrent subarachnoid hemorrhage, leading to significant neurological impairments including sensorineural hearing loss, cerebellar ataxia, and pyramidal signs. This case report presents a 50-year-old male patient with a history of pituitary tumor surgery, manifesting progressive neurological symptoms over 2 years, thereby highlighting the potential long-term complications associated with SSCNS. The atypical clinical presentation, coupled with a surgical background, underscores the diagnostic challenges faced by clinicians, who may misattribute symptoms to more common neurological conditions. Advanced imaging modalities, particularly susceptibility-weighted imaging (SWI), have proven essential in enhancing the diagnostic accuracy for SSCNS, revealing characteristic patterns of iron deposition that are often subtle and can lead to delayed recognition. This case not only contributes to the existing literature by documenting a rare presentation of SSCNS but also emphasizes the necessity for increased awareness and vigilance among healthcare providers regarding this condition's complex manifestations. The findings advocate for interdisciplinary collaboration between neurologists and radiologists to improve recognition and management strategies, ultimately leading to better patient outcomes. Despite the rarity and variability of SSCNS, which complicates the establishment of standardized treatment protocols, this case highlights the critical need for continued research into its underlying mechanisms and therapy efficacy, particularly in patients with previous neurological interventions. Enhanced educational initiatives may be pivotal in addressing the diagnostic challenges associated with this debilitating condition.