Frontiers in neuroimaging最新文献

This study compares volumetric measurements of various brain regions using different magnetic resonance imaging (MRI) modalities and deep learning models, specifically 3T MRI, ultra-low field (ULF) MRI at 64mT, and AI-enhanced ULF MRI using SynthSR and HiLoResGAN. The aim is to evaluate the alignment and agreement among field strengths and ULF MRI with and without AI. Descriptive statistics, paired t-tests, effect size analyses, and regression analyses are employed to assess the relationships and differences between modalities. The results indicate that volumetric measurements derived from 64mT MRI deviate significantly from those obtained using 3T MRI. By leveraging SynthSR and LoHiResGAN models, these deviations are reduced, bringing the volumetric estimates closer to those obtained from 3T MRI, which serves as the reference standard for brain volume quantification. These findings highlight that deep learning models can reduce systematic differences in brain volume measurements across field strengths, providing potential solutions to minimize bias in imaging studies.

Introduction: 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.

Methods: 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.

Results: 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.

Discussion: 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.

To facilitate high spatial-temporal resolution fMRI (≦1mm³) at more broadly available field strengths (3T) and to better understand the neural underpinnings of joy, we used SE-based generalized Slice Dithered Enhanced Resolution (gSLIDER). This sequence increases SNR efficiency utilizing sub-voxel shifts along the slice direction. To improve the effective temporal resolution of gSLIDER, we utilized the temporal information within individual gSLIDER RF encodings to develop gSLIDER with Sliding Window Accelerated Temporal resolution (gSLIDER-SWAT). We first validated gSLIDER-SWAT using a classic hemifield checkerboard paradigm, demonstrating robust activation in primary visual cortex even with stimulus frequency increased to the Nyquist frequency of gSLIDER (i.e., TR = block duration). gSLIDER provided ~2× gain in tSNR over traditional SE-EPI. GLM and ICA results suggest improved signal detection with gSLIDER-SWAT's nominal 5-fold higher temporal resolution that was not seen with simple temporal interpolation. Next, we applied gSLIDER-SWAT to investigate the neural networks underlying joy using naturalistic video stimuli. Regions significantly activated during joy included the left amygdala, specifically the basolateral subnuclei, and rostral anterior cingulate, both part of the salience network; the hippocampus, involved in memory; the striatum, part of the reward circuit; prefrontal cortex, part of the executive network and involved in emotion processing and regulation [bilateral mPFC/BA10/11, left MFG (BA46)]; and throughout visual cortex. This proof of concept study demonstrates the feasibility of measuring the networks underlying joy at high resolutions at 3T with gSLIDER-SWAT, and highlights the importance of continued innovation of imaging techniques beyond the limits of standard GE fMRI.

Several decades of research have investigated the neural connections between stroke-induced brain damage and language difficulties. Typically, lesion-symptom mapping (LSM) studies that address this connection have relied on mass univariate statistics, which do not account for multidimensional relationships between variables. Machine learning (ML) techniques, which can capture these intricate connections, offer a promising complement to LSM methods. To test this promise, we benchmarked ML models on structural and functional MRI to predict aphasia severity (N = 238) and naming impairment (N = 191) for a cohort of chronic-stage stroke survivors. We used nested cross-validation to examine performance along three dimensions: (1) parcellation schemes (JHU, AAL, BRO, and AICHA atlases), (2) neuroimaging modalities (resting-state functional connectivity, structural connectivity, mean diffusivity, fractional anisotropy, and lesion location) and (3) ML methods (Random Forest, Support Vector Regression, Decision Tree, K Nearest Neighbors, and Gradient Boosting). The best results were obtained by combining the JHU atlas, lesion location, and the Random Forest model. This combination yielded moderate to high correlations with the two different behavioral scores. Key regions identified included several perisylvian areas and pathways within the language network. This work complements existing LSM methods with new tools for improving the prediction of language outcomes in stroke survivors.

Introduction: Low-intensity focused ultrasound (LIFU) is a form of neuromodulation that offers increased depth of penetrance and improved spatial resolution over other non-invasive techniques, allowing for modulation of otherwise inaccessible subcortical structures that are implicated in neuropsychiatric pathologies. The amygdala is a target of great interest due to its involvement in numerous psychiatric conditions. While prior works have found that LIFU sonication of the amygdala can alter resting-state neural activation, only a few studies have investigated whether LIFU can selectively modulate the amygdala during task-based fMRI.

Methods: We aimed to address these gaps in literature in a cohort of 10 healthy individuals. We utilized the well-validated Emotional Face Assessment Task (EFAT), which is designed to robustly engage the amygdala. We selected the fusiform gyrus and the thalamus as our non-target regional comparison measures due to their roles in facial and emotional processing. In succession, participants completed a pre-LIFU baseline fMRI, received 10-min of LIFU neuromodulation, and then repeated the baseline fMRI. To test our hypothesis, we conducted paired-samples t-tests assessing changes in amygdala, fusiform gyrus, and thalamic activation from pre to post scan.

Results: We found that there was a significant decrease in left (t(9) = 2.286; p = 0.024) and right (t(9) = 2.240; p = 0.026) amygdala activation from pre-to-post sonication.

Discussion: Meanwhile, there were no differences in activation of the left or right fusiform gyrus or thalamus. Our results indicate that LIFU of the amygdala acutely dampens amygdala reactivity during active socio-emotional processing.

Introduction: 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.

Methods: 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%).

Results: 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.

Discussion: 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.

Low-field MRI is gaining interest, especially in low-resource settings, due to its low cost, portability, small footprint, and low power consumption. However, it suffers from significant noise, limiting its clinical utility. This study introduces native noise denoising (NND), which leverages the inherent noise characteristics of the acquired low-field data. By obtaining the noise characteristics from corner patches of low-field images, we iteratively added similar noise to high-field images to create a paired noisy-clean dataset. A U-Net based denoising autoencoder was trained on this dataset and evaluated on three low-field datasets: the M4Raw dataset (0.3T), in vivo brain MRI (0.05T), and phantom images (0.05T). The NND approach demonstrated improvements in signal-to-noise ratio (SNR) of 32.76%, 19.02%, and 8.16% across the M4Raw, in vivo and phantom datasets, respectively. Qualitative assessments, including difference maps, line intensity plots, and effective receptive fields, suggested that NND preserves structural details and edges compared to random noise denoising (RND), indicating potential enhancements in visual quality. This substantial improvement in low-field imaging quality addresses the fundamental challenge of diagnostic confidence in resource-constrained settings. By mitigating the primary technical limitation of these systems, our approach expands the clinical utility of low-field MRI scanners, potentially facilitating broader access to diagnostic imaging across resource-limited healthcare environments globally.

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.

Introduction: Functional connectivity (FC) is a metric of how different brain regions interact with each other. Although there have been some studies correlating learning and memory with FC, there have not yet been, to date, studies that use machine learning (ML) to explain how FC changes can be used to explain behavior not only in healthy mice, but also in mouse models of Alzheimer's Disease (AD). Here, we investigated changes in FC and their relationship to learning and memory in a mouse model of AD across disease progression.

Methods: We assessed the APP/PS1 mouse model of AD and wild-type controls at 3-, 6-, and 10-months of age. Using resting state functional magnetic resonance imaging (rs-fMRI) in awake, unanesthetized mice, we assessed FC between 30 brain regions. ML models were then used to define interactions between neuroimaging readouts with learning and memory performance.

Results: In the APP/PS1 mice, we identified a pattern of hyperconnectivity across all three time points, with 47 hyperconnected regions at 3 months, 46 at 6 months, and 84 at 10 months. Notably, FC changes were also observed in the Default Mode Network, exhibiting a loss of hyperconnectivity over time. Modeling revealed functional connections that support learning and memory performance differ between the 6- and 10-month groups.

Discussion: These ML models show potential for early disease detection by identifying connectivity patterns associated with cognitive decline. Additionally, ML may provide a means to begin to understand how FC translates into learning and memory performance.

Introduction: Anatomical variations in the posterior horns of the lateral ventricles are well-documented, with the horn presenting as open, constricted, or completely closed. However, the extent and nature of these variations across different demographics remain under-explored. This study aimed to investigate the anatomical variations of the posterior horn of the lateral ventricles across different age and sex groups and to compare the variations between the right and left sides.

Methods: We conducted a retrospective analysis of magnetic resonance imaging (MRI) scans from 217 adult participants across 15 age groups, utilizing a stratified random sampling from a radiology database. MRI scans were analyzed for ventricular dimensions, and horn types (open, constricted, and closed). Statistical significance was defined as p-value < 0.05.

Results: Variants of the posterior horn were observed frequently, with open posterior horn being the most common in the left lateral ventricle (41%) and constricted type being the most common in the right lateral ventricle (37%). A significant correlation existed between the right and left horn types, but in most cases, there was a difference in type between the right and the left horns in the same individual. No significant association between age and the type of the posterior horns was found. However, there was a significant difference in the width and length of the horns between the open and other types, with open horns being wider and longer. Lastly, the left horn appeared longer than the right one.

Discussion: The findings underline the high variability in posterior horn morphology, which is not significantly influenced by age or sex but varies between individuals and sides. Future studies should explore the functional impact of these anatomical variations.