Search

Experimental blood flow measurement techniques are invaluable for a better understanding of cardiovascular disease formation, progression, and treatment. One of the emerging methods is time‐resolved three‐dimensional phase‐contrast magnetic resonance imaging (4D flow MRI), which enables noninvasive time‐dependent velocity measurements within large vessels. However, several limitations hinder the usability of 4D flow MRI and other experimental methods for quantitative hemodynamics analysis. These mainly include measurement noise, corrupt or missing data, low spatiotemporal resolution, and other artifacts. Traditional filtering is routinely applied for denoising experimental blood flow data without any detailed discussion on why it is preferred over other methods. In this study, filtering is compared to different singular value decomposition (SVD)‐based machine learning and autoencoder‐type deep learning methods for denoising and filling in missing data (imputation). An artificially corrupted and voxelized computational fluid dynamics (CFD) simulation as well as in vitro 4D flow MRI data are used to test the methods. SVD‐based algorithms achieve excellent results for the idealized case but severely struggle when applied to in vitro data. The autoencoders are shown to be versatile and applicable to all investigated cases. For denoising, the in vitro 4D flow MRI data, the denoising autoencoder (DAE), and the Noise2Noise (N2N) autoencoder produced better reconstructions than filtering both qualitatively and quantitatively. Deep learning methods such as N2N can result in noise‐free velocity fields even though they did not use clean data during training. This work presents one of the first comprehensive assessments and comparisons of various classical and modern machine‐learning methods for enhancing corrupt cardiovascular flow data in diseased arteries for both synthetic and experimental test cases. [ABSTRACT FROM AUTHOR]

DNA‐based data storage is a new technology in computational and synthetic biology, that offers a solution for long‐term, high‐density data archiving. Given the critical importance of medical data in advancing human health, there is a growing interest in developing an effective medical data storage system based on DNA. Data integrity, accuracy, reliability, and efficient retrieval are all significant concerns. Therefore, this study proposes an Effective DNA Storage (EDS) approach for archiving medical MRI data. The EDS approach incorporates three key components (i) a novel fraction strategy to address the critical issue of rotating encoding, which often leads to data loss due to single base error propagation; (ii) a novel rule‐based quaternary transcoding method that satisfies bio‐constraints and ensure reliable mapping; and (iii) an indexing technique designed to simplify random search and access. The effectiveness of this approach is validated through computer simulations and biological experiments, confirming its practicality. The EDS approach outperforms existing methods, providing superior control over bio‐constraints and reducing computational time. The results and code provided in this study open new avenues for practical DNA storage of medical MRI data, offering promising prospects for the future of medical data archiving and retrieval. [ABSTRACT FROM AUTHOR]

Background: A variety of deep learning‐based and iterative approaches are available to predict Tracer Kinetic (TK) parameters from fully sampled or undersampled dynamic contrast‐enhanced (DCE) MRI data. However, both the methods offer distinct benefits and drawbacks. Purpose: To propose a hybrid algorithm (named as 'Greybox'), using both model‐ as well as DL‐based, for solving a multi‐parametric non‐linear inverse problem of directly estimating TK parameters from undersampled DCE MRI data, which is invariant to undersampling rate. Methods: The proposed algorithm was inspired by plug‐and‐play algorithms used for solving linear inverse imaging problems. This technique was tested for its effectiveness in solving the nonlinear ill‐posed inverse problem of generating 3D TK parameter maps from four‐dimensional (4D; Spatial + Temporal) retrospectively undersampled k‐space data. The algorithm learns a deep learning‐based prior using UNET to estimate the Ktrans$\mathbf {K_{trans}}$ and Vp$\mathbf {V_{p}}$ parameters based on the Patlak pharmacokinetic model, and this trained prior was utilized to estimate the TK parameter maps using an iterative gradient‐based optimization scheme. Unlike the existing DL models, this network is invariant to the undersampling rate of the input data. The proposed method was compared with the total variation‐based direct reconstruction technique on brain, breast, and prostate DCE‐MRI datasets for various undersampling rates using the Radial Golden Angle (RGA) scheme. For the breast dataset, an indirect estimation using the Fast Composite Splitting algorithm was utilized for comparison. Undersampling rates of 8×$\times$, 12×$\times$ and 20×$\times$ were used for the experiments, and the results were compared using the PSNR and SSIM as metrics. For the breast dataset of 10 patients, data from four patients were utilized for training (1032 samples), two for validation (752 samples), and the entire volume of four patients for testing. Similarly, for the prostate dataset of 18 patients, 10 patients were utilized for training (720 samples), five for validation (216 samples), and the whole volume of three patients for testing. For the brain dataset of nineteen patients, ten patients were used for training (3152 samples), five for validation (1168 samples), and the whole volume of four patients for testing. Statistical tests were also conducted to assess the significance of the improvement in performance. Results: The experiments showed that the proposed Greybox performs significantly better than other direct reconstruction methods. The proposed algorithm improved the estimated Ktrans$\mathbf {K_{trans}}$ and Vp$\mathbf {V_{p}}$ in terms of the peak signal‐to‐noise ratio by up to 3 dB compared to other standard reconstruction methods. Conclusion: The proposed hybrid reconstruction algorithm, Greybox, can provide state‐of‐the‐art performance in solving the nonlinear inverse problem of DCE‐MRI. This is also the first of its kind to utilize convolutional neural network‐based encodings as part of the plug‐and‐play priors to improve the performance of the reconstruction algorithm. [ABSTRACT FROM AUTHOR]

Energy landscape analysis is a data-driven method to analyse multidimensional time series, including functional magnetic resonance imaging (fMRI) data. It has been shown to be a useful characterization of fMRI data in health and disease. It fits an Ising model to the data and captures the dynamics of the data as movement of a noisy ball constrained on the energy landscape derived from the estimated Ising model. In the present study, we examine test-retest reliability of the energy landscape analysis. To this end, we construct a permutation test that assesses whether or not indices characterizing the energy landscape are more consistent across different sets of scanning sessions from the same participant (i.e. within-participant reliability) than across different sets of sessions from different participants (i.e. between-participant reliability). We show that the energy landscape analysis has significantly higher within-participant than between-participant test-retest reliability with respect to four commonly used indices. We also show that a variational Bayesian method, which enables us to estimate energy landscapes tailored to each participant, displays comparable test-retest reliability to that using the conventional likelihood maximization method. The proposed methodology paves the way to perform individual-level energy landscape analysis for given data sets with a statistically controlled reliability., (© 2024 The Authors. European Journal of Neuroscience published by Federation of European Neuroscience Societies and John Wiley & Sons Ltd.)

Brain gliomas, common in adults, pose significant diagnostic challenges. Accurate segmentation from multimodal magnetic resonance imaging (MRI) scans is critical for effective treatment planning. Traditional manual segmentation methods, labor‐intensive and error‐prone, often lead to inconsistent diagnoses. To overcome these limitations, our study presents a sophisticated framework for the automated segmentation of brain gliomas from multimodal MRI images. This framework consists of three integral components: a 3D UNet, a classifier, and a Classifier Weight Transformer (CWT). The 3D UNet, acting as both an encoder and decoder, is instrumental in extracting comprehensive features from MRI scans. The classifier, employing a streamlined 1 × 1 convolutional architecture, performs detailed pixel‐wise classification. The CWT integrates self‐attention mechanisms through three linear layers, a multihead attention module, and layer normalization, dynamically refining the classifier's parameters based on the features extracted by the 3D UNet, thereby improving segmentation accuracy. Our model underwent a two‐stage training process for maximum efficiency: in the first stage, supervised learning was used to pre‐train the encoder and decoder, focusing on robust feature representation. In the second stage, meta‐training was applied to the classifier, with the encoder and decoder remaining unchanged, ensuring precise fine‐tuning based on the initially developed features. Extensive evaluation of datasets such as BraTS2019, BraTS2020, BraTS2021, and a specialized private dataset (ZZU) underscored the robustness and clinical potential of our framework, highlighting its superiority and competitive advantage over several state‐of‐the‐art approaches across various segmentation metrics in training and validation sets. [ABSTRACT FROM AUTHOR]

Purpose: The interpretation of prostate multiparametric magnetic resonance imaging (MRI) is subjective in nature, and there is large inter‐observer variability among radiologists and up to 30% of clinically significant cancers are missed. This has motivated the development of new MRI techniques and sequences, especially quantitative approaches to improve prostate cancer diagnosis. Using hybrid multidimensional MRI, apparent diffusion coefficient (ADC) and T2 have been shown to change as a function of echo time (TE) and b‐values, and that this dependence is different for cancer and benign tissue, which can be exploited for prostate cancer diagnosis. The purpose of this study is to investigate whether four‐quadrant vector mapping of hybrid multidimensional MRI (HM‐MRI) data can be used to diagnose prostate cancer (PCa) and determine cancer aggressiveness. Methods: Twenty‐one patients with confirmed PCa underwent preoperative MRI prior to radical prostatectomy. Axial HM‐MRI were acquired with all combinations of TE = 47, 75, 100 ms and b‐values of 0, 750, 1500 s/mm2, resulting in a 3 × 3 data matrix associated with each voxel. Prostate Quadrant (PQ) mapping analysis represents HM‐MRI data for each voxel as a color‐coded vector in the four‐quadrant space of HM‐MRI parameters (a 2D matrix of signal values for each combination of b‐value and TE) with associated amplitude and angle information representing the change in T2 and ADC as a function of b‐value and TE, respectively. Results: Cancers have a higher PQ4 (22.50% ± 21.27%) and lower PQ2 (69.86% ± 28.24%) compared to benign tissue: peripheral, transition, and central zone (PQ4 = 0.13% ± 0.56%, 5.73% ± 15.07%, 2.66% ± 4.05%, and PQ2 = 98.51% ± 3.05%, 86.18% ± 21.75%, 93.38% ± 9.88%, respectively). Cancers have a higher vector angle (206.5 ± 41.8°) and amplitude (0.017 ± 0.013) compared to benign tissue. PQ metrics showed moderate correlation with Gleason score (|ρ| = 0.388–0.609), with more aggressive cancers being associated with increased PQ4 and angle and reduced PQ2 and amplitude. A combination of four‐quadrant analysis metrics provided an area under the curve of 0.904 (p < 0.001) for the differentiation of prostate cancer from benign prostatic tissue. Conclusions: Four‐quadrant vector mapping of HM‐MRI data provides effective cancer markers, with cancers associated with high PQ4 and high vector angle and lower PQ2 and vector amplitude. [ABSTRACT FROM AUTHOR]

Combining magnetic resonance imaging (MRI) data from multi-site studies is a popular approach for constructing larger datasets to greatly enhance the reliability and reproducibility of neuroscience research. However, the scanner/site variability is a significant confound that complicates the interpretation of the results, so effective and complete removal of the scanner/site variability is necessary to realise the full advantages of pooling multi-site datasets. Independent component analysis (ICA) and general linear model (GLM) based harmonisation methods are the two primary methods used to eliminate scanner/site effects. Unfortunately, there are challenges with both ICA-based and GLM-based harmonisation methods to remove site effects completely when the signals of interest and scanner/site effects-related variables are correlated, which may occur in neuroscience studies. In this study, we propose an effective and powerful harmonisation strategy that implements dual projection (DP) theory based on ICA to remove the scanner/site effects more completely. This method can separate the signal effects correlated with site variables from the identified site effects for removal without losing signals of interest. Both simulations and vivo structural MRI datasets, including a dataset from Autism Brain Imaging Data Exchange II and a travelling subject dataset from the Strategic Research Program for Brain Sciences, were used to test the performance of a DP-based ICA harmonisation method. Results show that DP-based ICA harmonisation has superior performance for removing site effects and enhancing the sensitivity to detect signals of interest as compared with GLM-based and conventional ICA harmonisation methods., (© 2023 Federation of European Neuroscience Societies and John Wiley & Sons Ltd.)

Summary: Alzheimer's Disease (AD) is the leading cause of dementia and impairment in various domains. Recent AD studies, (ie, Alzheimer's Disease Neuroimaging Initiative (ADNI) study), collect multimodal data, including longitudinal neurological assessments and magnetic resonance imaging (MRI) data, to better study the disease progression. Adopting early interventions is essential to slow AD progression for subjects with mild cognitive impairment (MCI). It is of particular interest to develop an AD predictive model that leverages multimodal data and provides accurate personalized predictions. In this article, we propose a multivariate functional mixed model with MRI data (MFMM‐MRI) that simultaneously models longitudinal neurological assessments, baseline MRI data, and the survival outcome (ie, dementia onset) for subjects with MCI at baseline. Two functional forms (the random‐effects model and instantaneous model) linking the longitudinal and survival process are investigated. We use Markov Chain Monte Carlo (MCMC) method based on No‐U‐Turn Sampling (NUTS) algorithm to obtain posterior samples. We develop a dynamic prediction framework that provides accurate personalized predictions of longitudinal trajectories and survival probability. We apply MFMM‐MRI to the ADNI study and identify significant associations among longitudinal outcomes, MRI data, and the risk of dementia onset. The instantaneous model with voxels from the whole brain has the best prediction performance among all candidate models. The simulation study supports the validity of the estimation and dynamic prediction method. [ABSTRACT FROM AUTHOR]

Magnetic resonance imaging (MRI) has been one of the primary instruments to measure the properties of the human brain non‐invasively in vivo. MRI data generally needs to go through a series of processing steps (i.e., a pipeline) before statistical analysis. Currently, the processing pipelines for multi‐modal MRI data are still rare, in contrast to single‐modal pipelines. Furthermore, the reliability and validity of the output of the pipelines are critical for the MRI studies. However, the reliability and validity measures are not available or adequate for almost all pipelines. Here, we present PhiPipe, a multi‐modal MRI processing pipeline. PhiPipe could process T1‐weighted, resting‐state BOLD, and diffusion‐weighted MRI data and generate commonly used brain features in neuroimaging. We evaluated the test–retest reliability of PhiPipe's brain features by computing intra‐class correlations (ICC) in four public datasets with repeated scans. We further evaluated the predictive validity by computing the correlation of brain features with chronological age in three public adult lifespan datasets. The multivariate reliability and predictive validity of the PhiPipe results were also evaluated. The results of PhiPipe were consistent with previous studies, showing comparable or better reliability and validity when compared with two popular single‐modality pipelines, namely DPARSF and PANDA. The publicly available PhiPipe provides a simple‐to‐use solution to multi‐modal MRI data processing. The accompanied reliability and validity assessments could help researchers make informed choices in experimental design and statistical analysis. Furthermore, this study provides a framework for evaluating the reliability and validity of image processing pipelines. [ABSTRACT FROM AUTHOR]

Background: Movement disorders such as Parkinson's disease are associated with structural and functional changes in specific brain regions. Advanced magnetic resonance imaging (MRI) techniques combined with machine learning (ML) are promising tools for identifying imaging biomarkers and patterns associated with these disorders. Purpose/Hypothesis: We aimed to systematically identify the brain regions most commonly affected in movement disorders using ML approaches applied to structural and functional MRI data. We searched the PubMed and Scopus databases using relevant keywords up to June 2023 for studies that used ML approaches to detect brain regions associated with movement disorders using MRI data. Study Type: A systematic review and diagnostic meta‐analysis. Population/Subjects: Sixty‐seven studies with 6,285 patients were included. Field Strength/Sequence: Studies utilizing 1.5T or 3T MR scanners and the acquisition of diffusion tensor imaging (DTI), structural MRI (sMRI), functional MRI (fMRI), or a combination of these were included. Assessment: The authors independently assessed the study quality using the CLAIM and QUADAS‐2 criteria and extracted data on diagnostic accuracy measures. Statistical Tests: Sensitivity, specificity, accuracy, and area under the curve were pooled using random‐effects models. Q statistics and the I2 index were used to evaluate heterogeneity, and Begg's funnel plot was used to identify publication bias. Results: sMRI showed the highest sensitivity (93%) and mixed modalities had the highest specificity (90%) for detecting regional abnormalities. sMRI had a 94% sensitivity for identifying subcortical changes. The support vector machine (93%) and logistic regression (91%) models exhibited high diagnostic accuracies. Data Conclusion: The combination of advanced MR neuroimaging techniques and ML is a promising approach for identifying brain biomarkers and affected regions in movement disorders with subcortical structures frequently implicated. Structural MRI, in particular, showed strong performance. Level of Evidence: 1 Technical Efficacy: Stage 2 [ABSTRACT FROM AUTHOR]

Purpose: To develop a fast denoising framework for high‐dimensional MRI data based on a self‐supervised learning scheme, which does not require ground truth clean image. Theory and Methods: Quantitative MRI faces limitations in SNR, because the variation of signal amplitude in a large set of images is the key mechanism for quantification. In addition, the complex non‐linear signal models make the fitting process vulnerable to noise. To address these issues, we propose a fast deep‐learning framework for denoising, which efficiently exploits the redundancy in multidimensional MRI data. A self‐supervised model was designed to use only noisy images for training, bypassing the challenge of clean data paucity in clinical practice. For validation, we used two different datasets of simulated magnetization transfer contrast MR fingerprinting (MTC‐MRF) dataset and in vivo DWI image dataset to show the generalizability. Results: The proposed method drastically improved denoising performance in the presence of mild‐to‐severe noise regardless of noise distributions compared to previous methods of the BM3D, tMPPCA, and Patch2self. The improvements were even pronounced in the following quantification results from the denoised images. Conclusion: The proposed MD‐S2S (Multidimensional‐Self2Self) denoising technique could be further applied to various multi‐dimensional MRI data and improve the quantification accuracy of tissue parameter maps. [ABSTRACT FROM AUTHOR]

Background: Early biomarkers are needed to identify individuals at high risk of preclinical Alzheimer's disease (AD) (Jack et al., 2018). Electroencephalography (EEG) is a non‐invasive and cheap technique that would be an interesting screening tool for the preclinical stage of AD. Method: We included participants from the INSIGHT‐preAD cohort, which is an ongoing single‐center multimodal observational study designed to identify risk factors and markers of progression to clinical AD in 318 cognitively normal individuals aged 70–85 years with a subjective memory complaint (Dubois et al., 2018). We divided the subjects into four groups, according to their amyloid status (on 18F‐florbetapir PET) and neurodegeneration status (on 18F‐fluorodeoxyglucose PET). We analysed 314 baseline 256‐channel high‐density eyes‐closed 1‐minute resting‐state EEG recordings. We extracted 10 quantitative EEG biomarkers, including spectral measures, algorithmic complexity and functional connectivity metrics. We evaluated three different classifiers (random forest, decision tree and logistic regression) to predict amyloid and neurodegeneration status from multimodal data, combining EEG, ApoE4 genotype, demographic, neuropsychological and MRI data. We divided the dataset into a training set (n=258) and validation set (n=46). We used a 5‐fold cross‐validation on the training set to identify the optimal combination of features. Result: As recently published (Gaubert et al., 2019), the most prominent effects of neurodegeneration on EEG metrics were localized in fronto‐central regions with an increase in high‐frequency oscillations (higher beta and gamma power) and a decrease in low‐frequency oscillations (lower delta power), higher spectral entropy, higher complexity and increased functional connectivity measured by weighted Symbolic Mutual Information in theta band. In the machine learning analysis, among the different features, EEG was the most strongly predictive of neurodegeneration. Similar predictive performance was obtained when reducing the number of electrodes from 256 to 8. EEG biomarkers combined with hippocampal volumetry had a high negative predictive value (91%) and high sensitivity (82%) to predict neurodegeneration. The combination of ApoE4 genotype with demographic data was most strongly predictive of amyloid status. Conclusion: EEG could be a valuable tool to screen for or rule out neurodegeneration in elderly memory complainers. This procedure has been patented under the following PCT number: PCT/EP2019/086629. [ABSTRACT FROM AUTHOR]

Purpose: To propose a robust time and space invariant deep learning (DL) method to directly estimate the pharmacokinetic/tracer kinetic (PK/TK) parameters from undersampled dynamic contrast-enhanced (DCE) magnetic resonance imaging (MRI) data. Methods: DCE-MRI consists of 4D (3D-spatial + temporal) data and has been utilized to estimate 3D (spatial) tracer kinetic maps.Existing DL architecture for this task needs retraining for variation in temporal and/or spatial dimensions. This work proposes a DL algorithm that is invariant to training and testing in both temporal and spatial dimensions. The proposed network was based on a 2.5-dimensional Unet architecture,where the encoder consists of a 3D convolutional layer and the decoder consists of a 2D convolutional layer. The proposed VTDCE-Net was evaluated for solving the ill-posed inverse problem of directly estimating TK parameters from undersampled k − t space data of breast cancer patients, and the results were systematically compared with a total variation (TV) regularization based direct parameter estimation scheme. In the breast dataset, the training was performed on patients with 32 time samples, and testing was carried out on patients with 26 and 32 time samples. Translation of the proposed VTDCE-Net for brain dataset to show the generalizability was also carried out. Undersampling rates (R) of 8×, 12×, and 20× were utilized with PSNR and SSIM as the figures of merit in this evaluation. TK parameter maps estimated from fully sampled data were utilized as ground truth. Results: Experiments carried out in this work demonstrate that the proposed VTDCE-Net outperforms the TV scheme on both breast and brain datasets across all undersampling rates. For Ktrans and Vp maps, the improvement over TV is as high as 2 and 5 dB, respectively, using the proposed VTDCE-Net. Conclusion: Temporal points invariant DL network that was proposed in this work to estimate the TK-parameters using DCE-MRI data has provided state-of-the-art performance compared to standard image reconstruction methods and is shown to work across all undersampling rates. [ABSTRACT FROM AUTHOR]

Machine-learning prediction studies have shown potential to inform treatment stratification, but recent efforts to predict psychotherapy outcomes with clinical routine data have only resulted in moderate prediction accuracies. Neuroimaging data showed promise to predict treatment outcome, but previous prediction attempts have been exploratory and reported small clinical sample sizes. Herein, we aimed to examine the incremental predictive value of neuroimaging data in contrast to clinical and demographic data alone (for which results were previously published), using a two-level multimodal ensemble machine-learning strategy. We used pretreatment structural and task-based fMRI data to predict virtual reality exposure therapy outcome in a bicentric sample of N = 190 patients with spider phobia. First, eight 1st-level random forest classifications were conducted using separate data modalities (clinical questionnaire scores and sociodemographic data, cortical thickness and gray matter volumes, functional activation, connectivity, connectivity-derived graph metrics, and BOLD signal variance). Then, the resulting predictions were used to train a 2nd-level classifier that produced a final prediction. No 1st-level or 2nd-level classifier performed above chance level except BOLD signal variance, which showed potential as a contributor to higher-level prediction from multiple regions across the brain (1st-level balanced accuracy = 0.63). Overall, neuroimaging data did not provide any incremental accuracy for treatment outcome prediction in patients with spider phobia with respect to clinical and sociodemographic data alone. Thus, we advise caution in the interpretation of prediction performances from small-scale, single-site patient samples. Larger multimodal datasets are needed to further investigate individual-level neuroimaging predictors of therapy response in anxiety disorders. [ABSTRACT FROM AUTHOR]

Background and Purpose: Resting‐state functional magnetic resonance imaging (rsfMRI) has been proposed as an alternative to task‐based fMRI including clinical situations such as preoperative brain tumor planning, due to advantages including ease of performance and time savings. However, one of its drawbacks is the limited ability to accurately lateralize language function. Methods: Using the rsfMRI data of healthy controls, we carried out a power spectra analysis on three regions of interest (ROIs): Broca's area (BA) in the frontal cortex for language, hand motor (HM) area in the primary motor cortex, and the primary visual cortex (V1). Spike removal, motion correction, linear trend removal, and spatial smoothing were applied. Spontaneous low‐frequency fluctuations (0.01‐0.1 Hz) were filtered to enable functional integration. Results: BA showed greater power on the left hemisphere relative to the right (p =.0055), while HM (p =.1563) and V1 (p =.4681) were not statistically significant. A novel index, termed the power laterality index (PLI), computed to estimate the degree of power lateralization for each brain region, revealed a statistically significant difference between BA and V1 (p <.00001), where V1 was used as a control since the primary visual cortex does not lateralize. Validation studies used to compare PLI to a laterality index computed using phonemic fluency, a task‐based, language fMRI paradigm, demonstrated good correlation. Conclusions: The power spectra for BA revealed left language lateralization, which was not replicated in HM or V1. This work demonstrates the feasibility and validity of an ROI‐based power spectra analysis on rsfMRI data for language lateralization. [ABSTRACT FROM AUTHOR]

Diffusion‐weighted magnetic resonance imaging (DW‐MRI) has evolved to provide increasingly sophisticated investigations of the human brain's structural connectome in vivo. Restriction spectrum imaging (RSI) is a method that reconstructs the orientation distribution of diffusion within tissues over a range of length scales. In its original formulation, RSI represented the signal as consisting of a spectrum of Gaussian diffusion response functions. Recent technological advances have enabled the use of ultra‐high b‐values on human MRI scanners, providing higher sensitivity to intracellular water diffusion in the living human brain. To capture the complex diffusion time dependence of the signal within restricted water compartments, we expand upon the RSI approach to represent restricted water compartments with non‐Gaussian response functions, in an extended analysis framework called linear multi‐scale modeling (LMM). The LMM approach is designed to resolve length scale and orientation‐specific information with greater specificity to tissue microstructure in the restricted and hindered compartments, while retaining the advantages of the RSI approach in its implementation as a linear inverse problem. Using multi‐shell, multi‐diffusion time DW‐MRI data acquired with a state‐of‐the‐art 3 T MRI scanner equipped with 300 mT/m gradients, we demonstrate the ability of the LMM approach to distinguish different anatomical structures in the human brain and the potential to advance mapping of the human connectome through joint estimation of the fiber orientation distributions and compartment size characteristics. [ABSTRACT FROM AUTHOR]

Deep learning presents a generalizable solution for motion correction requiring no pulse sequence modifications or additional hardware, but previous networks have all been applied to coil‐combined data. Multichannel MRI data provide a degree of spatial encoding that may be useful for motion correction. We hypothesize that incorporating deep learning for motion correction prior to coil combination will improve results. A conditional generative adversarial network was trained using simulated rigid motion artifacts in brain images acquired at multiple sites with multiple contrasts (not limited to healthy subjects). We compared the performance of deep‐learning‐based motion correction on individual channel images (single‐channel model) with that performed after coil combination (channel‐combined model). We also investigate simultaneous motion correction of all channel data from an image volume (multichannel model). The single‐channel model significantly (p < 0.0001) improved mean absolute error, with an average 50.9% improvement compared with the uncorrected images. This was significantly (p < 0.0001) better than the 36.3% improvement achieved by the channel‐combined model (conventional approach). The multichannel model provided no significant improvement in quantitative measures of image quality compared with the uncorrected images. Results were independent of the presence of pathology, and generalizable to a new center unseen during training. Performing motion correction on single‐channel images prior to coil combination provided an improvement in performance compared with conventional deep‐learning‐based motion correction. Improved deep learning methods for retrospective correction of motion‐affected MR images could reduce the need for repeat scans if applied in a clinical setting. [ABSTRACT FROM AUTHOR]

Background: Cognitive aging is described as the age‐related declines in multiple cognitive abilities such as processing speed, memory, and executive function. However, old individuals with cognitive reserve are observed. Moreover, there are older adults aged more than 80 years, called Super‐Agers, who show cognitive abilities at least as good as middle‐aged adults, particularly in memory. Understanding the differences in brain structure between Super‐agers and Cognitive decliners can provide practical knowledge for middle‐aged adults to reserve the cognitive functions as they age and decrease Alzheimer's Disease risk. Method: Using the longitudinal cognitive tests of the UK Biobank, principal component analysis (PCA) was applied to combine the performance of the multiple cognitive domains to define the general cognitive ability (GCA) which was used to identify Super‐Ager and Cognitive Decline individuals. GCA 1 scores were extracted by processing speed, verbal and numeric reasoning, prospective memory, and visual declarative memory exams, and GCA 2 scores were extracted by executive function and processing speed exams. Random Forest (RF) classification models with a multi‐stage feature selection method were developed to distinguish Super‐Agers from Cognitive Decliners using the structural Magnetic Resonance (sMRI) features for two GCA 1 and 2 labels individually. Models' performance and feature importance ranking were analyzed. Result: RF model achieved AUC of 73% with 54 selected sMRI predictors by GCA 1 labels, and age, education, area of total surface, area of pars orbitalis, mean intensity of thalamus, area of superior frontal were the top six features in classification. Also, RF models achieved AUC of 76% with 42 selected sMRI features by GCA 2 labels, and age, grey‐white contrast in pars triangularis, volume of CA3‐head, grey‐white contrast in pars opercularis, grey‐white contrast in frontal pole, and mean intensity of choroid plexus were the top six features in classification. Conclusion: our results suggest that using PCA to combine various cognitive domains is an optimal method to analyze GCA in older adults and identify latent cognitive trajectory type. Furthermore, the multi‐stage feature selection model could find the most relevant sMRI features to classify Super‐Agers and Cognitive Decliners. [ABSTRACT FROM AUTHOR]

The preprocessing of diffusion magnetic resonance imaging (dMRI) data involve numerous steps, including the corrections for head motion, susceptibility distortion, low signal‐to‐noise ratio, and signal drifting. Researchers or clinical practitioners often need to configure different preprocessing steps depending on disparate image acquisition schemes, which increases the technical threshold for dMRI analysis for nonexpert users. This could cause disparities in data processing approaches and thus hinder the comparability between studies. To make the dMRI data processing steps transparent and adapt to various dMRI acquisition schemes for researchers, we propose a semi‐automated pipeline tool for dMRI named integrated diffusion image operator or iDIO. This pipeline integrates features from a wide range of advanced dMRI software tools and targets at providing a one‐click solution for dMRI data analysis, via adaptive configuration for a set of suggested processing steps based on the image header of the input data. Additionally, the pipeline provides options for post‐processing, such as estimation of diffusion tensor metrics and whole‐brain tractography‐based connectomes reconstruction using common brain atlases. The iDIO pipeline also outputs an easy‐to‐interpret quality control report to facilitate users to assess the data quality. To keep the transparency of data processing, the execution log and all the intermediate images produced in the iDIO's workflow are accessible. The goal of iDIO is to reduce the barriers for clinical or nonspecialist users to adopt the state‐of‐art dMRI processing steps. [ABSTRACT FROM AUTHOR]

Introduction: The false discovery rate (FDR) procedure does not incorporate the geometry of the random field and requires high statistical power at each voxel, a requirement not satisfied by the limited number of participants in imaging studies. Topological FDR, threshold free cluster enhancement (TFCE), and probabilistic TFCE improve statistical power by incorporating local geometry. However, topological FDR requires specifying a cluster defining threshold and TFCE requires specifying transformation weights. Methods: Geometry‐derived statistical significance (GDSS) procedure overcomes these limitations by combining voxelwise p‐values for the test statistic with the probabilities computed from the local geometry for the random field, thereby providing substantially greater statistical power than the procedures currently used to control for multiple comparisons. We use synthetic data and real‐world data to compare its performance against the performance of these other, previously developed procedures. Results: GDSS provided substantially greater statistical power relative to the comparator procedures, which was less variable to the number of participants. GDSS was more conservative than TFCE: that is, it rejected null hypotheses at voxels with much higher effect sizes than TFCE. Our experiments also showed that the Cohen's D effect size decreases as the number of participants increases. Therefore, sample size calculations from small studies may underestimate the participants required in larger studies. Our findings also suggest effect size maps should be presented along with p‐value maps for correct interpretation of findings. Conclusions: GDSS compared with the other procedures provides considerably greater statistical power for detecting true positives while limiting false positives, especially in small sized (<40 participants) imaging cohorts. [ABSTRACT FROM AUTHOR]

Background: Deep‐learning is widely used for lesion classification. However, in the clinic patient data often has missing images. Purpose: To evaluate the use of generated, duplicate and empty(black) images for replacing missing MRI data in AI brain tumor classification tasks. Study Type: Retrospective. Population: 224 patients (local‐dataset; low‐grade‐glioma (LGG) = 37, high‐grade‐glioma (HGG) = 187) and 335 patients (public‐dataset (BraTS); LGG = 76, HGG = 259). The local‐dataset was divided into training (64), validation (16), and internal‐test‐data (20), while the public‐dataset was an independent test‐set. Field Strength/Sequence: T1WI, T1WI+C, T2WI, and FLAIR images (1.5T/3.0T‐MR), obtained from different suppliers. Assessment: Three image‐to‐image translation generative‐adversarial‐network (Pix2Pix‐GAN) models were trained on the local‐dataset, to generate T1WI, T2WI, and FLAIR images. The rating‐and‐preference‐judgment assessment was performed by three human‐readers (radiologist (MD) and two MRI‐technicians). Resnet152 was used for classification, and inference was performed on both datasets, with baseline input, and with missing data replaced by 1) generated images; 2) duplication of existing images; and 3) black images. Statistical Tests: The similarity between the generated and the original images was evaluated using the peak‐signal‐to‐noise‐ratio (PSNR) and the structural‐similarity‐index‐measure (SSIM). Classification results were evaluated using accuracy, F1‐score and the Kolmogorov–Smirnov test and distance. Results: For baseline‐state, the classification model reached to accuracy = 0.93,0.82 on the local and public‐datasets. For the missing‐data methods, high similarity was obtained between the generated and the original images with mean PSNR = 35.65,32.94 and SSIM = 0.87,0.91 on the local and public‐datasets; 39% of the generated‐images were labeled as real images by the human‐readers. The classification model using generated‐images to replace missing images produced the highest results with mean accuracy = 0.91,0.82 compared to 0.85,0.79 for duplicated and 0.77,0.68 for use of black images; Data Conclusion: The feasibility for inference classification model on an MRI dataset with missing images using the Pix2pix‐GAN generated images, was shown. The stability and generalization ability of the model was demonstrated by producing consistent results on two independent datasets. Level of Evidence: 3 Technical Efficacy: Stage 5 [ABSTRACT FROM AUTHOR]

Integrating datasets from multiple sites and scanners can increase statistical power for neuroimaging studies but can also introduce significant inter‐site confounds. We evaluated the effectiveness of ComBat, an empirical Bayes approach, to combine longitudinal preclinical MRI data acquired at 4.7 or 9.4 T at two different sites in Australia. Male Sprague Dawley rats underwent MRI on Days 2, 9, 28, and 150 following moderate/severe traumatic brain injury (TBI) or sham injury as part of Project 1 of the NIH/NINDS‐funded Centre Without Walls EpiBioS4Rx project. Diffusion‐weighted and multiple‐gradient‐echo images were acquired, and outcomes included QSM, FA, and ADC. Acute injury measures including apnea and self‐righting reflex were consistent between sites. Mixed‐effect analysis of ipsilateral and contralateral corpus callosum (CC) summary values revealed a significant effect of site on FA and ADC values, which was removed following ComBat harmonization. Bland–Altman plots for each metric showed reduced variability across sites following ComBat harmonization, including for QSM, despite appearing to be largely unaffected by inter‐site differences and no effect of site observed. Following harmonization, the combined inter‐site data revealed significant differences in the imaging metrics consistent with previously reported outcomes. TBI resulted in significantly reduced FA and increased susceptibility in the ipsilateral CC, and significantly reduced FA in the contralateral CC compared with sham‐injured rats. Additionally, TBI rats also exhibited a reversal in ipsilateral CC ADC values over time with significantly reduced ADC at Day 9, followed by increased ADC 150 days after injury. Our findings demonstrate the need for harmonizing multi‐site preclinical MRI data and show that this can be successfully achieved using ComBat while preserving phenotypical changes due to TBI. [ABSTRACT FROM AUTHOR]

Motion during the acquisition of magnetic resonance imaging (MRI) data degrades image quality, hindering our capacity to characterise disease in patient populations. Quality control procedures allow the exclusion of the most affected images from analysis. However, the criterion for exclusion is difficult to determine objectively and exclusion can lead to a suboptimal compromise between image quality and sample size. We provide an alternative, data‐driven solution that assigns weights to each image, computed from an index of image quality using restricted maximum likelihood. We illustrate this method through the analysis of quantitative MRI data. The proposed method restores the validity of statistical tests, and performs near optimally in all brain regions, despite local effects of head motion. This method is amenable to the analysis of a broad type of MRI data and can accommodate any measure of image quality. [ABSTRACT FROM AUTHOR]

Objectives: Acute carbon monoxide (CO) poisoning survivors may experience persistent neurological sequelae (PNS) and delayed neurological sequelae (DNS). This study evaluated the clinical features, laboratory results, acute brain lesions (ABLs) on diffusion-weighted imaging (DWI) at presentation, and long-term outcomes and explored differences between patients with PNS and DNS., Methods: The study included 443 patients who had experienced CO poisoning, underwent DWI and completed 1-year follow-ups. The demographics, comorbidities, symptomatology, laboratory results, ABLs on DWI at presentation, and long-term outcomes were compared between patients with PNS and those with DNS., Results: The 42 (9.5%) and 96 (21.7%) patients with PNS and DNS, respectively, showed no significant differences in demographics, duration of CO exposure, initial conscious level, symptomatology, and laboratory results. ABLs on DWI were observed in 33 patients (33/42) with PNS and 62 patients (62/96) with DNS. The most common region of ABLs was the globus pallidus (60.6% and 56.6% in PNS and DNS, respectively). The proportion of ABLs present and lesion distribution did not differ significantly between the two groups. At 1 year, a significantly higher proportion of patients in the PNS group showed a good outcome (defined as modified Rankin Scale [mRS] scores of 0-2, 81%) compared with the DNS group (81% vs. 56.3%, p = .047)., Conclusion: Demographics, clinical features, laboratory results, and acute brain lesions on MRI at presentation did not differ between the PNS and DNS groups. However, the long-term outcome of PNS was better than that of DNS., (© 2022 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd.)

Heavy cannabis use (HCU) is frequently associated with a plethora of cognitive, psychopathological and sensorimotor phenomena. Although HCU is frequent, specific patterns of abnormal brain structure and function underlying HCU in individuals presenting without cannabis-use disorder or other current and life-time major mental disorders are unclear at present. This multimodal magnetic resonance imaging (MRI) study examined resting-state functional MRI (rs-fMRI) and structural MRI (sMRI) data from 24 persons with HCU and 16 controls. Parallel independent component analysis (p-ICA) was used to examine covarying components among grey matter volume (GMV) maps computed from sMRI and intrinsic neural activity (INA), as derived from amplitude of low-frequency fluctuations (ALFF) maps computed from rs-fMRI data. Further, we used JuSpace toolbox for cross-modal correlations between MRI-based modalities with nuclear imaging derived estimates, to examine specific neurotransmitter system changes underlying HCU. We identified two transmodal components, which significantly differed between the HCU and controls (GMV: p = 0.01, ALFF p = 0.03, respectively). The GMV component comprised predominantly cerebello-temporo-thalamic regions, whereas the INA component included fronto-parietal regions. Across HCU, loading parameters of both components were significantly associated with distinct HCU behavior. Finally, significant associations between GMV and the serotonergic system as well as between INA and the serotonergic, dopaminergic and μ-opioid receptor system were detected. This study provides novel multimodal neuromechanistic insights into HCU suggesting co-altered structure/function-interactions in neural systems subserving cognitive and sensorimotor functions., (© 2021 The Authors. Addiction Biology published by John Wiley & Sons Ltd on behalf of Society for the Study of Addiction.)

Assess the contribution of preoperative MRI data in improving grading of adult astrocytomas reclassified according to the WHO 2016 and cIMPACT‐NOW update 3. Retrospective unicentric cohort study of 679 adult patients treated for newly diagnosed diffuse astrocytic and oligodendroglial tumors (January 2006–December 2016). We first systematically compared radiological (contrast enhancement present [CE+] vs. absent [CE−]) and histopathological findings (microvascular proliferation present [MPV+] vs. absent [MPV−]) to validate whether this comparing step of neoangiogenesis represents an efficient method to appreciate the representativity of the tumoral sampling. We focused on 629 cases of astrocytomas for radio‐histological integrated analyses. In 598 cases (95.1%), neoangiogenesis evaluated by MRI or histology (CE+/MPV+ or CE−/MPV−) was identical. For the CE+/MPV− and CE−/MPV+ groups (23 cases), the radio‐histological face‐to‐face evaluation allowed us to assess that for 13 cases (56.5%) the reason for this discrepancy was an undersampled tumor. We analyzed the group of CE+/MPV− (n = 8) and CE−/MPV+ (n = 2) in verified image‐guided tumoral samples. Finally, we identified three new prognostic subgroups for molecular glioblastomas: (1) "non‐representative sampling" (n = 9), (2) "Non neoangiogenic glioblastoma at the time of diagnosis, without contrast enhancement and microvascular proliferation" (n = 8), and (3) "contrast enhancing glioblastoma but without microvascular proliferation in a representative sample" (n = 4). Neoangiogenesis processes should be assessed to improve the prognosis accuracy of the current integrated diagnosis. We suggest adding imaging analyses during the neuropathological analysis of astrocytomas in adults. [ABSTRACT FROM AUTHOR]

Background: Several prior studies have used diffusion MRI to investigate the relationship between white matter microstructure and aging; however, many of these studies used conventional diffusion MRI measures and single site data. The goal of the study is to leverage multi‐site harmonized diffusion MRI data in conjunction with a novel post‐processing technique [i.e., free‐water (FW) correction] to quantify the aging related tract‐specific changes in white matter microstructure. Method: The dataset used in this study was collated using several well‐established longitudinal cohorts of aging [Alzheimer's Neuroimaging Initiative (ADNI), Baltimore Longitudinal Study of Aging (BLSA), Vanderbilt Memory & Aging Project (VMAP)]. In total, this dataset included 1,909 participants (mean age at baseline: 72±9 years, 59% female) and 4,844 imaging sessions (mean number of visits: 4 ± 2 years, interval range: 1–12 years). Data was processed using standard approaches and uncorrected fractional anisotropy (FAU) was quantified with seven white matter tractography atlases (see Figure 1A). Data was then post‐processed using the FW correction technique and FW and FW‐corrected FA (FAT) values were quantified. Data were then harmonized using the ComBat technique and linear mixed‐effects regression was conducted on each microstructural measure, covarying for age at baseline, sex, cognitive status, education, APOE‐ε4 carrier status, and APOE‐ε2 carrier status. The effect of aging was modelled using an Age_at_Baseline x Interval interaction term. Result: Age was associated with lower FAU in all seven tracts (pFDR range: 4.77x10‐10 to ‐0.01). In the FW analysis, age was associated with higher FW in all seven tracts (pFDR range: 7.81x10‐12 to 1.33x10‐6). For FAT, however, age was only associated with lower FAT in the limbic tracts (pFDR=0.024) in addition to the occipital (pFDR=1.19x10‐8), parietal (pFDR=2.32x10‐4), and prefrontal (pFDR=0.024) TC tracts. Figure 1B‐D illustrates our findings. Conclusion: This study suggests that while there are global associations with FAU and age, these associations are attenuated once correcting for partial volume effects. Leveraging FW analysis, we found a global association with FW and age, whereby age is associated with higher FW. Further, we found that age is associated with reductions in FAT in the limbic and occipital/parietal/prefrontal TC tracts. [ABSTRACT FROM AUTHOR]

Background and objectives: Excessive smartphone use, also referred to as "smartphone addiction" (SPA), has increasingly attracted neuroscientific interest due to its similarities with other behavioral addictions, particularly internet gaming disorder. Little is known about the neural mechanisms underlying smartphone addiction. We explored interrelationships between brain structure and function to specify neurobiological correlates of SPA on a neural system level. Methods: Gray matter volume (GMV) and intrinsic neural activity (INA) were investigated in individuals with SPA (n = 20) and controls (n = 24), using multimodal magnetic resonance imaging and multivariate data fusion techniques, that is, parallel independent component analysis. Results: The joint analysis of both data modalities explored shared information between GMV and INA. In particular, two amplitudes of low frequency fluctuations‐based independent neural systems significantly differed between individuals with SPA and controls. A medial/dorsolateral prefrontal system exhibited lower functional network strength in individuals with SPA versus controls, whereas the opposite pattern was detected in a parietal cortical/cerebellar system. Neural network strength was significantly related to duration of smartphone use and sleep difficulties. Discussion and conclusions: We show modality‐specific associations of the brain's resting‐state activity with distinct and shared SPA symptom dimensions. In particular, the data suggest contributions of aberrant prefrontal and parietal neural network strength as a possible signature of deficient executive control in SPA. Scientific significance: This study suggests distinct neural mechanisms underlying specific biological and behavioral dimensions of excessive smartphone use. [ABSTRACT FROM AUTHOR]

Purpose: Brain metastases (BM) occur frequently in patients with metastatic cancer. Early and accurate detection of BM is essential for treatment planning and prognosis in radiation therapy. Due to their tiny sizes and relatively low contrast, small BM are very difficult to detect manually. With the recent development of deep learning technologies, several res earchers have reported promising results in automated brain metastasis detection. However, the detection sensitivity is still not high enough for tiny BM, and integration into clinical practice in regard to differentiating true metastases from false positives (FPs) is challenging. Methods: The DeepMedic network with the binary cross‐entropy (BCE) loss is used as our baseline method. To improve brain metastasis detection performance, a custom detection loss called volume‐level sensitivity–specificity (VSS) is proposed, which rates metastasis detection sensitivity and specificity at a (sub)volume level. As sensitivity and precision are always a trade‐off, either a high sensitivity or a high precision can be achieved for brain metastasis detection by adjusting the weights in the VSS loss without decline in dice score coefficient for segmented metastases. To reduce metastasis‐like structures being detected as FP metastases, a temporal prior volume is proposed as an additional input of DeepMedic. The modified network is called DeepMedic+ for distinction. Combining a high‐sensitivity VSS loss and a high specificity loss for DeepMedic+, the majority of true positive metastases are confirmed with high specificity, while additional metastases candidates in each patient are marked with high sensitivity for detailed expert evaluation. Results: Our proposed VSS loss improves the sensitivity of brain metastasis detection, increasing the sensitivity from 85.3% for DeepMedic with BCE to 97.5% for DeepMedic with VSS. Alternatively, the precision is improved from 69.1% for DeepMedic with BCE to 98.7% for DeepMedic with VSS. Comparing DeepMedic+ with DeepMedic with the same VSS loss, 44.4% of the FP metastases are reduced in the high‐sensitivity model and the precision reaches 99.6% for the high‐specificity model. The mean dice coefficient for all metastases is about 0.81. With the ensemble of the high‐sensitivity and high‐specificity models, on average only 1.5 FP metastases per patient need further check, while the majority of true positive metastases are confirmed. Conclusions: Our proposed VSS loss and temporal prior improve brain metastasis detection sensitivity and precision. The ensemble learning is able to distinguish high confidence true positive metastases from metastases candidates that require special expert review or further follow‐up, being particularly well‐fit to the requirements of expert support in real clinical practice. This facilitates metastasis detection and segmentation for neuroradiologists in diagnostic and radiation oncologists in therapeutic clinical applications. [ABSTRACT FROM AUTHOR]

We propose a surrogate model for the fluid–structure interaction (FSI) problem for the study of blood dynamics in carotid arteries in presence of plaque. This is based on the integration of a numerical model with subject‐specific data and clinical imaging. We propose to model the plaque as part of the tissues surrounding the vessel wall through the application of an elastic support boundary condition. In order to characterize the plaque and other surrounding tissues, such as the close‐by jugular vein, the elastic parameters of the boundary condition were spatially differentiated and their values were estimated by minimizing the discrepancies between computed vessel displacements and reference values obtained from CINE Magnetic Resonance Imaging data. We applied the model to three subjects with a degree of stenosis greater than 70%. We found that accounting for both plaque and jugular vein in the estimation of the elastic parameters increases the accuracy. In particular, in all patients, mismatches between computed and in vivo measured wall displacements were one to two orders of magnitude lower than the spatial resolution of the original MRI data. These results confirmed the validity of the proposed surrogate plaque model. We also compared fluid‐dynamics results with those obtained in a fixed wall setting and in a full FSI model, used as gold standard, highlighting the better accordance of our results in comparison to the rigid ones. [ABSTRACT FROM AUTHOR]

Introduction: It is unclear how rates of white matter microstructural decline differ between normal aging and abnormal aging. Methods: Diffusion MRI data from several well‐established longitudinal cohorts of aging (Alzheimer's Disease Neuroimaging Initiative [ADNI], Baltimore Longitudinal Study of Aging [BLSA], Vanderbilt Memory & Aging Project [VMAP]) were free‐water corrected and harmonized. This dataset included 1723 participants (age at baseline: 72.8 ± 8.87 years, 49.5% male) and 4605 imaging sessions (follow‐up time: 2.97 ± 2.09 years, follow‐up range: 1–13 years, mean number of visits: 4.42 ± 1.98). Differences in white matter microstructural decline in normal and abnormal agers was assessed. Results: While we found a global decline in white matter in normal/abnormal aging, we found that several white matter tracts (e.g., cingulum bundle) were vulnerable to abnormal aging. Conclusions: There is a prevalent role of white matter microstructural decline in aging, and future large‐scale studies in this area may further refine our understanding of the underlying neurodegenerative processes. HIGHLIGHTS: Longitudinal data were free‐water corrected and harmonized.Global effects of white matter decline were seen in normal and abnormal aging.The free‐water metric was most vulnerable to abnormal aging.Cingulum free‐water was the most vulnerable to abnormal aging. [ABSTRACT FROM AUTHOR]

Constrained spherical deconvolution (CSD) of diffusion‐weighted MRI (DW‐MRI) is a popular analysis method that extracts the full white matter (WM) fiber orientation density function (fODF) in the living human brain, noninvasively. It assumes that the DW‐MRI signal on the sphere can be represented as the spherical convolution of a single‐fiber response function (RF) and the fODF, and recovers the fODF through the inverse operation. CSD approaches typically require that the DW‐MRI data is sampled shell‐wise, and estimate the RF in a purely spherical manner using spherical basis functions, such as spherical harmonics (SH), disregarding any radial dependencies. This precludes analysis of data acquired with nonspherical sampling schemes, for example, Cartesian sampling. Additionally, nonspherical sampling can also arise due to technical issues, for example, gradient nonlinearities, resulting in a spatially dependent bias of the apparent tissue densities and connectivity information. Here, we adopt a compact model for the RFs that also describes their radial dependency. We demonstrate that the proposed model can accurately predict the tissue response for a wide range of b‐values. On shell‐wise data, our approach provides fODFs and tissue densities indistinguishable from those estimated using SH. On Cartesian data, fODF estimates and apparent tissue densities are on par with those obtained from shell‐wise data, significantly broadening the range of data sets that can be analyzed using CSD. In addition, gradient nonlinearities can be accounted for using the proposed model, resulting in much more accurate apparent tissue densities and connectivity metrics. [ABSTRACT FROM AUTHOR]

Echo planar imaging (EPI) is widely used in functional and diffusion‐weighted MRI, but suffers from significant geometric distortions in the phase encoding direction caused by inhomogeneities in the static magnetic field (B0). This is a particular challenge for EPI at very high field (≥7 T), as distortion increases with higher field strength. A number of techniques for distortion correction exist, including those based on B0 field mapping and acquiring EPI scans with opposite phase encoding directions. However, few quantitative comparisons of distortion compensation methods have been performed using human EPI data, especially at very high field. Here, we compared distortion compensation using B0 field maps and opposite phase encoding scans in two different software packages (FSL and AFNI) applied to 7 T gradient echo (GE) EPI data from 31 human participants. We assessed distortion compensation quality by quantifying alignment to anatomical reference scans using Dice coefficients and mutual information. Performance between FSL and AFNI was equivalent. In our whole‐brain analyses, we found superior distortion compensation using GE scans with opposite phase encoding directions, versus B0 field maps or spin echo (SE) opposite phase encoding scans. However, SE performed better when analyses were limited to ventromedial prefrontal cortex, a region with substantial dropout. Matching the type of opposite phase encoding scans to the EPI data being corrected (e.g., SE‐to‐SE) also yielded better distortion correction. While the ideal distortion compensation approach likely varies depending on methodological differences across experiments, this study provides a framework for quantitative comparison of different distortion compensation methods. [ABSTRACT FROM AUTHOR]

Purpose: Online adaptive radiotherapy would greatly benefit from the development of reliable auto‐segmentation algorithms for organs‐at‐risk and radiation targets. Current practice of manual segmentation is subjective and time‐consuming. While deep learning‐based algorithms offer ample opportunities to solve this problem, they typically require large datasets. However, medical imaging data are generally sparse, in particular annotated MR images for radiotherapy. In this study, we developed a method to exploit the wealth of publicly available, annotated CT images to generate synthetic MR images, which could then be used to train a convolutional neural network (CNN) to segment the parotid glands on MR images of head and neck cancer patients. Methods: Imaging data comprised 202 annotated CT and 27 annotated MR images. The unpaired CT and MR images were fed into a 2D CycleGAN network to generate synthetic MR images from the CT images. Annotations of axial slices of the synthetic images were generated by propagating the CT contours. These were then used to train a 2D CNN. We assessed the segmentation accuracy using the real MR images as test dataset. The accuracy was quantified with the 3D Dice similarity coefficient (DSC), Hausdorff distance (HD), and mean surface distance (MSD) between manual and auto‐generated contours. We benchmarked the approach by a comparison to the interobserver variation determined for the real MR images, as well as to the accuracy when training the 2D CNN to segment the CT images. Results: The determined accuracy (DSC: 0.77±0.07, HD: 18.04±12.59mm, MSD: 2.51±1.47mm) was close to the interobserver variation (DSC: 0.84±0.06, HD: 10.85±5.74mm, MSD: 1.50±0.77mm), as well as to the accuracy when training the 2D CNN to segment the CT images (DSC: 0.81±0.07, HD: 13.00±7.61mm, MSD: 1.87±0.84mm). Conclusions: The introduced cross‐modality learning technique can be of great value for segmentation problems with sparse training data. We anticipate using this method with any nonannotated MRI dataset to generate annotated synthetic MR images of the same type via image style transfer from annotated CT images. Furthermore, as this technique allows for fast adaptation of annotated datasets from one imaging modality to another, it could prove useful for translating between large varieties of MRI contrasts due to differences in imaging protocols within and between institutions. [ABSTRACT FROM AUTHOR]

Diffusion magnetic resonance imaging can indirectly infer the microstructure of tissues and provide metrics subject to normal variability in a population. Potentially abnormal values may yield essential information to support analysis of controls and patients cohorts, but subtle confounds could be mistaken for purely biologically driven variations amongst subjects. In this work, we propose a new harmonization algorithm based on adaptive dictionary learning to mitigate the unwanted variability caused by different scanner hardware while preserving the natural biological variability of the data. Our harmonization algorithm does not require paired training data sets, nor spatial registration or matching spatial resolution. Overcomplete dictionaries are learned iteratively from all data sets at the same time with an adaptive regularization criterion, removing variability attributable to the scanners in the process. The obtained mapping is applied directly in the native space of each subject toward a scanner‐space. The method is evaluated with a public database which consists of two different protocols acquired on three different scanners. Results show that the effect size of the four studied diffusion metrics is preserved while removing variability attributable to the scanner. Experiments with alterations using a free water compartment, which is not simulated in the training data, shows that the modifications applied to the diffusion weighted images are preserved in the diffusion metrics after harmonization, while still reducing global variability at the same time. The algorithm could help multicenter studies pooling their data by removing scanner specific confounds, and increase statistical power in the process. [ABSTRACT FROM AUTHOR]

Objective: The objective of this study was to determine the ability of 7T‐MRI for characterizing brain tissue integrity in early relapsing‐remitting MS patients compared to conventional 3T‐MRI and to investigate whether 7T‐MRI improves the performance for detecting cortical gray matter neurodegeneration and its associated network reorganization dynamics. Methods: Seven early relapsing‐remitting MS patients and seven healthy individuals received MRI at 7T and 3T, whereas 30 and 40 healthy controls underwent separate 3T‐ and 7T‐MRI sessions, respectively. Surface‐based cortical thickness (CT) and gray‐to‐white contrast (GWc) measures were used to model morphometric networks, analyzed with graph theory by means of modularity, clustering coefficient, path length, and small‐worldness. Results: 7T‐MRI had lower CT and higher GWc compared to 3T‐MRI in MS. CT and GWc measures robustly differentiated MS from controls at 3T‐MRI. 7T‐ and 3T‐MRI showed high regional correspondence for CT (r = 0.72, P = 2e‐78) and GWc (r = 0.83, P = 5.5e‐121) in MS patients. MS CT and GWc morphometric networks at 7T‐MRI showed higher modularity, clustering coefficient, and small‐worldness than 3T, also compared to controls. Interpretation: 7T‐MRI allows to more precisely quantify morphometric alterations across the cortical mantle and captures more sensitively MS‐related network reorganization. Our findings open new avenues to design more accurate studies quantifying brain tissue loss and test treatment effects on tissue repair. [ABSTRACT FROM AUTHOR]

This article discusses the importance of multimodal magnetic resonance imaging (MRI) in brain tumor classification and segmentation tasks for accurate diagnosis and treatment decisions. However, missing image sequences in MRI data pose a challenge for machine learning and deep learning algorithms. The article explores different methodologies for handling missing data, including image synthesis-based methods using general adversarial networks (GAN) models. The authors evaluate the performance of synthetic images generated by a pix2pix GAN model for brain tumor classification tasks and find that it effectively represents missing data. This study suggests that the pix2pix GAN model could become an effective tool for the accurate diagnosis of brain tumors. [Extracted from the article]

Background: Deep learning (DL) has provided impressive results in numerous domains in recent years, including medical image analysis. Training DL models requires large data sets to yield good performance. Since medical data can be difficult to acquire, most studies rely on public research cohorts, which often have harmonized scanning protocols and strict exclusion criteria. This is not representative of a clinical setting. In this study, we investigated the performance of a DL model in out‐of‐distribution data from multiple memory clinics and research cohorts. Method: We trained multiple versions of AVRA: a DL model trained to predict visual ratings of Scheltens' medial temporal atrophy (MTA) scale (Mårtensson et al., 2019). This was done on different combinations of training data—starting with only harmonized MRI data from public research cohorts, and further increasing image heterogeneity in the training set by including external memory clinic data. We assessed the performance in multiple test sets by comparing AVRA's MTA ratings to an experienced radiologist's (who rated all images in this study). Data came from Alzheimer's Disease Neuroimaging Initiative (ADNI), AddNeuroMed, and images from 13 European memory clinics in the E‐DLB consortium. Results: Models trained only on research cohorts generalized well to new data acquired with similar protocols as the training data (weighted kappa κw between 0.70‐0.72), but worse to memory clinic data with more image variability (κw between 0.34‐0.66). This was most prominent in one specific memory clinic, where the DL model systematically predicted too low MTA scores. When including data from a wider range of scanners and protocols during training, the agreement to the radiologist's ratings in external memory clinics increased (κw between 0.51‐0.71). Conclusion: In this study we showed that increasing heterogeneity in training data improves generalization to out‐of‐distribution data. Our findings suggest that studies assessing reliability of a DL model should be done in multiple cohorts, and that softwares based on DL need to be rigorously evaluated prior to being certified for deployment to clinics. References: Mårtensson, G. et al. (2019) 'AVRA: Automatic Visual Ratings of Atrophy from MRI images using Recurrent Convolutional Neural Networks', NeuroImage: Clinical. Elsevier, 23(March), p. 101872. [ABSTRACT FROM AUTHOR]

Abstract: Flow in the great arteries (aorta, pulmonary artery) is normally laminar with a parabolic velocity profile. Eccentric flow jets are linked to various diseases like aneurysms. Cardiac 4D PC‐MRI data provide spatio‐temporally resolved blood flow information for the whole cardiac cycle. In this work, we establish a time‐dependent visualization and quantification of flow jets. For this purpose, equidistant measuring planes are automatically placed along the vessel's centerline. The flow jet position and region with highest velocities are extracted for every plane in each time step. This is done during pre‐processing and without user‐defined parameters. We visualize the main flow jet as geometric tube. High‐velocity areas are depicted as a net around this tube. Both geometries are time‐dependent and can be animated. Quantitative values are provided during cross‐sectional measuring plane‐based evaluation. Moreover, we offer a plot visualization as summary of flow jet characteristics for the selected plane. Our physiologically plausible results are in accordance with medical findings. Our clinical collaborators appreciate the possibility to view the flow jet in the whole vessel at once, which normally requires repeated pathline filtering due to varying velocities along the vessel course. The overview plots are considered as valuable for documentation purposes. [ABSTRACT FROM AUTHOR]

Machine learning has increasingly been applied to classification of schizophrenia in neuroimaging research. However, direct replication studies and studies seeking to investigate generalizability are scarce. To address these issues, we assessed within‐site and between‐site generalizability of a machine learning classification framework which achieved excellent performance in a previous study using two independent resting‐state functional magnetic resonance imaging data sets collected from different sites and scanners. We established within‐site generalizability of the classification framework in the main data set using cross‐validation. Then, we trained a model in the main data set and investigated between‐site generalization in the validated data set using external validation. Finally, recognizing the poor between‐site generalization performance, we updated the unsupervised algorithm to investigate if transfer learning using additional unlabeled data were able to improve between‐site classification performance. Cross‐validation showed that the published classification procedure achieved an accuracy of 0.73 using majority voting across all selected components. External validation found a classification accuracy of 0.55 (not significant) and 0.70 (significant) using the direct and transfer learning procedures, respectively. The failure of direct generalization from one site to another demonstrates the limitation of within‐site cross‐validation and points toward the need to incorporate efforts to facilitate application of machine learning across multiple data sets. The improvement in performance with transfer learning highlights the importance of taking into account the properties of data when constructing predictive models across samples and sites. Our findings suggest that machine learning classification result based on a single study should be interpreted cautiously. [ABSTRACT FROM AUTHOR]

Accurate stroke lesion segmentation is a critical step in the neuroimaging processing pipeline for assessing the relationship between poststroke brain structure, function, and behavior. Many multimodal segmentation algorithms have been developed for acute stroke neuroimaging, yet few algorithms are effective with only a single T1‐weighted (T1w) anatomical MRI. This is a critical gap because multimodal MRI is not commonly available due to time and cost constraints in the stroke rehabilitation setting. Although several attempts to automate the segmentation of chronic lesions on single‐channel T1w MRI have been made, these approaches have not been systematically evaluated on a large dataset. We performed an exhaustive review of the literature and identified one semiautomated and three fully automated approaches for segmentation of chronic stroke lesions using T1w MRI within the last 10 years: Clusterize, automated lesion identification (ALI), Gaussian naïve Bayes lesion detection (lesionGnb), and lesion identification with neighborhood data analysis (LINDA). We evaluated each method on a large T1w stroke dataset (N = 181). LINDA was the most computationally expensive approach, but performed best across the three main evaluation metrics (median values: dice coefficient = 0.50, Hausdorff's distance = 36.34 mm, and average symmetric surface distance = 4.97 mm). lesionGnb had the highest recall/least false negatives (median = 0.80). However, across the automated methods, many lesions were either misclassified (ALI: 28, lesionGnb: 39, LINDA: 45) or not identified (ALI: 24, LINDA: 23, lesionGnb: 0). Segmentation accuracy in all automated methods were influenced by size (small: worst) and stroke territory (brainstem, cerebellum: worst) of the lesion. To facilitate reproducible science, our analysis files have been made publicly available online. [ABSTRACT FROM AUTHOR]

To compare the magnetization transfer ratio asymmetry (MTRasym) values calculated with polynomial and spline fitting methods for B0 correction of the full Z‐spectrum images, full Z‐spectrum data was obtained from 12 elderly healthy subjects by using a 3D segmented EPI sequence, as well as from 17 other subjects by using a 3D GRASE sequence with two different 3 T MRI systems. The full Z‐spectra were analyzed to map MTRasym with three different fitting methods, namely the 10th and 14th polynomial and spline methods, for B0 correction. The MTRasym values for each offset frequency were compared among the three fitting methods. For the 3D segmented EPI sequence, the MTRasym values significantly differed among the three fitting methods at 0.86, 2.14, 3.00, and 3.43 ppm offset frequencies. For the 3D GRASE sequence, the MTRasym values obtained by 14th polynomial and spline fitting methods significantly differed at arbitrary offset frequencies. The MTRasym values are sensitive to the fitting methods of B0 correction and to the type of acquisition sequence for the full Z‐spectrum. Therefore, an appropriate fitting method should be used to analyze the full Z‐spectrum obtained from the brain. [ABSTRACT FROM AUTHOR]

Purpose: To develop a denoising strategy leveraging redundancy in high‐dimensional data. Theory and Methods: The SNR fundamentally limits the information accessible by MRI. This limitation has been addressed by a host of denoising techniques, recently including the so‐called MPPCA: principal component analysis of the signal followed by automated rank estimation, exploiting the Marchenko‐Pastur distribution of noise singular values. Operating on matrices comprised of data patches, this popular approach objectively identifies noise components and, ideally, allows noise to be removed without introducing artifacts such as image blurring, or nonlocal averaging. The MPPCA rank estimation, however, relies on a large number of noise singular values relative to the number of signal components to avoid such ill effects. This condition is unlikely to be met when data patches and therefore matrices are small, for example due to spatially varying noise. Here, we introduce tensor MPPCA (tMPPCA) for the purpose of denoising multidimensional data, such as from multicontrast acquisitions. Rather than combining dimensions in matrices, tMPPCA uses each dimension of the multidimensional data's inherent tensor‐structure to better characterize noise, and to recursively estimate signal components. Results: Relative to matrix‐based MPPCA, tMPPCA requires no additional assumptions, and comparing the two in a numerical phantom and a multi‐TE diffusion MRI data set, tMPPCA dramatically improves denoising performance. This is particularly true for small data patches, suggesting that tMPPCA can be especially beneficial in such cases. Conclusions: The MPPCA denoising technique can be extended to high‐dimensional data with improved performance for smaller patch sizes. [ABSTRACT FROM AUTHOR]

Calhoun , V.D. , Adali , T. , Pearlson , G.D. , & Pekar , J.J. (2001). Spatial and temporal independent component analysis of functional MRI data containing a pair of task‐related waveforms. Human Brain Mapping , 13 , 43–53. https://doi.org/10.1002/hbm.1024In the author affiliations section, the affiliation of T. Adali was incorrect. It was stated as "Department of CSEE, University of Maryland, Baltimore, Maryland." This should have read: "Department of CSEE, University of Maryland Baltimore County, Maryland."We apologize for this error. [Extracted from the article]

Cardiac four-dimensional phase-contrast magnetic resonance imaging (4D PC-MRI) acquisitions have gained increasing clinical interest in recent years. They allow to non-invasively obtain extensive information about patient-specific hemodynamics, and thus have a great potential to improve the diagnosis, prognosis and therapy planning of cardiovascular diseases. A dataset contains time-resolved, three-dimensional blood flow directions and strengths, making comprehensive qualitative and quantitative data analysis possible. Quantitative measures, such as stroke volumes, help to assess the cardiac function and to monitor disease progression. Qualitative analysis allows to investigate abnormal flow characteristics, such as vortices, which are correlated to different pathologies. Processing the data comprises complex image processing methods, as well as flow analysis and visualization. In this work, we mainly focus on the aorta. We provide an overview of data measurement and pre-processing, as well as current visualization and quantification methods. This allows other researchers to quickly catch up with the topic and take on new challenges to further investigate the potential of 4D PC-MRI data. [ABSTRACT FROM AUTHOR]

Objective: In adulthood, the diagnosis of attention-deficit/hyperactivity disorder (ADHD) has been subject of recent controversy. We searched for a neuroanatomical signature associated with ADHD spectrum symptoms in adults by applying, for the first time, machine learning-based pattern classification methods to structural MRI and diffusion tensor imaging (DTI) data obtained from stimulant-naïve adults with childhood-onset ADHD and healthy controls (HC)., Method: Sixty-seven ADHD patients and 66 HC underwent high-resolution T1-weighted and DTI acquisitions. A support vector machine (SVM) classifier with a non-linear kernel was applied on multimodal image features extracted on regions of interest placed across the whole brain., Results: The discrimination between a mixed-gender ADHD subgroup and individually matched HC (n = 58 each) yielded area-under-the-curve (AUC) and diagnostic accuracy (DA) values of up to 0.71% and 66% (P = 0.003) respectively. AUC and DA values increased to 0.74% and 74% (P = 0.0001) when analyses were restricted to males (52 ADHD vs. 44 HC)., Conclusion: Although not at the level of clinically definitive DA, the neuroanatomical signature identified herein may provide additional, objective information that could influence treatment decisions in adults with ADHD spectrum symptoms., (© 2017 John Wiley & Sons A/S. Published by John Wiley & Sons Ltd.)

Tensor‐based morphometry (TBM) performed using T1‐weighted images (T1WIs) is a well‐established method for analyzing local morphological changes occurring in the brain due to normal aging and disease. However, in white matter regions that appear homogeneous on T1WIs, T1W‐TBM may be inadequate for detecting changes that affect specific pathways. In these regions, diffusion tensor MRI (DTI) can identify white matter pathways on the basis of their different anisotropy and orientation. In this study, we propose performing TBM using deformation fields constructed using all scalar and directional information provided by the diffusion tensor (DTBM) with the goal of increasing sensitivity in detecting morphological abnormalities of specific white matter pathways. Previously, mostly fractional anisotropy (FA) has been used to drive registration in diffusion MRI‐based TBM (FA‐TBM). However, FA does not have the directional information that the tensors contain, therefore, the registration based on tensors provides better alignment of brain structures and better localization of volume change. We compare our DTBM method to both T1W‐TBM and FA‐TBM in investigating differences in brain morphology between patients with complicated hereditary spastic paraplegia of type 11 (SPG11) and a group of healthy controls. Effect size maps of T1W‐TBM of SPG11 patients showed diffuse atrophy of white matter. However, DTBM indicated that atrophy was more localized, predominantly affecting several long‐range pathways. The results of our study suggest that DTBM could be a powerful tool for detecting morphological changes of specific white matter pathways in normal brain development and aging, as well as in degenerative disorders. [ABSTRACT FROM AUTHOR]