SCoRS--A Method Based on Stability for Feature Selection and Mapping inNeuroimaging [corrected]

Stable multivariate lesion symptom mapping

Multivariate lesion-symptom mapping (MLSM) considers lesion information across the entire brain to predict impairments. The strength of this approach is also its weakness-considering many brain features together synergistically can uncover complex brain-behavior relationships but exposes a high-dimensional feature space that a model is expected to learn. Successfully distinguishing between features in this landscape can be difficult for models, particularly in the presence of irrelevant or redundant features. Here, we propose stable multivariate lesion-symptom mapping (sMLSM), which integrates the identification of reliable features with stability selection into conventional MLSM and describe our open-source MATLAB implementation. Usage is showcased with our publicly available dataset of chronic stroke survivors (N=167) and further validated in our independent public acute stroke dataset (N = 1106). We demonstrate that sMLSM eliminates inconsistent features highlighted by MLSM, reduces variation in feature weights, enables the model to learn more complex patterns of brain damage, and improves model accuracy for predicting aphasia severity in a way that tends to be robust regarding the choice of parameters for identifying reliable features. Critically, sMLSM more consistently outperforms predictions based on lesion size alone. This advantage is evident starting at modest sample sizes (N>75). Spatial distribution of feature importance is different in sMLSM, which highlights the features identified by univariate lesion symptom mapping while also implicating select regions emphasized by MLSM. Beyond improved prediction accuracy, sMLSM can offer deeper insight into reliable biomarkers of impairment, informing our understanding of neurobiology.

Groupwise structural sparsity for discriminative voxels identification

This paper investigates the selection of voxels for functional Magnetic Resonance Imaging (fMRI) brain data. We aim to identify a comprehensive set of discriminative voxels associated with human learning when exposed to a neutral visual stimulus that predicts an aversive outcome. However, due to the nature of the unconditioned stimuli (typically a noxious stimulus), it is challenging to obtain sufficient sample sizes for psychological experiments, given the tolerability of the subjects and ethical considerations. We propose a stable hierarchical voting (SHV) mechanism based on stability selection to address this challenge. This mechanism enables us to evaluate the quality of spatial random sampling and minimizes the risk of false and missed detections. We assess the performance of the proposed algorithm using simulated and publicly available datasets. The experiments demonstrate that the regularization strategy choice significantly affects the results' interpretability. When applying our algorithm to our collected fMRI dataset, it successfully identifies sparse and closely related patterns across subjects and displays stable weight maps for three experimental phases under the fear conditioning paradigm. These findings strongly support the causal role of aversive conditioning in altering visual-cortical activity.

A novel neighborhood rough set-based feature selection method and its application to biomarker identification of schizophrenia

Feature selection can disclose biomarkers of mental disorders that have unclear biological mechanisms. Although neighborhood rough set (NRS) has been applied to discover important sparse features, it has hardly ever been utilized in neuroimaging-based biomarker identification, probably due to the inadequate feature evaluation metric and incomplete information provided under a single-granularity. Here, we propose a new NRS-based feature selection method and successfully identify brain functional connectivity biomarkers of schizophrenia (SZ) using functional magnetic resonance imaging (fMRI) data. Specifically, we develop a new weighted metric based on NRS combined with information entropy to evaluate the capacity of features in distinguishing different groups. Inspired by multi-granularity information maximization theory, we further take advantage of the complementary information from different neighborhood sizes via a multi-granularity fusion to obtain the most discriminative and stable features. For validation, we compare our method with six popular feature selection methods using three public omics datasets as well as resting-state fMRI data of 393 SZ patients and 429 healthy controls. Results show that our method obtained higher classification accuracies on both omics data (100.0%, 88.6%, and 72.2% for three omics datasets, respectively) and fMRI data (93.9% for main dataset, and 76.3% and 83.8% for two independent datasets, respectively). Moreover, our findings reveal biologically meaningful substrates of SZ, notably involving the connectivity between the thalamus and superior temporal gyrus as well as between the postcentral gyrus and calcarine gyrus. Taken together, we propose a new NRS-based feature selection method that shows the potential of exploring effective and sparse neuroimaging-based biomarkers of mental disorders.

A new scheme of global feature management improved the performance and stability of radiomics model: a study based on CT images of acute brainstem infarction

OBJECTIVE

The performance and stability of radiomics model caused by dimension reduction remain being confronted with major challenges. In this study, we aimed to propose a new scheme of global feature management independent of dimension reduction to improve it.

METHODS

The non-contrast computed tomography (NCCT) images of acute brainstem infarction (ABI) from two medical centers were used as test and validation sets. A new scheme was constructed based on global feature management, and the traditional scheme dependent on dimension reduction was used as control. The radiomic features of NCCT images were extracted in Matlab R2013a. The performance of prediction model was evaluated by the generalized linear model (GLM) and multivariate logistic regression. And, the stability of radiomics model was evaluated with the difference of area under curve (AUC) between the test and validation sets.

RESULTS

Compared with the traditional scheme, the new scheme presented a similar detection performance (AUC: 0.875 vs. 0.883), yet a better performance in predicting prognosis (AUC: 0.864, OR = 0.917, p = 0.021 vs. AUC:0.806, OR = 0.972, p = 0.007). All these results were well verified in an independent validation set. Moreover, the new scheme showed stronger stability in both the detection model (ΔAUC: 0.013 vs. 0.039) and prediction model (ΔAUC = 0.004 vs. 0.044).

CONCLUSION

Although there might be several limitations, this study proved that the scheme of global feature management independent of dimension reduction could be a powerful supplement to the radiomics methodology.

KEY POINTS

• The new scheme (S_wavelet) presented similar detection performances for ABI with the traditional scheme. • A better predictive performance for END was found in the new scheme (S_wavelet) compared with the traditional scheme. • Stronger model stability was found in both the detection and prediction models based on the new scheme.

A neural network approach to optimising treatments for depression using data from specialist and community psychiatric services in Australia, New Zealand and Japan

This study investigated the application of a recurrent neural network for optimising pharmacological treatment for depression. A clinical dataset of 458 participants from specialist and community psychiatric services in Australia, New Zealand and Japan were extracted from an existing custom-built, web-based tool called Psynary . This data, which included baseline and self-completed reviews, was used to train and refine a novel algorithm which was a fully connected network feature extractor and long short-term memory algorithm was firstly trained in isolation and then integrated and annealed using slow learning rates due to the low dimensionality of the data. The accuracy of predicting depression remission before processing patient review data was 49.8%. After processing only 2 reviews, the accuracy was 76.5%. When considering a change in medication, the precision of changing medications was 97.4% and the recall was 71.4% . The medications with predicted best results were antipsychotics (88%) and selective serotonin reuptake inhibitors (87.9%). This is the first study that has created an all-in-one algorithm for optimising treatments for all subtypes of depression. Reducing treatment optimisation time for patients suffering with depression may lead to earlier remission and hence reduce the high levels of disability associated with the condition. Furthermore, in a setting where mental health conditions are increasing strain on mental health services, the utilisation of web-based tools for remote monitoring and machine/deep learning algorithms may assist clinicians in both specialist and primary care in extending specialist mental healthcare to a larger patient community.

Connectome-based evidence for creative thinking as an emergent property of ordinary cognitive operations

Creative thinking is a hallmark of human cognition, which enables us to generate novel and useful ideas. Nevertheless, its emergence within the macro-scale neurocognitive circuitry remains largely unknown. Using resting-state fMRI data from two large population samples (SWU: n = 931; HCP: n = 1001) and a novel "travelling pattern prediction analysis", here we identified the modularized functional connectivity patterns linked to creative thinking ability, which concurrently explained individual variability across ordinary cognitive abilities such as episodic memory, working memory and relational processing. Further interrogation of this neural pattern with graph theoretical tools revealed both hub-like brain structures and globally-efficient information transfer paths that together may facilitate higher creative thinking ability through the convergence of distinct cognitive operations. Collectively, our results provide reliable evidence for the hypothesized emergence of creative thinking from core cognitive components through neural integration, and thus allude to a significant theoretical advancement in the study of creativity.

Denouements of machine learning and multimodal diagnostic classification of Alzheimer's disease

Alzheimer's disease (AD) is the most common type of dementia. The exact cause and treatment of the disease are still unknown. Different neuroimaging modalities, such as magnetic resonance imaging (MRI), positron emission tomography, and single-photon emission computed tomography, have played a significant role in the study of AD. However, the effective diagnosis of AD, as well as mild cognitive impairment (MCI), has recently drawn large attention. Various technological advancements, such as robots, global positioning system technology, sensors, and machine learning (ML) algorithms, have helped improve the diagnostic process of AD. This study aimed to determine the influence of implementing different ML classifiers in MRI and analyze the use of support vector machines with various multimodal scans for classifying patients with AD/MCI and healthy controls. Conclusions have been drawn in terms of employing different classifier techniques and presenting the optimal multimodal paradigm for the classification of AD.

Towards a brain-based predictome of mental illness

Neuroimaging-based approaches have been extensively applied to study mental illness in recent years and have deepened our understanding of both cognitively healthy and disordered brain structure and function. Recent advancements in machine learning techniques have shown promising outcomes for individualized prediction and characterization of patients with psychiatric disorders. Studies have utilized features from a variety of neuroimaging modalities, including structural, functional, and diffusion magnetic resonance imaging data, as well as jointly estimated features from multiple modalities, to assess patients with heterogeneous mental disorders, such as schizophrenia and autism. We use the term "predictome" to describe the use of multivariate brain network features from one or more neuroimaging modalities to predict mental illness. In the predictome, multiple brain network-based features (either from the same modality or multiple modalities) are incorporated into a predictive model to jointly estimate features that are unique to a disorder and predict subjects accordingly. To date, more than 650 studies have been published on subject-level prediction focusing on psychiatric disorders. We have surveyed about 250 studies including schizophrenia, major depression, bipolar disorder, autism spectrum disorder, attention-deficit hyperactivity disorder, obsessive-compulsive disorder, social anxiety disorder, posttraumatic stress disorder, and substance dependence. In this review, we present a comprehensive review of recent neuroimaging-based predictomic approaches, current trends, and common shortcomings and share our vision for future directions.

Regularized Bagged Canonical Component Analysis for Multiclass Learning in Brain Imaging

A fundamental problem of supervised learning algorithms for brain imaging applications is that the number of features far exceeds the number of subjects. In this paper, we propose a combined feature selection and extraction approach for multiclass problems. This method starts with a bagging procedure which calculates the sign consistency of the multivariate analysis (MVA) projection matrix feature-wise to determine the relevance of each feature. This relevance measure provides a parsimonious matrix, which is combined with a hypothesis test to automatically determine the number of selected features. Then, a novel MVA regularized with the sign and magnitude consistency of the features is used to generate a reduced set of summary components providing a compact data description. We evaluated the proposed method with two multiclass brain imaging problems: 1) the classification of the elderly subjects in four classes (cognitively normal, stable mild cognitive impairment (MCI), MCI converting to AD in 3 years, and Alzheimer's disease) based on structural brain imaging data from the ADNI cohort; 2) the classification of children in 3 classes (typically developing, and 2 types of Attention Deficit/Hyperactivity Disorder (ADHD)) based on functional connectivity. Experimental results confirmed that each brain image (defined by 29.852 features in the ADNI database and 61.425 in the ADHD) could be represented with only 30 - 45% of the original features. Furthermore, this information could be redefined into two or three summary components, providing not only a gain of interpretability but also classification rate improvements when compared to state-of-art reference methods.

Identifying factors associated with opioid cessation in a biracial sample using machine learning

AIM

Racial disparities in opioid use disorder (OUD) management exist, however, and there is limited research on factors that influence opioid cessation in different population groups.

METHODS

We employed multiple machine learning prediction algorithms least absolute shrinkage and selection operator, random forest, deep neural network, and support vector machine to assess factors associated with ceasing opioid use in a sample of 1,192 African Americans (AAs) and 2,557 individuals of European ancestry (EAs) who met Diagnostic and Statistical Manual of Mental Disorders, 5th Edition criteria for OUD. Values for nearly 4,000 variables reflecting demographics, alcohol and other drug use, general health, non-drug use behaviors, and diagnoses for other psychiatric disorders, were obtained for each participant from the Semi-Structured Assessment for Drug Dependence and Alcoholism, a detailed semi-structured interview.

RESULTS

Support vector machine models performed marginally better on average than other machine learning methods with maximum prediction accuracies of 75.4% in AAs and 79.4% in EAs. Subsequent stepwise regression considered the 83 most highly ranked variables across all methods and models and identified less recent cocaine use (AAs: odds ratio (OR) = 1.82, P = 9.19 × 10-5; EAs: OR = 1.91, P = 3.30 × 10-15), shorter duration of opioid use (AAs: OR = 0.55, P = 5.78 × 10-6; EAs: OR = 0.69, P = 3.01 × 10-7), and older age (AAs: OR = 2.44, P = 1.41 × 10-12; EAs: OR = 2.00, P = 5.74 × 10-9) as the strongest independent predictors of opioid cessation in both AAs and EAs. Attending self-help groups for OUD was also an independent predictor (P < 0.05) in both population groups, while less gambling severity (OR = 0.80, P = 3.32 × 10-2) was specific to AAs and post-traumatic stress disorder recovery (OR = 1.93, P = 7.88 × 10-5), recent antisocial behaviors (OR = 0.64, P = 2.69 × 10-3), and atheism (OR = 1.45, P = 1.34 × 10-2) were specific to EAs. Factors related to drug use comprised about half of the significant independent predictors in both AAs and EAs, with other predictors related to non-drug use behaviors, psychiatric disorders, overall health, and demographics.

CONCLUSIONS

These proof-of-concept findings provide avenues for hypothesis-driven analysis, and will lead to further research on strategies to improve OUD management in EAs and AAs.

Multiple Holdouts With Stability: Improving the Generalizability of Machine Learning Analyses of Brain-Behavior Relationships

BACKGROUND

In 2009, the National Institute of Mental Health launched the Research Domain Criteria, an attempt to move beyond diagnostic categories and ground psychiatry within neurobiological constructs that combine different levels of measures (e.g., brain imaging and behavior). Statistical methods that can integrate such multimodal data, however, are often vulnerable to overfitting, poor generalization, and difficulties in interpreting the results.

METHODS

We propose an innovative machine learning framework combining multiple holdouts and a stability criterion with regularized multivariate techniques, such as sparse partial least squares and kernel canonical correlation analysis, for identifying hidden dimensions of cross-modality relationships. To illustrate the approach, we investigated structural brain-behavior associations in an extensively phenotyped developmental sample of 345 participants (312 healthy and 33 with clinical depression). The brain data consisted of whole-brain voxel-based gray matter volumes, and the behavioral data included item-level self-report questionnaires and IQ and demographic measures.

RESULTS

Both sparse partial least squares and kernel canonical correlation analysis captured two hidden dimensions of brain-behavior relationships: one related to age and drinking and the other one related to depression. The applied machine learning framework indicates that these results are stable and generalize well to new data. Indeed, the identified brain-behavior associations are in agreement with previous findings in the literature concerning age, alcohol use, and depression-related changes in brain volume.

CONCLUSIONS

Multivariate techniques (such as sparse partial least squares and kernel canonical correlation analysis) embedded in our novel framework are promising tools to link behavior and/or symptoms to neurobiology and thus have great potential to contribute to a biologically grounded definition of psychiatric disorders.

Recommendations and future directions for supervised machine learning in psychiatry

Machine learning methods hold promise for personalized care in psychiatry, demonstrating the potential to tailor treatment decisions and stratify patients into clinically meaningful taxonomies. Subsequently, publication counts applying machine learning methods have risen, with different data modalities, mathematically distinct models, and samples of varying size being used to train and test models with the promise of clinical translation. Consequently, and in part due to the preliminary nature of such works, many studies have reported largely varying degrees of accuracy, raising concerns over systematic overestimation and methodological inconsistencies. Furthermore, a lack of procedural evaluation guidelines for non-expert medical professionals and funding bodies leaves many in the field with no means to systematically evaluate the claims, maturity, and clinical readiness of a project. Given the potential of machine learning methods to transform patient care, albeit, contingent on the rigor of employed methods and their dissemination, we deem it necessary to provide a review of current methods, recommendations, and future directions for applied machine learning in psychiatry. In this review we will cover issues of best practice for model training and evaluation, sources of systematic error and overestimation, model explainability vs. trust, the clinical implementation of AI systems, and finally, future directions for our field.

MultiLink Analysis: Brain Network Comparison via Sparse Connectivity Analysis

The analysis of the brain from a connectivity perspective is revealing novel insights into brain structure and function. Discovery is, however, hindered by the lack of prior knowledge used to make hypotheses. Additionally, exploratory data analysis is made complex by the high dimensionality of data. Indeed, to assess the effect of pathological states on brain networks, neuroscientists are often required to evaluate experimental effects in case-control studies, with hundreds of thousands of connections. In this paper, we propose an approach to identify the multivariate relationships in brain connections that characterize two distinct groups, hence permitting the investigators to immediately discover the subnetworks that contain information about the differences between experimental groups. In particular, we are interested in data discovery related to connectomics, where the connections that characterize differences between two groups of subjects are found. Nevertheless, those connections do not necessarily maximize the accuracy in classification since this does not guarantee reliable interpretation of specific differences between groups. In practice, our method exploits recent machine learning techniques employing sparsity to deal with weighted networks describing the whole-brain macro connectivity. We evaluated our technique on functional and structural connectomes from human and murine brain data. In our experiments, we automatically identified disease-relevant connections in datasets with supervised and unsupervised anatomy-driven parcellation approaches and by using high-dimensional datasets.

Chained regularization for identifying brain patterns specific to HIV infection

Human Immunodeficiency Virus (HIV) infection continues to have major adverse public health and clinical consequences despite the effectiveness of combination Antiretroviral Therapy (cART) in reducing HIV viral load and improving immune function. As successfully treated individuals with HIV infection age, their cognition declines faster than reported for normal aging. This phenomenon underlines the importance of improving long-term care, which requires a better understanding of the impact of HIV on the brain. In this paper, automated identification of patients and brain regions affected by HIV infection are modeled as a classification problem, whose solution is determined in two steps within our proposed Chained-Regularization framework. The first step focuses on selecting the HIV pattern (i.e., the most informative constellation of brain region measurements for distinguishing HIV infected subjects from healthy controls) by constraining the search for the optimal parameter setting of the classifier via group sparsity (ℓ2,1-norm). The second step improves classification accuracy by constraining the parameterization with respect to the selected measurements and the Euclidean regularization (ℓ2-norm). When applied to the cortical and subcortical structural Magnetic Resonance Images (MRI) measurements of 65 controls and 65 HIV infected individuals, this approach is more accurate in distinguishing the two cohorts than more common models. Finally, the brain regions of the identified HIV pattern concur with the HIV literature that uses traditional group analysis models.

Machine learning in major depression: From classification to treatment outcome prediction

AIMS

Major depression disorder (MDD) is the single greatest cause of disability and morbidity, and affects about 10% of the population worldwide. Currently, there are no clinically useful diagnostic biomarkers that are able to confirm a diagnosis of MDD from bipolar disorder (BD) in the early depressive episode. Therefore, exploring translational biomarkers of mood disorders based on machine learning is in pressing need, though it is challenging, but with great potential to improve our understanding of these disorders.

DISCUSSIONS

In this study, we review popular machine-learning methods used for brain imaging classification and predictions, and provide an overview of studies, specifically for MDD, that have used magnetic resonance imaging data to either (a) classify MDDs from controls or other mood disorders or (b) investigate treatment outcome predictors for individual patients. Finally, challenges, future directions, and potential limitations related to MDD biomarker identification are also discussed, with a goal of offering a comprehensive overview that may help readers to better understand the applications of neuroimaging data mining in depression.

CONCLUSIONS

We hope such efforts may highlight the need for an urgently needed paradigm shift in treatment, to guide personalized optimal clinical care.

MIDAS: Regionally linear multivariate discriminative statistical mapping

Statistical parametric maps formed via voxel-wise mass-univariate tests, such as the general linear model, are commonly used to test hypotheses about regionally specific effects in neuroimaging cross-sectional studies where each subject is represented by a single image. Despite being informative, these techniques remain limited as they ignore multivariate relationships in the data. Most importantly, the commonly employed local Gaussian smoothing, which is important for accounting for registration errors and making the data follow Gaussian distributions, is usually chosen in an ad hoc fashion. Thus, it is often suboptimal for the task of detecting group differences and correlations with non-imaging variables. Information mapping techniques, such as searchlight, which use pattern classifiers to exploit multivariate information and obtain more powerful statistical maps, have become increasingly popular in recent years. However, existing methods may lead to important interpretation errors in practice (i.e., misidentifying a cluster as informative, or failing to detect truly informative voxels), while often being computationally expensive. To address these issues, we introduce a novel efficient multivariate statistical framework for cross-sectional studies, termed MIDAS, seeking highly sensitive and specific voxel-wise brain maps, while leveraging the power of regional discriminant analysis. In MIDAS, locally linear discriminative learning is applied to estimate the pattern that best discriminates between two groups, or predicts a variable of interest. This pattern is equivalent to local filtering by an optimal kernel whose coefficients are the weights of the linear discriminant. By composing information from all neighborhoods that contain a given voxel, MIDAS produces a statistic that collectively reflects the contribution of the voxel to the regional classifiers as well as the discriminative power of the classifiers. Critically, MIDAS efficiently assesses the statistical significance of the derived statistic by analytically approximating its null distribution without the need for computationally expensive permutation tests. The proposed framework was extensively validated using simulated atrophy in structural magnetic resonance imaging (MRI) and further tested using data from a task-based functional MRI study as well as a structural MRI study of cognitive performance. The performance of the proposed framework was evaluated against standard voxel-wise general linear models and other information mapping methods. The experimental results showed that MIDAS achieves relatively higher sensitivity and specificity in detecting group differences. Together, our results demonstrate the potential of the proposed approach to efficiently map effects of interest in both structural and functional data.

Embedding Anatomical or Functional Knowledge in Whole-Brain Multiple Kernel Learning Models

Pattern recognition models have been increasingly applied to neuroimaging data over the last two decades. These applications have ranged from cognitive neuroscience to clinical problems. A common limitation of these approaches is that they do not incorporate previous knowledge about the brain structure and function into the models. Previous knowledge can be embedded into pattern recognition models by imposing a grouping structure based on anatomically or functionally defined brain regions. In this work, we present a novel approach that uses group sparsity to model the whole brain multivariate pattern as a combination of regional patterns. More specifically, we use a sparse version of Multiple Kernel Learning (MKL) to simultaneously learn the contribution of each brain region, previously defined by an atlas, to the decision function. Our application of MKL provides two beneficial features: (1) it can lead to improved overall generalisation performance when the grouping structure imposed by the atlas is consistent with the data; (2) it can identify a subset of relevant brain regions for the predictive model. In order to investigate the effect of the grouping in the proposed MKL approach we compared the results of three different atlases using three different datasets. The method has been implemented in the new version of the open-source Pattern Recognition for Neuroimaging Toolbox (PRoNTo).

Selecting the most relevant brain regions to discriminate Alzheimer's disease patients from healthy controls using multiple kernel learning: A comparison across functional and structural imaging modalities and atlases

BACKGROUND

Machine learning techniques such as support vector machine (SVM) have been applied recently in order to accurately classify individuals with neuropsychiatric disorders such as Alzheimer's disease (AD) based on neuroimaging data. However, the multivariate nature of the SVM approach often precludes the identification of the brain regions that contribute most to classification accuracy. Multiple kernel learning (MKL) is a sparse machine learning method that allows the identification of the most relevant sources for the classification. By parcelating the brain into regions of interest (ROI) it is possible to use each ROI as a source to MKL (ROI-MKL).

METHODS

We applied MKL to multimodal neuroimaging data in order to: 1) compare the diagnostic performance of ROI-MKL and whole-brain SVM in discriminating patients with AD from demographically matched healthy controls and 2) identify the most relevant brain regions to the classification. We used two atlases (AAL and Brodmann's) to parcelate the brain into ROIs and applied ROI-MKL to structural (T1) MRI, 18F-FDG-PET and regional cerebral blood flow SPECT (rCBF-SPECT) data acquired from the same subjects (20 patients with early AD and 18 controls). In ROI-MKL, each ROI received a weight (ROI-weight) that indicated the region's relevance to the classification. For each ROI, we also calculated whether there was a predominance of voxels indicating decreased or increased regional activity (for 18F-FDG-PET and rCBF-SPECT) or volume (for T1-MRI) in AD patients.

RESULTS

Compared to whole-brain SVM, the ROI-MKL approach resulted in better accuracies (with either atlas) for classification using 18F-FDG-PET (92.5% accuracy for ROI-MKL versus 84% for whole-brain), but not when using rCBF-SPECT or T1-MRI. Although several cortical and subcortical regions contributed to discrimination, high ROI-weights and predominance of hypometabolism and atrophy were identified specially in medial parietal and temporo-limbic cortical regions. Also, the weight of discrimination due to a pattern of increased voxel-weight values in AD individuals was surprisingly high (ranging from approximately 20% to 40% depending on the imaging modality), located mainly in primary sensorimotor and visual cortices and subcortical nuclei.

CONCLUSION

The MKL-ROI approach highlights the high discriminative weight of a subset of brain regions of known relevance to AD, the selection of which contributes to increased classification accuracy when applied to 18F-FDG-PET data. Moreover, the MKL-ROI approach demonstrates that brain regions typically spared in mild stages of AD also contribute substantially in the individual discrimination of AD patients from controls.

FReM - Scalable and stable decoding with fast regularized ensemble of models

Brain decoding relates behavior to brain activity through predictive models. These are also used to identify brain regions involved in the cognitive operations related to the observed behavior. Training such multivariate models is a high-dimensional statistical problem that calls for suitable priors. State of the art priors -eg small total-variation- enforce spatial structure on the maps to stabilize them and improve prediction. However, they come with a hefty computational cost. We build upon very fast dimension reduction with spatial structure and model ensembling to achieve decoders that are fast on large datasets and increase the stability of the predictions and the maps. Our approach, fast regularized ensemble of models (FReM), includes an implicit spatial regularization by using a voxel grouping with a fast clustering algorithm. In addition, it aggregates different estimators obtained across splits of a cross-validation loop, each time keeping the best possible model. Experiments on a large number of brain imaging datasets show that our combination of voxel clustering and model ensembling improves decoding maps stability and reduces the variance of prediction accuracy. Importantly, our method requires less samples than state-of-the-art methods to achieve a given level of prediction accuracy. Finally, FreM is much faster than other spatially-regularized methods and, in addition, it can better exploit parallel computing resources.

Support vector machine-based classification of neuroimages in Alzheimer's disease: direct comparison of FDG-PET, rCBF-SPECT and MRI data acquired from the same individuals

Objective:

To conduct the first support vector machine (SVM)-based study comparing the diagnostic accuracy of T1-weighted magnetic resonance imaging (T1-MRI), F-fluorodeoxyglucose-positron emission tomography (FDG-PET) and regional cerebral blood flow single-photon emission computed tomography (rCBF-SPECT) in Alzheimer’s disease (AD).

Method:

Brain T1-MRI, FDG-PET and rCBF-SPECT scans were acquired from a sample of mild AD patients (n=20) and healthy elderly controls (n=18). SVM-based diagnostic accuracy indices were calculated using whole-brain information and leave-one-out cross-validation.

Results:

The accuracy obtained using PET and SPECT data were similar. PET accuracy was 68∼71% and area under curve (AUC) 0.77∼0.81; SPECT accuracy was 68∼74% and AUC 0.75∼0.79, and both had better performance than analysis with T1-MRI data (accuracy of 58%, AUC 0.67). The addition of PET or SPECT to MRI produced higher accuracy indices (68∼74%; AUC: 0.74∼0.82) than T1-MRI alone, but these were not clearly superior to the isolated neurofunctional modalities.

Conclusion:

In line with previous evidence, FDG-PET and rCBF-SPECT more accurately identified patients with AD than T1-MRI, and the addition of either PET or SPECT to T1-MRI data yielded increased accuracy. The comparable SPECT and PET performances, directly demonstrated for the first time in the present study, support the view that rCBF-SPECT still has a role to play in AD diagnosis.

Euler Elastica Regularized Logistic Regression for Whole-Brain Decoding of fMRI Data

OBJECTIVE

Multivariate pattern analysis methods have been widely applied to functional magnetic resonance imaging (fMRI) data to decode brain states. Due to the "high features, low samples" in fMRI data, machine learning methods have been widely regularized using various regularizations to avoid overfitting. Both total variation (TV) using the gradients of images and Euler's elastica (EE) using the gradient and the curvature of images are the two popular regulations with spatial structures. In contrast to TV, EE regulation is able to overcome the disadvantage of TV regulation that favored piecewise constant images over piecewise smooth images. In this study, we introduced EE to fMRI-based decoding for the first time and proposed the EE regularized multinomial logistic regression (EELR) algorithm for multi-class classification.

METHODS

We performed experimental tests on both simulated and real fMRI data to investigate the feasibility and robustness of EELR. The performance of EELR was compared with sparse logistic regression (SLR) and TV regularized LR (TVLR).

RESULTS

The results showed that EELR was more robustness to noises and showed significantly higher classification performance than TVLR and SLR. Moreover, the forward models and weights patterns revealed that EELR detected larger brain regions that were discriminative to each task and activated by each task than TVLR.

CONCLUSION

The results suggest that EELR not only performs well in brain decoding but also reveals meaningful discriminative and activation patterns.

SIGNIFICANCE

This study demonstrated that EELR showed promising potential in brain decoding and discriminative/activation pattern detection.

Detecting Neuroimaging Biomarkers for Depression: A Meta-analysis of Multivariate Pattern Recognition Studies

BACKGROUND

Multiple studies have examined functional and structural brain alteration in patients diagnosed with major depressive disorder (MDD). The introduction of multivariate statistical methods allows investigators to utilize data concerning these brain alterations to generate diagnostic models that accurately differentiate patients with MDD from healthy control subjects (HCs). However, there is substantial heterogeneity in the reported results, the methodological approaches, and the clinical characteristics of participants in these studies.

METHODS

We conducted a meta-analysis of all studies using neuroimaging (volumetric measures derived from T1-weighted images, task-based functional magnetic resonance imaging [MRI], resting-state MRI, or diffusion tensor imaging) in combination with multivariate statistical methods to differentiate patients diagnosed with MDD from HCs.

RESULTS

Thirty-three (k = 33) samples including 912 patients with MDD and 894 HCs were included in the meta-analysis. Across all studies, patients with MDD were separated from HCs with 77% sensitivity and 78% specificity. Classification based on resting-state MRI (85% sensitivity, 83% specificity) and on diffusion tensor imaging data (88% sensitivity, 92% specificity) outperformed classifications based on structural MRI (70% sensitivity, 71% specificity) and task-based functional MRI (74% sensitivity, 77% specificity).

CONCLUSIONS

Our results demonstrate the high representational capacity of multivariate statistical methods to identify neuroimaging-based biomarkers of depression. Future studies are needed to elucidate whether multivariate neuroimaging analysis has the potential to generate clinically useful tools for the differential diagnosis of affective disorders and the prediction of both treatment response and functional outcome.

Kernel-based Joint Feature Selection and Max-Margin Classification for Early Diagnosis of Parkinson's Disease

Feature selection methods usually select the most compact and relevant set of features based on their contribution to a linear regression model. Thus, these features might not be the best for a non-linear classifier. This is especially crucial for the tasks, in which the performance is heavily dependent on the feature selection techniques, like the diagnosis of neurodegenerative diseases. Parkinson's disease (PD) is one of the most common neurodegenerative disorders, which progresses slowly while affects the quality of life dramatically. In this paper, we use the data acquired from multi-modal neuroimaging data to diagnose PD by investigating the brain regions, known to be affected at the early stages. We propose a joint kernel-based feature selection and classification framework. Unlike conventional feature selection techniques that select features based on their performance in the original input feature space, we select features that best benefit the classification scheme in the kernel space. We further propose kernel functions, specifically designed for our non-negative feature types. We use MRI and SPECT data of 538 subjects from the PPMI database, and obtain a diagnosis accuracy of 97.5%, which outperforms all baseline and state-of-the-art methods.

Exploring stability-based voxel selection methods in MVPA using cognitive neuroimaging data: a comprehensive study

Feature selection plays a key role in multi-voxel pattern analysis because functional magnetic resonance imaging data are typically noisy, sparse, and high-dimensional. Although the conventional evaluation criterion is the classification accuracy, selecting a stable feature set that is not sensitive to the variance in dataset may provide more scientific insights. In this study, we aim to investigate the stability of feature selection methods and test the stability-based feature selection scheme on two benchmark datasets. Top-k feature selection with a ranking score of mutual information and correlation, recursive feature elimination integrated with support vector machine, and L1 and L2-norm regularizations were adapted to a bootstrapped stability selection framework, and the selected algorithms were compared based on both accuracy and stability scores. The results indicate that regularization-based methods are generally more stable in StarPlus dataset, but in Haxby dataset they failed to perform as well as others.

A Bayesian probit model with spatially varying coefficients for brain decoding using fMRI data

Recent advances in human neuroimaging have shown that it is possible to accurately decode how the brain perceives information based only on non-invasive functional magnetic resonance imaging measurements of brain activity. Two commonly used statistical approaches, namely, univariate analysis and multivariate pattern analysis often lead to distinct patterns of selected voxels. One current debate in brain decoding concerns whether the brain's representation of sound categories is localized or distributed. We hypothesize that the distributed pattern of voxels selected by most multivariate pattern analysis models can be an artifact due to the spatial correlation among voxels. Here, we propose a Bayesian spatially varying coefficient model, where the spatial correlation is modeled through the variance-covariance matrix of the model coefficients. Combined with a proposed region selection strategy, we demonstrate that our approach is effective in identifying the truly localized patterns of the voxels while maintaining robustness to discover truly distributed pattern. In addition, we show that localized or clustered patterns can be artificially identified as distributed if without proper usage of the spatial correlation information in fMRI data. Copyright © 2016 John Wiley & Sons, Ltd.

Profiling persistent tubercule bacilli from patient sputa during therapy predicts early drug efficacy

BACKGROUND

New treatment options are needed to maintain and improve therapy for tuberculosis, which caused the death of 1.5 million people in 2013 despite potential for an 86 % treatment success rate. A greater understanding of Mycobacterium tuberculosis (M.tb) bacilli that persist through drug therapy will aid drug development programs. Predictive biomarkers for treatment efficacy are also a research priority.

METHODS AND RESULTS

Genome-wide transcriptional profiling was used to map the mRNA signatures of M.tb from the sputa of 15 patients before and 3, 7 and 14 days after the start of standard regimen drug treatment. The mRNA profiles of bacilli through the first 2 weeks of therapy reflected drug activity at 3 days with transcriptional signatures at days 7 and 14 consistent with reduced M.tb metabolic activity similar to the profile of pre-chemotherapy bacilli. These results suggest that a pre-existing drug-tolerant M.tb population dominates sputum before and after early drug treatment, and that the mRNA signature at day 3 marks the killing of a drug-sensitive sub-population of bacilli. Modelling patient indices of disease severity with bacterial gene expression patterns demonstrated that both microbiological and clinical parameters were reflected in the divergent M.tb responses and provided evidence that factors such as bacterial load and disease pathology influence the host-pathogen interplay and the phenotypic state of bacilli. Transcriptional signatures were also defined that predicted measures of early treatment success (rate of decline in bacterial load over 3 days, TB test positivity at 2 months, and bacterial load at 2 months).

CONCLUSIONS

This study defines the transcriptional signature of M.tb bacilli that have been expectorated in sputum after two weeks of drug therapy, characterizing the phenotypic state of bacilli that persist through treatment. We demonstrate that variability in clinical manifestations of disease are detectable in bacterial sputa signatures, and that the changing M.tb mRNA profiles 0-2 weeks into chemotherapy predict the efficacy of treatment 6 weeks later. These observations advocate assaying dynamic bacterial phenotypes through drug therapy as biomarkers for treatment success.

Studying depression using imaging and machine learning methods

Depression is a complex clinical entity that can pose challenges for clinicians regarding both accurate diagnosis and effective timely treatment. These challenges have prompted the development of multiple machine learning methods to help improve the management of this disease. These methods utilize anatomical and physiological data acquired from neuroimaging to create models that can identify depressed patients vs. non-depressed patients and predict treatment outcomes. This article (1) presents a background on depression, imaging, and machine learning methodologies; (2) reviews methodologies of past studies that have used imaging and machine learning to study depression; and (3) suggests directions for future depression-related studies.

•

Past studies successfully studied depression using machine learning and imaging.

•

Past studies have limitations in their methods.

•

Methods for future studies can be improved.

•

Future studies could yield more robust models to diagnosis and treat depression.

Discriminative Structured Feature Engineering for Macroscale Brain Connectomes

Neuroimaging techniques can measure structural and functional brain connectivity with unprecedented detail in vivo. This so-called brain connectome can be represented as high dimensional matrices corresponding to edge weights in graphs. After measuring the matrices of two cohorts (i.e., patients and healthy controls), one is often required to formulate computational network models for effective feature engineering to draw discriminative distinctions between the cohorts, as well as estimate the associated statistical significance. We designed a novel method to reveal the intrinsic features of functional matrices of discriminative power for group comparison. More specifically, by encouraging co-selection of edges connected to the same node, we preserved the discriminative edges to maximum extent. To reduce the false positive rate of the extracted discriminative edges, an optimization procedure was developed to evaluate the significance of these edges and remove trivial ones. We validated the proposed method using both synthetic data and real benchmarks, and compared it to ℓ1 regularized logistic regression, univariate t-test and stability selection. The experimental results clearly showed that the proposed approach outperformed the three competing methods under various settings. In addition to increasing the F-measure of feature selection, our approach captured the endogenous, discriminative connectivity patterns consistent with recent findings in biomedical literature. This data-driven method paves a new avenue of enquiry into the inherent nature of network models for functional brain connectomes.

Relevant feature set estimation with a knock-out strategy and random forests

Group analysis of neuroimaging data is a vital tool for identifying anatomical and functional variations related to diseases as well as normal biological processes. The analyses are often performed on a large number of highly correlated measurements using a relatively smaller number of samples. Despite the correlation structure, the most widely used approach is to analyze the data using univariate methods followed by post-hoc corrections that try to account for the data's multivariate nature. Although widely used, this approach may fail to recover from the adverse effects of the initial analysis when local effects are not strong. Multivariate pattern analysis (MVPA) is a powerful alternative to the univariate approach for identifying relevant variations. Jointly analyzing all the measures, MVPA techniques can detect global effects even when individual local effects are too weak to detect with univariate analysis. Current approaches are successful in identifying variations that yield highly predictive and compact models. However, they suffer from lessened sensitivity and instabilities in identification of relevant variations. Furthermore, current methods' user-defined parameters are often unintuitive and difficult to determine. In this article, we propose a novel MVPA method for group analysis of high-dimensional data that overcomes the drawbacks of the current techniques. Our approach explicitly aims to identify all relevant variations using a "knock-out" strategy and the Random Forest algorithm. In evaluations with synthetic datasets the proposed method achieved substantially higher sensitivity and accuracy than the state-of-the-art MVPA methods, and outperformed the univariate approach when the effect size is low. In experiments with real datasets the proposed method identified regions beyond the univariate approach, while other MVPA methods failed to replicate the univariate results. More importantly, in a reproducibility study with the well-known ADNI dataset the proposed method yielded higher stability and power than the univariate approach.

Randomized structural sparsity via constrained block subsampling for improved sensitivity of discriminative voxel identification

In this paper, we consider voxel selection for functional Magnetic Resonance Imaging (fMRI) brain data with the aim of finding a more complete set of probably correlated discriminative voxels, thus improving interpretation of the discovered potential biomarkers. The main difficulty in doing this is an extremely high dimensional voxel space and few training samples, resulting in unreliable feature selection. In order to deal with the difficulty, stability selection has received a great deal of attention lately, especially due to its finite sample control of false discoveries and transparent principle for choosing a proper amount of regularization. However, it fails to make explicit use of the correlation property or structural information of these discriminative features and leads to large false negative rates. In other words, many relevant but probably correlated discriminative voxels are missed. Thus, we propose a new variant on stability selection "randomized structural sparsity", which incorporates the idea of structural sparsity. Numerical experiments demonstrate that our method can be superior in controlling for false negatives while also keeping the control of false positives inherited from stability selection.

Feature Selection Based on the SVM Weight Vector for Classification of Dementia

Computer-aided diagnosis of dementia using a support vector machine (SVM) can be improved with feature selection. The relevance of individual features can be quantified from the SVM weights as a significance map (p-map). Although these p-maps previously showed clusters of relevant voxels in dementia-related brain regions, they have not yet been used for feature selection. Therefore, we introduce two novel feature selection methods based on p-maps using a direct approach (filter) and an iterative approach (wrapper). To evaluate these p-map feature selection methods, we compared them with methods based on the SVM weight vector directly, t-statistics, and expert knowledge. We used MRI data from the Alzheimer's disease neuroimaging initiative classifying Alzheimer's disease (AD) patients, mild cognitive impairment (MCI) patients who converted to AD (MCIc), MCI patients who did not convert to AD (MCInc), and cognitively normal controls (CN). Features for each voxel were derived from gray matter morphometry. Feature selection based on the SVM weights gave better results than t-statistics and expert knowledge. The p-map methods performed slightly better than those using the weight vector. The wrapper method scored better than the filter method. Recursive feature elimination based on the p-map improved most for AD-CN: the area under the receiver-operating-characteristic curve (AUC) significantly increased from 90.3% without feature selection to 92.0% when selecting 1.5%-3% of the features. This feature selection method also improved the other classifications: AD-MCI 0.1% improvement in AUC (not significant), MCI-CN 0.7%, and MCIc-MCInc 0.1% (not significant). Although the performance improvement due to feature selection was limited, the methods based on the p-map generally had the best performance, and were therefore better in estimating the relevance of individual features.

Clinical prediction from structural brain MRI scans: a large-scale empirical study

Multivariate pattern analysis (MVPA) methods have become an important tool in neuroimaging, revealing complex associations and yielding powerful prediction models. Despite methodological developments and novel application domains, there has been little effort to compile benchmark results that researchers can reference and compare against. This study takes a significant step in this direction. We employed three classes of state-of-the-art MVPA algorithms and common types of structural measurements from brain Magnetic Resonance Imaging (MRI) scans to predict an array of clinically relevant variables (diagnosis of Alzheimer's, schizophrenia, autism, and attention deficit and hyperactivity disorder; age, cerebrospinal fluid derived amyloid-β levels and mini-mental state exam score). We analyzed data from over 2,800 subjects, compiled from six publicly available datasets. The employed data and computational tools are freely distributed ( https://www.nmr.mgh.harvard.edu/lab/mripredict), making this the largest, most comprehensive, reproducible benchmark image-based prediction experiment to date in structural neuroimaging. Finally, we make several observations regarding the factors that influence prediction performance and point to future research directions. Unsurprisingly, our results suggest that the biological footprint (effect size) has a dramatic influence on prediction performance. Though the choice of image measurement and MVPA algorithm can impact the result, there was no universally optimal selection. Intriguingly, the choice of algorithm seemed to be less critical than the choice of measurement type. Finally, our results showed that cross-validation estimates of performance, while generally optimistic, correlate well with generalization accuracy on a new dataset.

Diagnostic classification of specific phobia subtypes using structural MRI data: a machine-learning approach

While neuroimaging research has advanced our knowledge about fear circuitry dysfunctions in anxiety disorders, findings based on diagnostic groups do not translate into diagnostic value for the individual patient. Machine-learning generates predictive information that can be used for single subject classification. We applied Gaussian process classifiers to a sample of patients with specific phobia as a model disorder for pathological forms of anxiety to test for classification based on structural MRI data. Gray (GM) and white matter (WM) volumetric data were analyzed in 33 snake phobics (SP; animal subtype), 26 dental phobics (DP; blood-injection-injury subtype) and 37 healthy controls (HC). Results showed good accuracy rates for GM and WM data in predicting phobia subtypes (GM: 62 % phobics vs. HC, 86 % DP vs. HC, 89 % SP vs. HC, 89 % DP vs. SP; WM: 88 % phobics vs. HC, 89 % DP vs. HC, 79 % SP vs. HC, 79 % DP vs. HC). Regarding GM, classification improved when considering the subtype compared to overall phobia status. The discriminatory brain pattern was not solely based on fear circuitry structures but included widespread cortico-subcortical networks. Results demonstrate that multivariate pattern recognition represents a promising approach for the development of neuroimaging-based diagnostic markers that could support clinical decisions. Regarding the increasing number of fMRI studies on anxiety disorders, researchers are encouraged to use functional and structural data not only for studying phenotype characteristics on a group level, but also to evaluate their incremental value for diagnostic or prognostic purposes.

Bayesian multi-task learning for decoding multi-subject neuroimaging data

Decoding models based on pattern recognition (PR) are becoming increasingly important tools for neuroimaging data analysis. In contrast to alternative (mass-univariate) encoding approaches that use hierarchical models to capture inter-subject variability, inter-subject differences are not typically handled efficiently in PR. In this work, we propose to overcome this problem by recasting the decoding problem in a multi-task learning (MTL) framework. In MTL, a single PR model is used to learn different but related “tasks” simultaneously. The primary advantage of MTL is that it makes more efficient use of the data available and leads to more accurate models by making use of the relationships between tasks. In this work, we construct MTL models where each subject is modelled by a separate task. We use a flexible covariance structure to model the relationships between tasks and induce coupling between them using Gaussian process priors. We present an MTL method for classification problems and demonstrate a novel mapping method suitable for PR models. We apply these MTL approaches to classifying many different contrasts in a publicly available fMRI dataset and show that the proposed MTL methods produce higher decoding accuracy and more consistent discriminative activity patterns than currently used techniques. Our results demonstrate that MTL provides a promising method for multi-subject decoding studies by focusing on the commonalities between a group of subjects rather than the idiosyncratic properties of different subjects.

•

In mass-univariate analysis, mixed effects models can capture subject variability.

•

In pattern recognition (PR), subject variability is usually not modelled explicitly.

•

Multi-task learning (MTL) is proposed to accommodate subject variability in PR.

•

The proposed approach improves predictive accuracy and pattern reproducibility.

•

A novel brain mapping approach is also proposed for MTL and existing PR models.