Generalized Scalar-on-Image Regression Models via Total Variation

Generalized functional linear model with a point process predictor

Point process data have become increasingly popular these days. For example, many of the data captured in electronic health records (EHR) are in the format of point process data. It is of great interest to study the association between a point process predictor and a scalar response using generalized functional linear regression models. Various generalized functional linear regression models have been developed under different settings in the past decades. However, existing methods can only deal with functional or longitudinal predictors, not point process predictors. In this article, we propose a novel generalized functional linear regression model for a point process predictor. Our proposed model is based on the joint modeling framework, where we adopt a log-Gaussian Cox process model for the point process predictor and a generalized linear regression model for the outcome. We also develop a new algorithm for fast model estimation based on the Gaussian variational approximation method. We conduct extensive simulation studies to evaluate the performance of our proposed method and compare it to competing methods. The performance of our proposed method is further demonstrated on an EHR dataset of patients admitted into the intensive care units of the Beth Israel Deaconess Medical Center between 2001 and 2008.

Bayesian interaction selection model for multimodal neuroimaging data analysis

Multimodality or multiconstruct data arise increasingly in functional neuroimaging studies to characterize brain activity under different cognitive states. Relying on those high-resolution imaging collections, it is of great interest to identify predictive imaging markers and intermodality interactions with respect to behavior outcomes. Currently, most of the existing variable selection models do not consider predictive effects from interactions, and the desired higher-order terms can only be included in the predictive mechanism following a two-step procedure, suffering from potential misspecification. In this paper, we propose a unified Bayesian prior model to simultaneously identify main effect features and intermodality interactions within the same inference platform in the presence of high-dimensional data. To accommodate the brain topological information and correlation between modalities, our prior is designed by compiling the intermediate selection status of sequential partitions in light of the data structure and brain anatomical architecture, so that we can improve posterior inference and enhance biological plausibility. Through extensive simulations, we show the superiority of our approach in main and interaction effects selection, and prediction under multimodality data. Applying the method to the Adolescent Brain Cognitive Development (ABCD) study, we characterize the brain functional underpinnings with respect to general cognitive ability under different memory load conditions.

Identifying regions of interest in mammogram images

Screening mammography is the primary preventive strategy for early detection of breast cancer and an essential input to breast cancer risk prediction and application of prevention/risk management guidelines. Identifying regions of interest within mammogram images that are associated with 5- or 10-year breast cancer risk is therefore clinically meaningful. The problem is complicated by the irregular boundary issue posed by the semi-circular domain of the breast area within mammograms. Accommodating the irregular domain is especially crucial when identifying regions of interest, as the true signal comes only from the semi-circular domain of the breast region, and noise elsewhere. We address these challenges by introducing a proportional hazards model with imaging predictors characterized by bivariate splines over triangulation. The model sparsity is enforced with the group lasso penalty function. We apply the proposed method to the motivating Joanne Knight Breast Health Cohort to illustrate important risk patterns and show that the proposed method is able to achieve higher discriminatory performance.

Additive hazards model with time-varying coefficients and imaging predictors

Conventional hazard regression analyses frequently assume constant regression coefficients and scalar covariates. However, some covariate effects may vary with time. Moreover, medical imaging has become an increasingly important tool in screening, diagnosis, and prognosis of various diseases, given its information visualization and quantitative assessment. This study considers an additive hazards model with time-varying coefficients and imaging predictors to examine the dynamic effects of potential scalar and imaging risk factors for the failure of interest. We develop a two-stage approach that comprises the high-dimensional functional principal component analysis technique in the first stage and the counting process-based estimating equation approach in the second stage. In addition, we construct the pointwise confidence intervals for the proposed estimators and provide a significance test for the effects of scalar and imaging covariates. Simulation studies demonstrate the satisfactory performance of the proposed method. An application to the Alzheimer's disease neuroimaging initiative study further illustrates the utility of the methodology.

Robust High-Dimensional Regression with Coefficient Thresholding and its Application to Imaging Data Analysis

It is important to develop statistical techniques to analyze high-dimensional data in the presence of both complex dependence and possible heavy tails and outliers in real-world applications such as imaging data analyses. We propose a new robust high-dimensional regression with coefficient thresholding, in which an efficient nonconvex estimation procedure is proposed through a thresholding function and the robust Huber loss. The proposed regularization method accounts for complex dependence structures in predictors and is robust against heavy tails and outliers in outcomes. Theoretically, we rigorously analyze the landscape of the population and empirical risk functions for the proposed method. The fine landscape enables us to establish both statistical consistency and computational convergence under the high-dimensional setting. We also present an extension to incorporate spatial information into the proposed method. Finite-sample properties of the proposed methods are examined by extensive simulation studies. An application concerns a scalar-on-image regression analysis for an association of psychiatric disorder measured by the general factor of psychopathology with features extracted from the task functional MRI data in the Adolescent Brain Cognitive Development (ABCD) study.

Multi-task Learning with High-Dimensional Noisy Images

Recent medical imaging studies have given rise to distinct but inter-related datasets corresponding to multiple experimental tasks or longitudinal visits. Standard scalar-on-image regression models that fit each dataset separately are not equipped to leverage information across inter-related images, and existing multi-task learning approaches are compromised by the inability to account for the noise that is often observed in images. We propose a novel joint scalar-on-image regression framework involving wavelet-based image representations with grouped penalties that are designed to pool information across inter-related images for joint learning, and which explicitly accounts for noise in high-dimensional images via a projection-based approach. In the presence of non-convexity arising due to noisy images, we derive non-asymptotic error bounds under non-convex as well as convex grouped penalties, even when the number of voxels increases exponentially with sample size. A projected gradient descent algorithm is used for computation, which is shown to approximate the optimal solution via well-defined non-asymptotic optimization error bounds under noisy images. Extensive simulations and application to a motivating longitudinal Alzheimer's disease study illustrate significantly improved predictive ability and greater power to detect true signals, that are simply missed by existing methods without noise correction due to the attenuation to null phenomenon.

From multivariate to functional data analysis: fundamentals, recent developments, and emerging areas

Functional data analysis (FDA), which is a branch of statistics on modeling infinite dimensional random vectors resided in functional spaces, has become a major research area for Journal of Multivariate Analysis. We review some fundamental concepts of FDA, their origins and connections from multivariate analysis, and some of its recent developments, including multi-level functional data analysis, high-dimensional functional regression, and dependent functional data analysis. We also discuss the impact of these new methodology developments on genetics, plant science, wearable device data analysis, image data analysis, and business analytics. Two real data examples are provided to motivate our discussions.

Brain Regions Identified as Being Associated with Verbal Reasoning through the Use of Imaging Regression via Internal Variation

Brain-imaging data have been increasingly used to understand intellectual disabilities. Despite significant progress in biomedical research, the mechanisms for most of the intellectual disabilities remain unknown. Finding the underlying neurological mechanisms has been proved difficult, especially in children due to the rapid development of their brains. We investigate verbal reasoning, which is a reliable measure of individuals' general intellectual abilities, and develop a class of high-order imaging regression models to identify brain subregions which might be associated with this specific intellectual ability. A key novelty of our method is to take advantage of spatial brain structures, and specifically the piecewise smooth nature of most imaging coefficients in the form of high-order tensors. Our approach provides an effective and urgently needed method for identifying brain subregions potentially underlying certain intellectual disabilities. The idea behind our approach is a carefully constructed concept called Internal Variation (IV). The IV employs tensor decomposition and provides a computationally feasible substitution for Total Variation (TV), which has been considered in the literature to deal with similar problems but is problematic in high order tensor regression. Before applying our method to analyze the real data, we conduct comprehensive simulation studies to demonstrate the validity of our method in imaging signal identification. Then, we present our results from the analysis of a dataset based on the Philadelphia Neurodevelopmental Cohort for which we preprocessed the data including re-orienting, bias-field correcting, extracting, normalizing and registering the magnetic resonance images from 978 individuals. Our analysis identified a subregion across the cingulate cortex and the corpus callosum as being associated with individuals' verbal reasoning ability, which, to the best of our knowledge, is a novel region that has not been reported in the literature. This finding is useful in further investigation of functional mechansims for verbal reasoning.

TENSOR QUANTILE REGRESSION WITH APPLICATION TO ASSOCIATION BETWEEN NEUROIMAGES AND HUMAN INTELLIGENCE

Human intelligence is usually measured by well-established psychometric tests through a series of problem solving. The recorded cognitive scores are continuous but usually heavy-tailed with potential outliers and violating the normality assumption. Meanwhile, magnetic resonance imaging (MRI) provides an unparalleled opportunity to study brain structures and cognitive ability. Motivated by association studies between MRI images and human intelligence, we propose a tensor quantile regression model, which is a general and robust alternative to the commonly used scalar-on-image linear regression. Moreover, we take into account rich spatial information of brain structures, incorporating low-rankness and piece-wise smoothness of imaging coefficients into a regularized regression framework. We formulate the optimization problem as a sequence of penalized quantile regressions with a generalized Lasso penalty based on tensor decomposition, and develop a computationally efficient alternating direction method of multipliers algorithm (ADMM) to estimate the model components. Extensive numerical studies are conducted to examine the empirical performance of the proposed method and its competitors. Finally, we apply the proposed method to a large-scale important dataset: the Human Connectome Project. We find that the tensor quantile regression can serve as a prognostic tool to assess future risk of cognitive impairment progression. More importantly, with the proposed method, we are able to identify the most activated brain subregions associated with quantiles of human intelligence. The prefrontal and anterior cingulate cortex are found to be mostly associated with lower and upper quantile of fluid intelligence. The insular cortex associated with median of fluid intelligence is a rarely reported region.

Partially Observed Dynamic Tensor Response Regression

In modern data science, dynamic tensor data prevail in numerous applications. An important task is to characterize the relationship between dynamic tensor datasets and external covariates. However, the tensor data are often only partially observed, rendering many existing methods inapplicable. In this article, we develop a regression model with a partially observed dynamic tensor as the response and external covariates as the predictor. We introduce the low-rankness, sparsity, and fusion structures on the regression coefficient tensor, and consider a loss function projected over the observed entries. We develop an efficient nonconvex alternating updating algorithm, and derive the finite-sample error bound of the actual estimator from each step of our optimization algorithm. Unobserved entries in the tensor response have imposed serious challenges. As a result, our proposal differs considerably in terms of estimation algorithm, regularity conditions, as well as theoretical properties, compared to the existing tensor completion or tensor response regression solutions. We illustrate the efficacy of our proposed method using simulations and two real applications, including a neuroimaging dementia study and a digital advertising study.

High-Dimensional Spatial Quantile Function-on-Scalar Regression

This article develops a novel spatial quantile function-on-scalar regression model, which studies the conditional spatial distribution of a high-dimensional functional response given scalar predictors. With the strength of both quantile regression and copula modeling, we are able to explicitly characterize the conditional distribution of the functional or image response on the whole spatial domain. Our method provides a comprehensive understanding of the effect of scalar covariates on functional responses across different quantile levels and also gives a practical way to generate new images for given covariate values. Theoretically, we establish the minimax rates of convergence for estimating coefficient functions under both fixed and random designs. We further develop an efficient primal-dual algorithm to handle high-dimensional image data. Simulations and real data analysis are conducted to examine the finite-sample performance.

A spatial Bayesian latent factor model for image-on-image regression

Image-on-image regression analysis, using images to predict images, is a challenging task, due to (1) the high dimensionality and (2) the complex spatial dependence structures in image predictors and image outcomes. In this work, we propose a novel image-on-image regression model, by extending a spatial Bayesian latent factor model to image data, where low-dimensional latent factors are adopted to make connections between high-dimensional image outcomes and image predictors. We assign Gaussian process priors to the spatially varying regression coefficients in the model, which can well capture the complex spatial dependence among image outcomes as well as that among the image predictors. We perform simulation studies to evaluate the out-of-sample prediction performance of our method compared with linear regression and voxel-wise regression methods for different scenarios. The proposed method achieves better prediction accuracy by effectively accounting for the spatial dependence and efficiently reduces image dimensions with latent factors. We apply the proposed method to analysis of multimodal image data in the Human Connectome Project where we predict task-related contrast maps using subcortical volumetric seed maps.

Low-Rank Tensor Train Coefficient Array Estimation for Tensor-on-Tensor Regression

The tensor-on-tensor regression can predict a tensor from a tensor, which generalizes most previous multilinear regression approaches, including methods to predict a scalar from a tensor, and a tensor from a scalar. However, the coefficient array could be much higher dimensional due to both high-order predictors and responses in this generalized way. Compared with the current low CANDECOMP/PARAFAC (CP) rank approximation-based method, the low tensor train (TT) approximation can further improve the stability and efficiency of the high or even ultrahigh-dimensional coefficient array estimation. In the proposed low TT rank coefficient array estimation for tensor-on-tensor regression, we adopt a TT rounding procedure to obtain adaptive ranks, instead of selecting ranks by experience. Besides, an l₂ constraint is imposed to avoid overfitting. The hierarchical alternating least square is used to solve the optimization problem. Numerical experiments on a synthetic data set and two real-life data sets demonstrate that the proposed method outperforms the state-of-the-art methods in terms of prediction accuracy with comparable computational complexity, and the proposed method is more computationally efficient when the data are high dimensional with small size in each mode.

Nonparametric matrix response regression with application to brain imaging data analysis

With the rapid growth of neuroimaging technologies, a great effort has been dedicated recently to investigate the dynamic changes in brain activity. Examples include time course calcium imaging and dynamic brain functional connectivity. In this paper, we propose a novel nonparametric matrix response regression model to characterize the nonlinear association between 2D image outcomes and predictors such as time and patient information. Our estimation procedure can be formulated as a nuclear norm regularization problem, which can capture the underlying low-rank structure of the dynamic 2D images. We present a computationally efficient algorithm, derive the asymptotic theory, and show that the method outperforms other existing approaches in simulations. We then apply the proposed method to a calcium imaging study for estimating the change of fluorescent intensities of neurons, and an electroencephalography study for a comparison in the dynamic connectivity covariance matrices between alcoholic and control individuals. For both studies, the method leads to a substantial improvement in prediction error.

Regularized matrix data clustering and its application to image analysis

We propose a novel regularized mixture model for clustering matrix-valued data. The proposed method assumes a separable covariance structure for each cluster and imposes a sparsity structure (eg, low rankness, spatial sparsity) for the mean signal of each cluster. We formulate the problem as a finite mixture model of matrix-normal distributions with regularization terms, and then develop an expectation maximization type of algorithm for efficient computation. In theory, we show that the proposed estimators are strongly consistent for various choices of penalty functions. Simulation and two applications on brain signal studies confirm the excellent performance of the proposed method including a better prediction accuracy than the competitors and the scientific interpretability of the solution.

Logistic regression with image covariates via the combination of L1 and Sobolev regularizations

The use of image covariates to build a classification model has lots of impact in various fields, such as computer science, medicine, and so on. The aim of this paper is to develop an estimation method for logistic regression model with image covariates. We propose a novel regularized estimation approach, where the regularization is a combination of L1 regularization and Sobolev norm regularization. The L1 penalty can perform variable selection, while the Sobolev norm penalty can capture the shape edges information of image data. We develop an efficient algorithm for the optimization problem. We also establish a nonasymptotic error bound on parameter estimation. Simulated studies and a real data application demonstrate that our proposed method performs very well.

Bayesian Scalar on Image Regression With Nonignorable Nonresponse

Medical imaging has become an increasingly important tool in screening, diagnosis, prognosis, and treatment of various diseases given its information visualization and quantitative assessment. The aim of this article is to develop a Bayesian scalar-on-image regression model to integrate high-dimensional imaging data and clinical data to predict cognitive, behavioral, or emotional outcomes, while allowing for nonignorable missing outcomes. Such a nonignorable nonresponse consideration is motivated by examining the association between baseline characteristics and cognitive abilities for 802 Alzheimer patients enrolled in the Alzheimer's Disease Neuroimaging Initiative 1 (ADNI1), for which data are partially missing. Ignoring such missing data may distort the accuracy of statistical inference and provoke misleading results. To address this issue, we propose an imaging exponential tilting model to delineate the data missing mechanism and incorporate an instrumental variable to facilitate model identifiability followed by a Bayesian framework with Markov chain Monte Carlo algorithms to conduct statistical inference. This approach is validated in simulation studies where both the finite sample performance and asymptotic properties are evaluated and compared with the model with fully observed data and that with a misspecified ignorable missing mechanism. Our proposed methods are finally carried out on the ADNI1 dataset, which turns out to capture both of those clinical risk factors and imaging regions consistent with the existing literature that exhibits clinical significance. Supplementary materials for this article, including a standardized description of the materials available for reproducing the work, are available as an online supplement.

Matrix Linear Discriminant Analysis

We propose a novel linear discriminant analysis (LDA) approach for the classification of high-dimensional matrix-valued data that commonly arises from imaging studies. Motivated by the equivalence of the conventional LDA and the ordinary least squares, we consider an efficient nuclear norm penalized regression that encourages a low-rank structure. Theoretical properties including a nonasymptotic risk bound and a rank consistency result are established. Simulation studies and an application to electroencephalography data show the superior performance of the proposed method over the existing approaches.

A Bayesian Spatial Model to Predict Disease Status Using Imaging Data From Various Modalities

Relating disease status to imaging data stands to increase the clinical significance of neuroimaging studies. Many neurological and psychiatric disorders involve complex, systems-level alterations that manifest in functional and structural properties of the brain and possibly other clinical and biologic measures. We propose a Bayesian hierarchical model to predict disease status, which is able to incorporate information from both functional and structural brain imaging scans. We consider a two-stage whole brain parcellation, partitioning the brain into 282 subregions, and our model accounts for correlations between voxels from different brain regions defined by the parcellations. Our approach models the imaging data and uses posterior predictive probabilities to perform prediction. The estimates of our model parameters are based on samples drawn from the joint posterior distribution using Markov Chain Monte Carlo (MCMC) methods. We evaluate our method by examining the prediction accuracy rates based on leave-one-out cross validation, and we employ an importance sampling strategy to reduce the computation time. We conduct both whole-brain and voxel-level prediction and identify the brain regions that are highly associated with the disease based on the voxel-level prediction results. We apply our model to multimodal brain imaging data from a study of Parkinson's disease. We achieve extremely high accuracy, in general, and our model identifies key regions contributing to accurate prediction including caudate, putamen, and fusiform gyrus as well as several sensory system regions.

Scalar-on-Image Regression via the Soft-Thresholded Gaussian Process

This work concerns spatial variable selection for scalar-on-image regression. We propose a new class of Bayesian nonparametric models and develop an efficient posterior computational aigorithm. The proposed soft-thresholded Gaussian process provides large prior support over the class of piecewise-smooth, sparse, and continuous spatially-varying regression coefficient functions. In addition, under some mild regularity conditions the soft-thresholded Gaussian proess prior leads to the posterior consistency for parameter estimation and variable selection for scalar-on-image regression, even when the number of predictors is larger than the sample size. The proposed method is compared to alternatives via simulation and applied to an electroen-cephalography study of alcoholism.

Partition-based ultrahigh-dimensional variable screening

Traditional variable selection methods are compromised by overlooking useful information on covariates with similar functionality or spatial proximity, and by treating each covariate independently. Leveraging prior grouping information on covariates, we propose partition-based screening methods for ultrahigh-dimensional variables in the framework of generalized linear models. We show that partition-based screening exhibits the sure screening property with a vanishing false selection rate, and we propose a data-driven partition screening framework with unavailable or unreliable prior knowledge on covariate grouping and investigate its theoretical properties. We consider two special cases: correlation-guided partitioning and spatial location- guided partitioning. In the absence of a single partition, we propose a theoretically justified strategy for combining statistics from various partitioning methods. The utility of the proposed methods is demonstrated via simulation and analysis of functional neuroimaging data.