Statistical Learning Methods for Neuroimaging Data Analysis with Applications

This chapter will review the statistical methods used in RNA sequencing data analysis, including bulk RNA sequencing and single-cell RNA sequencing. RNA sequencing data analysis has been widely used in biomedical and biological research to identify genes associated with certain conditions or diseases. Many statistical methods have been proposed to analyze bulk and single-cell RNA sequencing data. Several studies have compared the performance of different statistical methods for RNA sequencing data analysis through simulation studies and real data evaluations. This chapter will summarize the statistical methods and the evaluation results for comparing different statistical analysis methods used for RNA sequencing data analysis. It will cover the statistical models, model assumptions, and challenges encountered in the RNA sequencing data analysis. It is hoped that this chapter will help researchers learn more about the statistical perspective of the RNA sequencing data analysis and enable them to choose appropriate statistical analysis methods for their own RNA sequencing data analysis.

The analysis of neuroimaging data is frequently used to assist the diagnosis of neurodegenerative disorders such as Alzheimer's disease (AD) or Parkinson's disease (PD) and has become a routine procedure in the clinical practice. During the past decade, the pattern recognition community has proposed a number of machine learning-based systems that automatically analyse neuroimaging data in order to improve the diagnosis. However, the high dimensionality of the data is still a challenge and there is room for improvement. The development of novel classification frameworks as TensorFlow, recently released as open source by Google Inc., represents an opportunity to continue evolving these systems. In this work, we demonstrate several computer-aided diagnosis (CAD) systems based on Deep Neural Networks that improve the diagnosis for AD and PD and outperform those based on classical classifiers. In order to address the small sample size problem we evaluate two dimensionality reduction algorithms based on Principal Component Analysis and Non-Negative Matrix Factorization (NNMF), respectively. The performance of developed CAD systems is assessed using 4 datasets with neuroimaging data of different modalities.

Bayesian model selection for group studies — Revisited

Schizophrenia is a chronic neurobiological disorder whose early detection has attracted significant attention from the clinical, psychiatric, and also artificial intelligence communities. This latter approach has been mainly focused on the analysis of neuroimaging and genetic data. A less explored strategy consists in exploiting the power of natural language processing (NLP) algorithms applied over narrative texts produced by schizophrenic subjects. In this paper, a novel dataset collected from a proper field study is presented. Also, grammatical traits discovered in narrative documents are used to build computational representations of texts, allowing an automatic classification of discourses generated by schizophrenic and non-schizophrenic subjects. The attained results showed that the use of the proposed computational representations along with machine learning techniques enables a novel and precise strategy to automatically detect texts produced by schizophrenic subjects.

The article presents a case of a 54-year-old female patient who, over the course of 2 years, suffered 5 cerebrovascular accidents (CVA) due to infectious arteritis of both internal carotid arteries (ICA) and basilar artery as a complication of sphenoiditis and otitis. According to neuroimaging data, the steno-occlusive process in the ICA developed gradually, starting with the intracranial ICA narrowing with the contrast enhancement by vessel wall, the development of its occlusion six months later, and the detection of the extracranial ICA occlusion with the formation of «flame sign» at its mouth a year later. Repeated examination of the cerebrospinal liquid at an early stage of the disease revealed cytosis up to 367/3 and protein 0.66 g/l. The correct diagnosis was established only after 3 years with a retrospective analysis of clinical, neuroimaging, and laboratory data. Therefore, targeted antibiotic therapy was not carried out, which led to the progression of ICA occlusion and repeated strokes. Infectious arteritis should be taken into account in the differential diagnosis of the causes of the ICA occlusive process.

We begin by reviewing the statistical framework of information theory as applicable to neuroimaging data analysis. A major factor hindering wider adoption of this framework in neuroimaging is the difficulty of estimating information theoretic quantities in practice. We present a novel estimation technique that combines the statistical theory of copulas with the closed form solution for the entropy of Gaussian variables. This results in a general, computationally efficient, flexible, and robust multivariate statistical framework that provides effect sizes on a common meaningful scale, allows for unified treatment of discrete, continuous, unidimensional and multidimensional variables, and enables direct comparisons of representations from behavioral and brain responses across any recording modality. We validate the use of this estimate as a statistical test within a neuroimaging context, considering both discrete stimulus classes and continuous stimulus features. We also present examples of analyses facilitated by these developments, including application of multivariate analyses to MEG planar magnetic field gradients, and pairwise temporal interactions in evoked EEG responses. We show the benefit of considering the instantaneous temporal derivative together with the raw values of M/EEG signals as a multivariate response, how we can separately quantify modulations of amplitude and direction for vector quantities, and how we can measure the emergence of novel information over time in evoked responses. Open‐source Matlab and Python code implementing the new methods accompanies this article. Hum Brain Mapp 38:1541–1573, 2017. © 2016 Wiley Periodicals, Inc.

Autism Spectrum Disorder (ASD) is a complex neurodevelopmental condition characterized by challenges in social interaction, communication, and behavior. Traditional diagnostic methods rely heavily on behavioral assessments, such as the Autism Diagnostic Observation Schedule (ADOS) and Autism Diagnostic Interview-Revised (ADI-R), which are often subjective and limited by the absence of definitive biomarkers. Recent advances in neuroimaging techniques, including structural MRI (sMRI), functional MRI (fMRI), and positron emission tomography (PET), have provided valuable insights into the structural and functional abnormalities associated with ASD. However, the analysis of neuroimaging data is complex, requiring sophisticated machine learning (ML) and artificial intelligence (AI) models. This paper reviews state-of-the-art machine learning approaches, such as Random Forests, Support Vector Machines (SVM), and deep learning models, aimed at improving the diagnostic accuracy of ASD. We particularly focus on multimodal deep learning models that integrate neuroimaging data with behavioral assessments. In this review, we critically evaluate the strengths and limitations of these approaches, and how emerging AI tools not only enhance classification accuracy but also hold potential for identifying biomarkers and tracking the disorder's progression. Our analysis underscores the need for advanced methods that exploit both spatial and temporal features of brain data to better understand ASD, paving the way for improved diagnostics and personalized treatments.

Neuroimaging data are an increasingly important part of etiological studies of neurological and psychiatric disorders. However, mitigating the influence of nuisance variables, including confounders, remains a challenge in image analysis. In studies of Alzheimer's disease, for example, an imbalance in disease rates by age and sex may make it difficult to distinguish between structural patterns in the brain (as measured by neuroimaging scans) attributable to disease progression and those characteristic of typical human aging or sex differences. Concerningly, when not properly accounted for, nuisance variables pose threats to the generalizability and interpretability of findings from these studies. Motivated by this critical issue, in this work, we examine the impact of nuisance variables on feature extraction methods and propose Penalized Decomposition Using Residuals (PeDecURe), a new method for obtaining nuisance variable-adjusted features. PeDecURe estimates primary directions of variation which maximize covariance between partially residualized imaging features and a variable of interest (e.g., Alzheimer's diagnosis) while simultaneously mitigating the influence of nuisance variation through a penalty on the covariance between partially residualized imaging features and those variables. Using features derived using PeDecURe's first direction of variation, we train a highly accurate and generalizable predictive model, as evidenced by its robustness in testing samples with different underlying nuisance variable distributions. We compare PeDecURe to commonly used decomposition methods (principal component analysis (PCA) and partial least squares) as well as a confounder-adjusted variation of PCA. We find that features derived from PeDecURe offer greater accuracy and generalizability and lower correlations with nuisance variables compared with the other methods. While PeDecURe is primarily motivated by challenges that arise in the analysis of neuroimaging data, it is broadly applicable to data sets with highly correlated features, where novel methods to handle nuisance variables are warranted.

Variable screening is a powerful and efficient tool for dimension reduction under ultrahigh dimensional settings. However, most existing methods overlook useful prior knowledge in specific applications. In this work, from a Bayesian modeling perspective, we develop a unified variable screening procedure for the linear regression model. We discuss different constructions of posterior mean screening (PMS) statistics to incorporate different types of prior knowledge according to specific applications. With non-informative prior specifications, PMS is equivalent to high-dimensional ordinary least-square projections (HOLP). We establish the screening consistency property for PMS with different types of prior knowledge. We show that PMS is robust to prior misspecifications; and when the prior knowledge provides correct information on summarizing the true parameter settings, PMS can substantially improve the selection accuracy compared to HOLP and other existing methods. We illustrate our method with extensive simulation studies and an analysis of neuroimaging data.

Ultra-high dimensional variable selection has become increasingly important in analysis of neuroimaging data. For example, in the Autism Brain Imaging Data Exchange (ABIDE) study, neuroscientists are interested in identifying important biomarkers for early detection of the autism spectrum disorder (ASD) using high resolution brain images that include hundreds of thousands voxels. However, most existing methods are not feasible for solving this problem due to their extensive computational costs. In this work, we propose a novel multiresolution variable selection procedure under a Bayesian probit regression framework. It recursively uses posterior samples for coarser-scale variable selection to guide the posterior inference on finer-scale variable selection, leading to very efficient Markov chain Monte Carlo (MCMC) algorithms. The proposed algorithms are computationally feasible for ultra-high dimensional data. Also, our model incorporates two levels of structural information into variable selection using Ising priors: the spatial dependence between voxels and the functional connectivity between anatomical brain regions. Applied to the resting state functional magnetic resonance imaging (R-fMRI) data in the ABIDE study, our methods identify voxel-level imaging biomarkers highly predictive of the ASD, which are biologically meaningful and interpretable. Extensive simulations also show that our methods achieve better performance in variable selection compared to existing methods.

In positron emission tomographic (PET) or magnetic resonance imaging (MRI) neuroimaging studies, spatial smoothing technique is commonly applied to increase the signal to noise ratio and to condition neuroimaging data for subsequent statistical analysis, and to reduce errors associated with registration or spatial normalizations. Usually, the smoothing step is applied to the images without any masking. Thus, some artifacts adjacent but outside of the brain will enter the brain volume. Masking the brain volume before smoothing has been suggested as one way to eliminate the introduced artifact, but it will introduce zero-in (intensities within-mask voxels are reduced) and nonzero-out (intensities outside the mask become non-zero) artifacts. Here we proposed an adaptive smoothing method to reduce the influence of such artifacts. Unlike the conventional smoothing method, the adaptive strategy did not introduce artificial addition (due to nonzero-out artifact) and deletion (due to zero-in), suggesting that such adaptive smoothing methods may be helpful in reducing the influence of non-brain tissue in the analysis of neuroimaging data.

The Enhancing NeuroImaging Genetics through Meta-Analysis (ENIGMA) Consortium is a collaborative network of researchers working together on a range of large-scale studies that integrate data from 70 institutions worldwide. Organized into Working Groups that tackle questions in neuroscience, genetics, and medicine, ENIGMA studies have analyzed neuroimaging data from over 12,826 subjects. In addition, data from 12,171 individuals were provided by the CHARGE consortium for replication of findings, in a total of 24,997 subjects. By meta-analyzing results from many sites, ENIGMA has detected factors that affect the brain that no individual site could detect on its own, and that require larger numbers of subjects than any individual neuroimaging study has currently collected. ENIGMA’s first project was a genome-wide association study identifying common variants in the genome associated with hippocampal volume or intracranial volume. Continuing work is exploring genetic associations with subcortical volumes (ENIGMA2) and white matter microstructure (ENIGMA-DTI). Working groups also focus on understanding how schizophrenia, bipolar illness, major depression and attention deficit/hyperactivity disorder (ADHD) affect the brain. We review the current progress of the ENIGMA Consortium, along with challenges and unexpected discoveries made on the way.

Classical measures of functional connectivity assume that the stationarity of the time courses and the time-invariance of functional connectivity under investigation. These assumptions may not be valid in the real cases. Also, they are bivariate measures and may not provide the directional information flow between brain units. A new approach is proposed to tackle these problems. A statistics reasoning shows that the short-length time course is more likely to be stationary than the long-length time course. Thus, the entire time course under investigation is divided into short segments with the proper length. Magnitude squared coherence (in spectrum domain) is computed to assess functional connectivity on these segments, hence, provides a dynamic measure of functional connectivity. The averaged magnitude squared coherence over the segments gives an overall measure of functional connectivity. This approach has been applied to several neuroimaging data analysis. The results and the interpretations / predictions are in good agreement. Mutual coherence (in time domain) is computed to assess functional connectivity, hence, provides an insight on directional information flow. By using grid computing, this approach will be extended from the bivariate to the multivariate.

When psychiatrists see a patient, they consider a diagnosis, estimate a prognosis and treat accordingly, but very few of these decisions are informed by objective tests. Recent advances in neuroimaging data analysis have shown that brain scans can make powerful diagnostic and prognostic predictions in patients with psychosis and depression.

Analysis of time-evolving data is crucial to understand the functioning of dynamic systems such as the brain. For instance, analysis of functional magnetic resonance imaging (fMRI) data collected during a task may reveal spatial regions of interest, and how they evolve during the task. However, capturing underlying spatial patterns as well as their change in time is challenging. The traditional approach in fMRI data analysis is to assume that underlying spatial regions of interest are static. In this article, using fractional amplitude of low-frequency fluctuations (fALFF) as an effective way to summarize the variability in fMRI data collected during a task, we arrange time-evolving fMRI data as a subjects by voxels by time windows tensor, and analyze the tensor using a tensor factorization-based approach called a PARAFAC2 model to reveal spatial dynamics. The PARAFAC2 model jointly analyzes data from multiple time windows revealing subject-mode patterns, evolving spatial regions (also referred to as networks) and temporal patterns. We compare the PARAFAC2 model with matrix factorization-based approaches relying on independent components, namely, joint independent component analysis (ICA) and independent vector analysis (IVA), commonly used in neuroimaging data analysis. We assess the performance of the methods in terms of capturing evolving networks through extensive numerical experiments demonstrating their modeling assumptions. In particular, we show that (i) PARAFAC2 provides a compact representation in all modes, i.e., subjects, time, and voxels, revealing temporal patterns as well as evolving spatial networks, (ii) joint ICA is as effective as PARAFAC2 in terms of revealing evolving networks but does not reveal temporal patterns, (iii) IVA's performance depends on sample size, data distribution and covariance structure of underlying networks. When these assumptions are satisfied, IVA is as accurate as the other methods, (iv) when subject-mode patterns differ from one time window to another, IVA is the most accurate. Furthermore, we analyze real fMRI data collected during a sensory motor task, and demonstrate that a component indicating statistically significant group difference between patients with schizophrenia and healthy controls is captured, which includes primary and secondary motor regions, cerebellum, and temporal lobe, revealing a meaningful spatial map and its temporal change.