BLENDS: Augmentation of Functional Magnetic Resonance Images for Machine Learning Using Anatomically Constrained Warping.

Introduction: Data augmentation improves the accuracy of deep learning models when training data are scarce by synthesizing additional samples. This work addresses the lack of validated augmentation methods specific for synthesizing anatomically realistic four-dimensional (4D) (three-dimensional [3D] + time) images for neuroimaging, such as functional magnetic resonance imaging (fMRI), by proposing a new augmentation method. Methods: The proposed method, Brain Library Enrichment through Nonlinear Deformation Synthesis (BLENDS), generates new nonlinear warp fields by combining intersubject coregistration maps, computed using symmetric normalization, through spatial blending. These new warp fields can be applied to existing 4D fMRI to create new augmented images. BLENDS was tested on two neuroimaging problems using de-identified data sets: (1) the prediction of antidepressant response from task-based fMRI (original data set n = 163), and (2) the prediction of Parkinson's disease (PD) symptom trajectory from baseline resting-state fMRI regional homogeneity (original data set n = 43). Results: BLENDS readily generates hundreds of new fMRI from existing images, with unique anatomical variations from the source images, that significantly improve prediction performance. For antidepressant response prediction, augmenting each original image once (2 × the original training data) significantly increased prediction R2 from 0.055 to 0.098 (p<1e-6), whereas at 10 × augmentation R2 increased to 0.103. For the prediction of PD trajectory, 10 × augmentation R2 increased from -0.044 to 0.472 (p<1e-6). Conclusions: Augmentation of fMRI through nonlinear transformations with BLENDS significantly improved the performance of deep learning models on clinically relevant predictive tasks. This method will help neuroimaging researchers overcome data set size limitations and achieve more accurate predictive models.

Pitfalls and Recommended Strategies and Metrics for Suppressing Motion Artifacts in Functional MRI.

In resting-state functional magnetic resonance imaging (rs-fMRI), artefactual signals arising from subject motion can dwarf and obfuscate the neuronal activity signal. Typical motion correction approaches involve the generation of nuisance regressors, which are timeseries of non-brain signals regressed out of the fMRI timeseries to yield putatively artifact-free data. Recent work suggests that concatenating all regressors into a single regression model is more effective than the sequential application of individual regressors, which may reintroduce previously removed artifacts. This work compares 18 motion correction pipelines consisting of head motion, independent components analysis, and non-neuronal physiological signal regressors in sequential or concatenated combinations. The pipelines are evaluated on a dataset of cognitively normal individuals with repeat imaging and on datasets of studies of Autism Spectrum Disorder, Major Depressive Disorder, and Parkinson's Disease. Extensive metrics of motion artifact removal are measured, including resting state network recovery, Quality Control-Functional Connectivity (QC-FC) correlation, distance-dependent artifact, network modularity, and test-retest reliability of multiple rs-fMRI analyses. The results reveal limitations in previously proposed metrics, including the QC-FC correlation and modularity quality, and identify more robust artifact removal metrics. The results also reveal limitations in the concatenated regression approach, which is outperformed by the sequential regression approach in the test-retest reliability metrics. Finally, pipelines are recommended that perform well based on quantitative and qualitative comparisons across multiple datasets and robust metrics. These new insights and recommendations help address the need for effective motion artifact correction to reduce noise and confounds in rs-fMRI.

Temporal and topographical changes in theta power between middle childhood and adolescence during sentence comprehension.

Time frequency analysis of the EEG is increasingly used to study the neural oscillations supporting language comprehension. Although this method holds promise for developmental research, most existing work focuses on adults. Theta power (4-8 Hz) in particular often corresponds to semantic processing of words in isolation and in ongoing text. Here we investigated how the timing and topography of theta engagement to individual words during written sentence processing changes between childhood and adolescence (8-15 years). Results show that topographically, the theta response is broadly distributed in children, occurring over left and right central-posterior and midline frontal areas, and localizes to left central-posterior areas by adolescence. There were two notable developmental shifts. First, in response to each word, early (150-300 msec) theta engagement over frontal areas significantly decreases between 8 and 9 years and 10-11 years. Second, throughout the sentence, theta engagement over the right parietal areas significantly decreases between 10 and 11 years and 12-13 years with younger children's theta response remaining significantly elevated between words compared to adolescents'. We found no significant differences between 12 and 13 years and 14-15 years. These findings indicate that children's engagement of the language network during sentence processing continues to change through middle childhood but stabilizes into adolescence.

Predicting Parkinson's disease trajectory using clinical and neuroimaging baseline measures.

INTRODUCTION: Predictive biomarkers of Parkinson's Disease progression are needed to expedite neuroprotective treatment development and facilitate prognoses for patients. This work uses measures derived from resting-state functional magnetic resonance imaging, including regional homogeneity (ReHo) and fractional amplitude of low frequency fluctuations (fALFF), to predict an individual's current and future severity over up to 4 years and to elucidate the most prognostic brain regions. METHODS: ReHo and fALFF are measured for 82 Parkinson's Disease subjects and used to train machine learning predictors of baseline clinical and future severity at 1 year, 2 years, and 4 years follow-up as measured by the Movement Disorder Society Unified Parkinson's Disease Rating Scale (MDS-UPDRS). Predictive performance is measured with nested cross-validation, validated on an external dataset, and again validated through leave-one-site-out cross-validation. Important predictive features are identified. RESULTS: The models explain up to 30.4% of the variance in current MDS-UPDRS scores, 55.8% of the variance in year 1 scores, and 47.1% of the variance in year 2 scores (p < 0.0001). For distinguishing high and low-severity individuals at each timepoint (MDS-UPDRS score above or below the median, respectively), the models achieve positive predictive values up to 79% and negative predictive values up to 80%. Higher ReHo and fALFF in several regions, including components of the default motor network, predicted lower severity across current and future timepoints. CONCLUSION: These results identify an accurate prognostic neuroimaging biomarker which may be used to better inform enrollment in trials of neuroprotective treatments and enable physicians to counsel their patients.

Prediction of Individual Progression Rate in Parkinson's Disease Using Clinical Measures and Biomechanical Measures of Gait and Postural Stability.

Parkinson's disease (PD) is a common neurological disorder characterized by gait impairment. PD has no cure, and an impediment to developing a treatment is the lack of any accepted method to predict disease progression rate. The primary aim of this study was to develop a model using clinical measures and biomechanical measures of gait and postural stability to predict an individual's PD progression over two years. Data from 160 PD subjects were utilized. Machine learning models, including XGBoost and Feed Forward Neural Networks, were developed using extensive model optimization and cross-validation. The highest performing model was a neural network that used a group of clinical measures, achieved a PPV of 71% in identifying fast progressors, and explained a large portion (37%) of the variance in an individual's progression rate on held-out test data. This demonstrates the potential to predict individual PD progression rate and enrich trials by analyzing clinical and biomechanical measures with machine learning.

Improved motion correction for functional MRI using an omnibus regression model.

Head motion during functional Magnetic Resonance Imaging acquisition can significantly contaminate the neural signal and introduce spurious, distance-dependent changes in signal correlations. This can heavily confound studies of development, aging, and disease. Previous approaches to suppress head motion artifacts have involved sequential regression of nuisance covariates, but this has been shown to reintroduce artifacts. We propose a new motion correction pipeline using an omnibus regression model that avoids this problem by simultaneously regressing out multiple artifacts using the best performing algorithms to estimate each artifact. We quantitatively evaluate its motion artifact suppression performance against sequential regression pipelines using a large heterogeneous dataset (n=151) which includes high-motion subjects and multiple disease phenotypes. The proposed concatenated regression pipeline significantly reduces the association between head motion and functional connectivity while significantly outperforming the traditional sequential regression pipelines in eliminating distance-dependent head motion artifacts.