Supervised machine learning for diagnostic classification from large-scale neuroimaging datasets

The application of machine learning techniques in posttraumatic stress disorder: a systematic review and meta-analysis

Posttraumatic stress disorder (PTSD) recently becomes one of the most important mental health concerns. However, no previous study has comprehensively reviewed the application of big data and machine learning (ML) techniques in PTSD. We found 873 studies meet the inclusion criteria and a total of 31 of those in a sample of 210,001 were included in quantitative analysis. ML algorithms were able to discriminate PTSD with an overall accuracy of 0.89. Pooled estimates of classification accuracy from multi-dimensional data (0.96) are higher than single data types (0.86 to 0.90). ML techniques can effectively classify PTSD and models using multi-dimensional data perform better than those using single data types. While selecting optimal combinations of data types and ML algorithms to be clinically applied at the individual level still remains a big challenge, these findings provide insights into the classification, identification, diagnosis and treatment of PTSD.

The Use of Generative Adversarial Network and Graph Convolution Network for Neuroimaging-Based Diagnostic Classification

Functional connectivity (FC) obtained from resting-state functional magnetic resonance imaging has been integrated with machine learning algorithms to deliver consistent and reliable brain disease classification outcomes. However, in classical learning procedures, custom-built specialized feature selection techniques are typically used to filter out uninformative features from FC patterns to generalize efficiently on the datasets. The ability of convolutional neural networks (CNN) and other deep learning models to extract informative features from data with grid structure (such as images) has led to the surge in popularity of these techniques. However, the designs of many existing CNN models still fail to exploit the relationships between entities of graph-structure data (such as networks). Therefore, graph convolution network (GCN) has been suggested as a means for uncovering the intricate structure of brain network data, which has the potential to substantially improve classification accuracy. Furthermore, overfitting in classifiers can be largely attributed to the limited number of available training samples. Recently, the generative adversarial network (GAN) has been widely used in the medical field for its generative aspect that can generate synthesis images to cope with the problems of data scarcity and patient privacy. In our previous work, GCN and GAN have been designed to investigate FC patterns to perform diagnosis tasks, and their effectiveness has been tested on the ABIDE-I dataset. In this paper, the models will be further applied to FC data derived from more public datasets (ADHD, ABIDE-II, and ADNI) and our in-house dataset (PTSD) to justify their generalization on all types of data. The results of a number of experiments show the powerful characteristic of GAN to mimic FC data to achieve high performance in disease prediction. When employing GAN for data augmentation, the diagnostic accuracy across ADHD-200, ABIDE-II, and ADNI datasets surpasses that of other machine learning models, including results achieved with BrainNetCNN. Specifically, in ADHD, the accuracy increased from 67.74% to 73.96% with GAN, in ABIDE-II from 70.36% to 77.40%, and in ADNI, reaching 52.84% and 88.56% for multiclass and binary classification, respectively. GCN also obtains decent results, with the best accuracy in ADHD datasets at 71.38% for multinomial and 75% for binary classification, respectively, and the second-best accuracy in the ABIDE-II dataset (72.28% and 75.16%, respectively). Both GAN and GCN achieved the highest accuracy for the PTSD dataset, reaching 97.76%. However, there are still some limitations that can be improved. Both methods have many opportunities for the prediction and diagnosis of diseases.

A review of the applications of generative adversarial networks to structural and functional MRI based diagnostic classification of brain disorders

Structural and functional MRI (magnetic resonance imaging) based diagnostic classification using machine learning has long held promise, but there are many roadblocks to achieving their potential. While traditional machine learning models suffered from their inability to capture the complex non-linear mapping, deep learning models tend to overfit the model. This is because there is data scarcity and imbalanced classes in neuroimaging; it is expensive to acquire data from human subjects and even more so in clinical populations. Due to their ability to augment data by learning underlying distributions, generative adversarial networks (GAN) provide a potential solution to this problem. Here, we provide a methodological primer on GANs and review the applications of GANs to classification of mental health disorders from neuroimaging data such as functional MRI and showcase the progress made thus far. We also highlight gaps in methodology as well as interpretability that are yet to be addressed. This provides directions about how the field can move forward. We suggest that since there are a range of methodological choices available to users, it is critical for users to interact with method developers so that the latter can tailor their development according to the users' needs. The field can be enriched by such synthesis between method developers and users in neuroimaging.

Neuroimaging-based classification of PTSD using data-driven computational approaches: A multisite big data study from the ENIGMA-PGC PTSD consortium

BACKGROUND

Recent advances in data-driven computational approaches have been helpful in devising tools to objectively diagnose psychiatric disorders. However, current machine learning studies limited to small homogeneous samples, different methodologies, and different imaging collection protocols, limit the ability to directly compare and generalize their results. Here we aimed to classify individuals with PTSD versus controls and assess the generalizability using a large heterogeneous brain datasets from the ENIGMA-PGC PTSD Working group.

METHODS

We analyzed brain MRI data from 3,477 structural-MRI; 2,495 resting state-fMRI; and 1,952 diffusion-MRI. First, we identified the brain features that best distinguish individuals with PTSD from controls using traditional machine learning methods. Second, we assessed the utility of the denoising variational autoencoder (DVAE) and evaluated its classification performance. Third, we assessed the generalizability and reproducibility of both models using leave-one-site-out cross-validation procedure for each modality.

RESULTS

We found lower performance in classifying PTSD vs. controls with data from over 20 sites (60 % test AUC for s-MRI, 59 % for rs-fMRI and 56 % for d-MRI), as compared to other studies run on single-site data. The performance increased when classifying PTSD from HC without trauma history in each modality (75 % AUC). The classification performance remained intact when applying the DVAE framework, which reduced the number of features. Finally, we found that the DVAE framework achieved better generalization to unseen datasets compared with the traditional machine learning frameworks, albeit performance was slightly above chance.

CONCLUSION

These results have the potential to provide a baseline classification performance for PTSD when using large scale neuroimaging datasets. Our findings show that the control group used can heavily affect classification performance. The DVAE framework provided better generalizability for the multi-site data. This may be more significant in clinical practice since the neuroimaging-based diagnostic DVAE classification models are much less site-specific, rendering them more generalizable.

Multimodal Autism Spectrum Disorder Diagnosis Method Based on DeepGCN

Multimodal data play an important role in the diagnosis of brain diseases. This study constructs a whole-brain functional connectivity network based on functional MRI data, uses non-imaging data with demographic information to complement the classification task for diagnosing subjects, and proposes a multimodal and across-site WL-DeepGCN-based method for classification to diagnose autism spectrum disorder (ASD). This method is used to resolve the existing problem that deep learning ASD identification cannot efficiently utilize multimodal data. In the WL-DeepGCN, a weight-learning network is used to represent the similarity of non-imaging data in the latent space, introducing a new approach for constructing population graph edge weights, and we find that it is beneficial and robust to define pairwise associations in the latent space rather than the input space. We propose a graph convolutional neural network residual connectivity approach to reduce the information loss due to convolution operations by introducing residual units to avoid gradient disappearance and gradient explosion. Furthermore, an EdgeDrop strategy makes the node connections sparser by randomly dropping edges in the raw graph, and its introduction can alleviate the overfitting and oversmoothing problems in the DeepGCN training process. We compare the WL-DeepGCN model with competitive models based on the same topics and nested 10-fold cross-validation show that our method achieves 77.27% accuracy and 0.83 AUC for ASD identification, bringing substantial performance gains.

Imaging Artificial Intelligence: A Framework for Radiologists to Address Health Equity, From the AJR Special Series on DEI

Artificial intelligence (AI) holds promise for helping patients access new and individualized health care pathways while increasing efficiencies for health care practitioners. Radiology has been at the forefront of this technology in medicine; many radiology practices are implementing and trialing AI-focused products. AI also holds great promise for reducing health disparities and promoting health equity. Radiology is ideally positioned to help reduce disparities given its central and critical role in patient care. The purposes of this article are to discuss the potential benefits and pitfalls of deploying AI algorithms in radiology, specifically highlighting the impact of AI on health equity; to explore ways to mitigate drivers of inequity; and to enhance pathways for creating better health care for all individuals, centering on a practical framework that helps radiologists address health equity during deployment of new tools.

Sampling inequalities affect generalization of neuroimaging-based diagnostic classifiers in psychiatry

BACKGROUND

The development of machine learning models for aiding in the diagnosis of mental disorder is recognized as a significant breakthrough in the field of psychiatry. However, clinical practice of such models remains a challenge, with poor generalizability being a major limitation.

METHODS

Here, we conducted a pre-registered meta-research assessment on neuroimaging-based models in the psychiatric literature, quantitatively examining global and regional sampling issues over recent decades, from a view that has been relatively underexplored. A total of 476 studies (n = 118,137) were included in the current assessment. Based on these findings, we built a comprehensive 5-star rating system to quantitatively evaluate the quality of existing machine learning models for psychiatric diagnoses.

RESULTS

A global sampling inequality in these models was revealed quantitatively (sampling Gini coefficient (G) = 0.81, p < .01), varying across different countries (regions) (e.g., China, G = 0.47; the USA, G = 0.58; Germany, G = 0.78; the UK, G = 0.87). Furthermore, the severity of this sampling inequality was significantly predicted by national economic levels (β =  - 2.75, p < .001, R²_adj = 0.40; r =  - .84, 95% CI: - .41 to - .97), and was plausibly predictable for model performance, with higher sampling inequality for reporting higher classification accuracy. Further analyses showed that lack of independent testing (84.24% of models, 95% CI: 81.0-87.5%), improper cross-validation (51.68% of models, 95% CI: 47.2-56.2%), and poor technical transparency (87.8% of models, 95% CI: 84.9-90.8%)/availability (80.88% of models, 95% CI: 77.3-84.4%) are prevailing in current diagnostic classifiers despite improvements over time. Relating to these observations, model performances were found decreased in studies with independent cross-country sampling validations (all p < .001, BF₁₀ > 15). In light of this, we proposed a purpose-built quantitative assessment checklist, which demonstrated that the overall ratings of these models increased by publication year but were negatively associated with model performance.

CONCLUSIONS

Together, improving sampling economic equality and hence the quality of machine learning models may be a crucial facet to plausibly translating neuroimaging-based diagnostic classifiers into clinical practice.

Error Consistency for Machine Learning Evaluation and Validation with Application to Biomedical Diagnostics

Supervised machine learning classification is the most common example of artificial intelligence (AI) in industry and in academic research. These technologies predict whether a series of measurements belong to one of multiple groups of examples on which the machine was previously trained. Prior to real-world deployment, all implementations need to be carefully evaluated with hold-out validation, where the algorithm is tested on different samples than it was provided for training, in order to ensure the generalizability and reliability of AI models. However, established methods for performing hold-out validation do not assess the consistency of the mistakes that the AI model makes during hold-out validation. Here, we show that in addition to standard methods, an enhanced technique for performing hold-out validation—that also assesses the consistency of the sample-wise mistakes made by the learning algorithm—can assist in the evaluation and design of reliable and predictable AI models. The technique can be applied to the validation of any supervised learning classification application, and we demonstrate the use of the technique on a variety of example biomedical diagnostic applications, which help illustrate the importance of producing reliable AI models. The validation software created is made publicly available, assisting anyone developing AI models for any supervised classification application in the creation of more reliable and predictable technologies.

Source space connectomics of neurodegeneration: One-metric approach does not fit all

Brain functional connectivity in dementia has been assessed with dissimilar EEG connectivity metrics and estimation procedures, thereby increasing results' heterogeneity. In this scenario, joint analyses integrating information from different metrics may allow for a more comprehensive characterization of brain functional interactions in different dementia subtypes. To test this hypothesis, resting-state electroencephalogram (rsEEG) was recorded in individuals with Alzheimer's Disease (AD), behavioral variant frontotemporal dementia (bvFTD), and healthy controls (HCs). Whole-brain functional connectivity was estimated in the EEG source space using 101 different types of functional connectivity, capturing linear and nonlinear interactions in both time and frequency-domains. Multivariate machine learning and progressive feature elimination was run to discriminate AD from HCs, and bvFTD from HCs, based on joint analyses of i) EEG frequency bands, ii) complementary frequency-domain metrics (e.g., instantaneous, lagged, and total connectivity), and iii) time-domain metrics with different linearity assumption (e.g., Pearson correlation coefficient and mutual information). <10% of all possible connections were responsible for the differences between patients and controls, and atypical connectivity was never captured by >1/4 of all possible connectivity measures. Joint analyses revealed patterns of hypoconnectivity (patientsHCs) in both groups was mainly identified in frontotemporal regions. These atypicalities were differently captured by frequency- and time-domain connectivity metrics, in a bandwidth-specific fashion. The multi-metric representation of source space whole-brain functional connectivity evidenced the inadequacy of single-metric approaches, and resulted in a valid alternative for the selection problem in EEG connectivity. These joint analyses reveal patterns of brain functional interdependence that are overlooked with single metrics approaches, contributing to a more reliable and interpretable description of atypical functional connectivity in neurodegeneration.

Generalizable neuromarker for autism spectrum disorder across imaging sites and developmental stages: A multi-site study

Autism spectrum disorder (ASD) is a lifelong condition, and its underlying biological mechanisms remain elusive. The complexity of various factors, including inter-site and development-related differences, makes it challenging to develop generalizable neuroimaging-based biomarkers for ASD. This study used a large-scale, multi-site dataset of 730 Japanese adults to develop a generalizable neuromarker for ASD across independent sites (U.S., Belgium, and Japan) and different developmental stages (children and adolescents). Our adult ASD neuromarker achieved successful generalization for the US and Belgium adults (area under the curve [AUC] = 0.70) and Japanese adults (AUC = 0.81). The neuromarker demonstrated significant generalization for children (AUC = 0.66) and adolescents (AUC = 0.71; all P < 0.05 , family-wise-error corrected). We identified 141 functional connections (FCs) important for discriminating individuals with ASD from TDCs. These FCs largely centered on social brain regions such as the amygdala, hippocampus, dorsomedial and ventromedial prefrontal cortices, and temporal cortices. Finally, we mapped schizophrenia (SCZ) and major depressive disorder (MDD) onto the biological axis defined by the neuromarker and explored the biological continuity of ASD with SCZ and MDD. We observed that SCZ, but not MDD, was located proximate to ASD on the biological dimension defined by the ASD neuromarker. The successful generalization in multifarious datasets and the observed relations of ASD with SCZ on the biological dimensions provide new insights for a deeper understanding of ASD.

Evaluation of Risk of Bias in Neuroimaging-Based Artificial Intelligence Models for Psychiatric Diagnosis: A Systematic Review

IMPORTANCE

Neuroimaging-based artificial intelligence (AI) diagnostic models have proliferated in psychiatry. However, their clinical applicability and reporting quality (ie, feasibility) for clinical practice have not been systematically evaluated.

OBJECTIVE

To systematically assess the risk of bias (ROB) and reporting quality of neuroimaging-based AI models for psychiatric diagnosis.

EVIDENCE REVIEW

PubMed was searched for peer-reviewed, full-length articles published between January 1, 1990, and March 16, 2022. Studies aimed at developing or validating neuroimaging-based AI models for clinical diagnosis of psychiatric disorders were included. Reference lists were further searched for suitable original studies. Data extraction followed the CHARMS (Checklist for Critical Appraisal and Data Extraction for Systematic Reviews of Prediction Modeling Studies) and PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) guidelines. A closed-loop cross-sequential design was used for quality control. The PROBAST (Prediction Model Risk of Bias Assessment Tool) and modified CLEAR (Checklist for Evaluation of Image-Based Artificial Intelligence Reports) benchmarks were used to systematically evaluate ROB and reporting quality.

FINDINGS

A total of 517 studies presenting 555 AI models were included and evaluated. Of these models, 461 (83.1%; 95% CI, 80.0%-86.2%) were rated as having a high overall ROB based on the PROBAST. The ROB was particular high in the analysis domain, including inadequate sample size (398 of 555 models [71.7%; 95% CI, 68.0%-75.6%]), poor model performance examination (with 100% of models lacking calibration examination), and lack of handling data complexity (550 of 555 models [99.1%; 95% CI, 98.3%-99.9%]). None of the AI models was perceived to be applicable to clinical practices. Overall reporting completeness (ie, number of reported items/number of total items) for the AI models was 61.2% (95% CI, 60.6%-61.8%), and the completeness was poorest for the technical assessment domain with 39.9% (95% CI, 38.8%-41.1%).

CONCLUSIONS AND RELEVANCE

This systematic review found that the clinical applicability and feasibility of neuroimaging-based AI models for psychiatric diagnosis were challenged by a high ROB and poor reporting quality. Particularly in the analysis domain, ROB in AI diagnostic models should be addressed before clinical application.

Multiple measurement analysis of resting-state fMRI for ADHD classification in adolescent brain from the ABCD study

Attention deficit hyperactivity disorder (ADHD) is one of the most common psychiatric disorders in school-aged children. Its accurate diagnosis looks after patients' interests well with effective treatment, which is important to them and their family. Resting-state functional magnetic resonance imaging (rsfMRI) has been widely used to characterize the abnormal brain function by computing the voxel-wise measures and Pearson's correlation (PC)-based functional connectivity (FC) for ADHD diagnosis. However, exploring the powerful measures of rsfMRI to improve ADHD diagnosis remains a particular challenge. To this end, this paper proposes an automated ADHD classification framework by fusion of multiple measures of rsfMRI in adolescent brain. First, we extract the voxel-wise measures and ROI-wise time series from the brain regions of rsfMRI after preprocessing. Then, to extract the multiple functional connectivities, we compute the PC-derived FCs including the topographical information-based high-order FC (tHOFC) and dynamics-based high-order FC (dHOFC), the sparse representation (SR)-derived FCs including the group SR (GSR), the strength and similarity guided GSR (SSGSR), and sparse low-rank (SLR). Finally, these measures are combined with multiple kernel learning (MKL) model for ADHD classification. The proposed method is applied to the Adolescent Brain and Cognitive Development (ABCD) dataset. The results show that the FCs of dHOFC and SLR perform better than the others. Fusing multiple measures achieves the best classification performance (AUC = 0.740, accuracy = 0.6916), superior to those from the single measure and the previous studies. We have identified the most discriminative FCs and brain regions for ADHD diagnosis, which are consistent with those of published literature.

Machine Learning and MRI-based Diagnostic Models for ADHD: Are We There Yet?

OBJECTIVE

Machine learning (ML) has been applied to develop magnetic resonance imaging (MRI)-based diagnostic classifiers for attention-deficit/hyperactivity disorder (ADHD). This systematic review examines this literature to clarify its clinical significance and to assess the implications of the various analytic methods applied.

METHODS

A comprehensive literature search on MRI-based diagnostic classifiers for ADHD was performed and data regarding the utilized models and samples were gathered.

RESULTS

We found that, although most studies reported the classification accuracies, they varied in choice of MRI modalities, ML models, cross-validation and testing methods, and sample sizes. We found that the accuracies of cross-validation methods inflated the performance estimation compared with those of a held-out test, compromising the model generalizability. Test accuracies have increased with publication year but were not associated with training sample sizes. Improved test accuracy over time was likely due to the use of better ML methods along with strategies to deal with data imbalances.

CONCLUSION

Ultimately, large multi-modal imaging datasets, and potentially the combination with other types of data, like cognitive data and/or genetics, will be essential to achieve the goal of developing clinically useful imaging classification tools for ADHD in the future.

Technologies to support the diagnosis and/or treatment of neurodevelopmental disorders: A systematic review

In recent years, there has been a great interest in utilizing technology in mental health research. The rapid technological development has encouraged researchers to apply technology as a part of a diagnostic process or treatment of Neurodevelopmental Disorders (NDDs). With the large number of studies being published comes an urgent need to inform clinicians and researchers about the latest advances in this field. Here, we methodically explore and summarize findings from studies published between August 2019 and February 2022. A search strategy led to the identification of 4108 records from PubMed and APA PsycInfo databases. 221 quantitative studies were included, covering a wide range of technologies used for diagnosis and/or treatment of NDDs, with the biggest focus on Autism Spectrum Disorder (ASD). The most popular technologies included machine learning, functional magnetic resonance imaging, electroencephalogram, magnetic resonance imaging, and neurofeedback. The results of the review indicate that technology-based diagnosis and intervention for NDD population is promising. However, given a high risk of bias of many studies, more high-quality research is needed.

Functional Connectome-Based Predictive Modeling in Autism

Autism is a heterogeneous neurodevelopmental condition, and functional magnetic resonance imaging-based studies have helped advance our understanding of its effects on brain network activity. We review how predictive modeling, using measures of functional connectivity and symptoms, has helped reveal key insights into this condition. We discuss how different prediction frameworks can further our understanding of the brain-based features that underlie complex autism symptomatology and consider how predictive models may be used in clinical settings. Throughout, we highlight aspects of study interpretation, such as data decay and sampling biases, that require consideration within the context of this condition. We close by suggesting exciting future directions for predictive modeling in autism.

Brain-phenotype models fail for individuals who defy sample stereotypes

Individual differences in brain functional organization track a range of traits, symptoms and behaviours1-12. So far, work modelling linear brain-phenotype relationships has assumed that a single such relationship generalizes across all individuals, but models do not work equally well in all participants13,14. A better understanding of in whom models fail and why is crucial to revealing robust, useful and unbiased brain-phenotype relationships. To this end, here we related brain activity to phenotype using predictive models-trained and tested on independent data to ensure generalizability15-and examined model failure. We applied this data-driven approach to a range of neurocognitive measures in a new, clinically and demographically heterogeneous dataset, with the results replicated in two independent, publicly available datasets16,17. Across all three datasets, we find that models reflect not unitary cognitive constructs, but rather neurocognitive scores intertwined with sociodemographic and clinical covariates; that is, models reflect stereotypical profiles, and fail when applied to individuals who defy them. Model failure is reliable, phenotype specific and generalizable across datasets. Together, these results highlight the pitfalls of a one-size-fits-all modelling approach and the effect of biased phenotypic measures18-20 on the interpretation and utility of resulting brain-phenotype models. We present a framework to address these issues so that such models may reveal the neural circuits that underlie specific phenotypes and ultimately identify individualized neural targets for clinical intervention.

Identifying neuroanatomical and behavioral features for autism spectrum disorder diagnosis in children using machine learning

Autism spectrum disorder (ASD) is a neurodevelopmental disorder that can cause significant social, communication, and behavioral challenges. Diagnosis of ASD is complicated and there is an urgent need to identify ASD-associated biomarkers and features to help automate diagnostics and develop predictive ASD models. The present study adopts a novel evolutionary algorithm, the conjunctive clause evolutionary algorithm (CCEA), to select features most significant for distinguishing individuals with and without ASD, and is able to accommodate datasets having a small number of samples with a large number of feature measurements. The dataset is unique and comprises both behavioral and neuroimaging measurements from a total of 28 children from 7 to 14 years old. Potential biomarker candidates identified include brain volume, area, cortical thickness, and mean curvature in specific regions around the cingulate cortex, frontal cortex, and temporal-parietal junction, as well as behavioral features associated with theory of mind. A separate machine learning classifier (i.e., k-nearest neighbors algorithm) was used to validate the CCEA feature selection and for ASD prediction. Study findings demonstrate how machine learning tools might help move the needle on improving diagnostic and predictive models of ASD.

Population heterogeneity in clinical cohorts affects the predictive accuracy of brain imaging

Brain imaging research enjoys increasing adoption of supervised machine learning for single-participant disease classification. Yet, the success of these algorithms likely depends on population diversity, including demographic differences and other factors that may be outside of primary scientific interest. Here, we capitalize on propensity scores as a composite confound index to quantify diversity due to major sources of population variation. We delineate the impact of population heterogeneity on the predictive accuracy and pattern stability in 2 separate clinical cohorts: the Autism Brain Imaging Data Exchange (ABIDE, n = 297) and the Healthy Brain Network (HBN, n = 551). Across various analysis scenarios, our results uncover the extent to which cross-validated prediction performances are interlocked with diversity. The instability of extracted brain patterns attributable to diversity is located preferentially in regions part of the default mode network. Collectively, our findings highlight the limitations of prevailing deconfounding practices in mitigating the full consequences of population diversity.

Machine Learning for Detecting Parkinson’s Disease by Resting-State Functional Magnetic Resonance Imaging: A Multicenter Radiomics Analysis

Parkinson’s disease (PD) is one of the most common progressive degenerative diseases, and its diagnosis is challenging on clinical grounds. Clinically, effective and quantifiable biomarkers to detect PD are urgently needed. In our study, we analyzed data from two centers, the primary set was used to train the model, and the independent external validation set was used to validate our model. We applied amplitude of low-frequency fluctuation (ALFF)-based radiomics method to extract radiomics features (including first- and high-order features). Subsequently, t-test and least absolute shrinkage and selection operator (LASSO) were harnessed for feature selection and data dimensionality reduction, and grid search method and nested 10-fold cross-validation were applied to determine the optimal hyper-parameter λ of LASSO and evaluate the performance of the model, in which a support vector machine was used to construct the classification model to classify patients with PD and healthy controls (HCs). We found that our model achieved good performance [accuracy = 81.45% and area under the curve (AUC) = 0.850] in the primary set and good generalization in the external validation set (accuracy = 67.44% and AUC = 0.667). Most of the discriminative features were high-order radiomics features, and the identified brain regions were mainly located in the sensorimotor network and lateral parietal cortex. Our study indicated that our proposed method can effectively classify patients with PD and HCs, ALFF-based radiomics features that might be potential biomarkers of PD, and provided further support for the pathological mechanism of PD, that is, PD may be related to abnormal brain activity in the sensorimotor network and lateral parietal cortex.

The Road to Personalized Medicine in Alzheimer’s Disease: The Use of Artificial Intelligence

Dementia remains an extremely prevalent syndrome among older people and represents a major cause of disability and dependency. Alzheimer’s disease (AD) accounts for the majority of dementia cases and stands as the most common neurodegenerative disease. Since age is the major risk factor for AD, the increase in lifespan not only represents a rise in the prevalence but also adds complexity to the diagnosis. Moreover, the lack of disease-modifying therapies highlights another constraint. A shift from a curative to a preventive approach is imminent and we are moving towards the application of personalized medicine where we can shape the best clinical intervention for an individual patient at a given point. This new step in medicine requires the most recent tools and analysis of enormous amounts of data where the application of artificial intelligence (AI) plays a critical role on the depiction of disease–patient dynamics, crucial in reaching early/optimal diagnosis, monitoring and intervention. Predictive models and algorithms are the key elements in this innovative field. In this review, we present an overview of relevant topics regarding the application of AI in AD, detailing the algorithms and their applications in the fields of drug discovery, and biomarkers.

ADHD classification using auto-encoding neural network and binary hypothesis testing

Attention Deficit Hyperactivity Disorder (ADHD) is a highly prevalent neurodevelopmental disease of school-age children. Early diagnosis is crucial for ADHD treatment, wherein its neurobiological diagnosis (or classification) is helpful and provides the objective evidence to clinicians. The existing ADHD classification methods suffer two problems, i.e., insufficient data and feature noise disturbance from other associated disorders. As an attempt to overcome these difficulties, a novel deep-learning classification architecture based on a binary hypothesis testing framework and a modified auto-encoding (AE) network is proposed in this paper. The binary hypothesis testing framework is introduced to cope with insufficient data of ADHD database. Brain functional connectivities (FCs) of test data (without seeing their labels) are incorporated during feature selection along with those of training data and affect the sequential deep learning procedure under binary hypotheses. On the other hand, the modified AE network is developed to capture more effective features from training data, such that the difference of inter- and intra-class variability scores between binary hypotheses can be enlarged and effectively alleviate the disturbance of feature noise. On the test of ADHD-200 database, our method significantly outperforms the existing classification methods. The average accuracy reaches 99.6% with the leave-one-out cross validation. Our method is also more robust and practically convenient for ADHD classification due to its uniform parameter setting across various datasets.

Functional Connectivity-Based Prediction of Autism on Site Harmonized ABIDE Dataset

OBJECTIVE

The larger sample sizes available from multi-site publicly available neuroimaging data repositories makes machine-learning based diagnostic classification of mental disorders more feasible by alleviating the curse of dimensionality. However, since multi-site data are aggregated post-hoc, i.e. they were acquired from different scanners with different acquisition parameters, non-neural inter-site variability may mask inter-group differences that are at least in part neural in origin. Hence, the advantages gained by the larger sample size in the context of machine-learning based diagnostic classification may not be realized.

METHODS

We address this issue using harmonization of multi-site neuroimaging data using the ComBat technique, which is based on an empirical Bayes formulation to remove inter-site differences in data distributions, to improve diagnostic classification accuracy. Specifically, we demonstrate this using ABIDE (Autism Brain Imaging Data Exchange) multi-site data for classifying individuals with Autism from healthy controls using resting state fMRI-based functional connectivity data.

RESULTS

Our results show that higher classification accuracies across multiple classification models can be obtained (especially for models based on artificial neural networks) from multi-site data post harmonization with the ComBat technique as compared to without harmonization, outperforming earlier results from existing studies using ABIDE. Furthermore, our network ablation analysis facilitated important insights into autism spectrum disorder pathology and the connectivity in networks shown to be important for classification covaried with verbal communication impairments in Autism.

CONCLUSION

Multi-site data harmonization using ComBat improves neuroimaging-based diagnostic classification of mental disorders.

SIGNIFICANCE

ComBat has the potential to make AI-based clinical decision-support systems more feasible in psychiatry.

A fundamental study assessing the generalized fitting method in conjunction with every possible coalition of N-combinations (G-EPOC) using the appendicitis detection task of computed tomography

Purpose

Increased use of deep learning (DL) in medical imaging diagnoses has led to more frequent use of 10-fold cross-validation (10-CV) for the evaluation of the performance of DL. To eliminate some of the (10-fold) repetitive processing in 10-CV, we proposed a “generalized fitting method in conjunction with every possible coalition of N-combinations (G-EPOC)”, to estimate the range of the mean accuracy of 10-CV using less than 10 results of 10-CV.

Material and methods

G-EPOC was executed as follows. We first provided (2N-1) coalition subsets using a specified N, which was 9 or less, out of 10 result datasets of 10-CV. We then obtained the estimation range of the accuracy by applying those subsets to the distribution fitting twice using a combination of normal, binominal, or Poisson distributions. Using datasets of 10-CVs acquired from the practical detection task of the appendicitis on CT by DL, we scored the estimation success rates if the range provided by G-EPOC included the true accuracy.

Results

G-EPOC successfully estimated the range of the mean accuracy by 10-CV at over 95% rates for datasets with N assigned as 2 to 9.

Conclusions

G-EPOC will help lessen the consumption of time and computer resources in the development of computerbased diagnoses in medical imaging and could become an option for the selection of a reasonable K value in K-CV.

Trends, Limits, and Challenges of Computer Technologies in Attention Deficit Hyperactivity Disorder Diagnosis and Treatment

Attention deficit hyperactivity disorder (ADHD) is a neurobiological condition that appears during an individual's childhood and may follow her/him for life. The research objective was to understand better how and which computer technologies have been applied to support ADHD diagnosis and treatment. The research used the systematic literature review method: a rigorous, verifiable, and repeatable approach that follows well-defined steps. Six well-known academic data sources have been consulted, including search engines and bibliographic databases, from technology and health care areas. After a rigorous research protocol, 1,239 articles were analyzed. For the diagnosis, the use of machine learning techniques was verified in 61 percent of the articles. Neurofeedback was ranked second with 9.3 percent participation, followed by serious games and eye tracking with 5.6 percent each. For the treatment, neurofeedback was present in 50 percent of the articles, whereas some studies combined both approaches, accounting for 31 percent of the total. Nine percent of the articles reported remote assistance technology, whereas another 9 percent have used virtual reality. By highlighting the leading computer technologies used, their applications, results, and challenges, this literature review breaks ground for further investigations. Moreover, the study highlighted the lack of consensus on ADHD biomarkers. The approaches using machine learning call attention to the probable occurrence of overfitting in several studies, thus demonstrating limitations of this technology on small-sized bases. This research also presented the convergence of evidence from different studies on the persistence of long-term effects of using neurofeedback in treating ADHD.

Machine learning with neuroimaging data to identify autism spectrum disorder: a systematic review and meta-analysis

PURPOSE

Autism Spectrum Disorder (ASD) is diagnosed through observation or interview assessments, which is time-consuming, subjective, and with questionable validity and reliability. Thus, we aimed to evaluate the role of machine learning (ML) with neuroimaging data to provide a reliable classification of ASD.

METHODS

A systematic search of PubMed, Scopus, and Embase was conducted to identify relevant publications. Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) was used to assess the studies' quality. A bivariate random-effects model meta-analysis was employed to evaluate the pooled sensitivity, the pooled specificity, and the diagnostic performance through the hierarchical summary receiver operating characteristic (HSROC) curve of ML with neuroimaging data in classifying ASD. Meta-regression was also performed.

RESULTS

Forty-four studies (5697 ASD and 6013 typically developing individuals [TD] in total) were included in the quantitative analysis. The pooled sensitivity for differentiating ASD from TD individuals was 86.25 95% confidence interval [CI] (81.24, 90.08), while the pooled specificity was 83.31 95% CI (78.12, 87.48) with a combined area under the HSROC (AUC) of 0.889. Higgins I² (> 90%) and Cochran's Q (p < 0.0001) suggest a high degree of heterogeneity. In the bivariate model meta-regression, a higher pooled specificity was observed in studies not using a brain atlas (90.91 95% CI [80.67, 96.00], p = 0.032). In addition, a greater pooled sensitivity was seen in studies recruiting both males and females (89.04 95% CI [83.84, 92.72], p = 0.021), and combining imaging modalities (94.12 95% [85.43, 97.76], p = 0.036).

CONCLUSION

ML with neuroimaging data is an exciting prospect in detecting individuals with ASD but further studies are required to improve its reliability for usage in clinical practice.

Using Clustering to Examine Inter-individual Variability in Topography of Auditory Event-Related Potentials in Autism and Typical Development

Although prior studies have compared sensory event-related potential (ERP) responses between groups of autistic and typically-developing participants, it is unclear how heterogeneity contributes to the results of these studies. The present study used examined individual differences in these responses. 130 autistic children and 81 typically-developing children, aged between 2 and 5 years, listened to tones at four identity levels while 61-channel electroencephalography was recorded. Hierarchical clustering was used to group participants based on rescaled ERP topographies between 51 and 350 ms. The hierarchical clustering analysis revealed substantial heterogeneity. Some of the seven clusters defined in this analysis were characterized by prolonged fronto-central positivities and/or weak or absent N2 negativities. However, many other participants fell into clusters in which N2 responses were present at varying latencies. Atypical response morphologies such as absent N2 responses and/or prolonged positive-going responses found in some autistic participants may account for prior research findings of attenuated N2 amplitudes in autism. However, there was also considerable overlap between groups, with participants of both groups appearing in all clusters. These results emphasize the utility of using clustering to explore individual differences in brain responses, which can expand on and clarify the results of analyses of group mean differences.

FMRI hemodynamic response function (HRF) as a novel marker of brain function: applications for understanding obsessive-compulsive disorder pathology and treatment response

The hemodynamic response function (HRF) represents the transfer function linking neural activity with the functional MRI (fMRI) signal, modeling neurovascular coupling. Since HRF is influenced by non-neural factors, to date it has largely been considered as a confound or has been ignored in many analyses. However, underlying biophysics suggests that the HRF may contain meaningful correlates of neural activity, which might be unavailable through conventional fMRI metrics. Here, we estimated the HRF by performing deconvolution on resting-state fMRI data from a longitudinal sample of 25 healthy controls scanned twice and 44 adults with obsessive-compulsive disorder (OCD) before and after 4-weeks of intensive cognitive-behavioral therapy (CBT). HRF response height, time-to-peak and full-width at half-maximum (FWHM) in OCD were abnormal before treatment and normalized after treatment in regions including the caudate. Pre-treatment HRF predicted treatment outcome (OCD symptom reduction) with 86.4% accuracy, using machine learning. Pre-treatment HRF response height in the caudate head and time-to-peak in the caudate tail were top-predictors of treatment response. Time-to-peak in the caudate tail, a region not typically identified in OCD studies using conventional fMRI activation or connectivity measures, may carry novel importance. Additionally, pre-treatment response height in caudate head predicted post-treatment OCD severity (R = -0.48, P = 0.001), and was associated with treatment-related OCD severity changes (R = -0.44, P = 0.0028), underscoring its relevance. With HRF being a reliable marker sensitive to brain function, OCD pathology, and intervention-related changes, these results could guide future studies towards novel discoveries not possible through conventional fMRI approaches like standard BOLD activation or connectivity.

Multimodal neurocognitive markers of frontal lobe epilepsy: Insights from ecological text processing

The pressing call to detect sensitive cognitive markers of frontal lobe epilepsy (FLE) remains poorly addressed. Standard frameworks prove nosologically unspecific (as they reveal deficits that also emerge across other epilepsy subtypes), possess low ecological validity, and are rarely supported by multimodal neuroimaging assessments. To bridge these gaps, we examined naturalistic action and non-action text comprehension, combined with structural and functional connectivity measures, in 19 FLE patients, 19 healthy controls, and 20 posterior cortex epilepsy (PCE) patients. Our analyses integrated inferential statistics and data-driven machine-learning classifiers. FLE patients were selectively and specifically impaired in action comprehension, irrespective of their neuropsychological profile. These deficits selectively and specifically correlated with (a) reduced integrity of the anterior thalamic radiation, a subcortical structure underlying motoric and action-language processing as well as epileptic seizure spread in this subtype; and (b) hypoconnectivity between the primary motor cortex and the left-parietal/supramarginal regions, two putative substrates of action-language comprehension. Moreover, machine-learning classifiers based on the above neurocognitive measures yielded 75% accuracy rates in discriminating individual FLE patients from both controls and PCE patients. Briefly, action-text assessments, combined with structural and functional connectivity measures, seem to capture ecological cognitive deficits that are specific to FLE, opening new avenues for discriminatory characterizations among epilepsy types.

Brain imaging-based machine learning in autism spectrum disorder: methods and applications

Autism spectrum disorder (ASD) is a neurodevelopmental condition with early childhood onset and high heterogeneity. As the pathogenesis is still elusive, ASD diagnosis is comprised of a constellation of behavioral symptoms. Non-invasive brain imaging techniques, such as magnetic resonance imaging (MRI), provide a valuable objective measurement of the brain. Many efforts have been devoted to developing imaging-based diagnostic tools for ASD based on machine learning (ML) technologies. In this survey, we review recent advances that utilize machine learning approaches to classify individuals with and without ASD. First, we provide a brief overview of neuroimaging-based ASD classification studies, including the analysis of publications and general classification pipeline. Next, representative studies are highlighted and discussed in detail regarding different imaging modalities, methods and sample sizes. Finally, we highlight several common challenges and provide recommendations on future directions. In summary, identifying discriminative biomarkers for ASD diagnosis is challenging, and further establishing more comprehensive datasets and dissecting the individual and group heterogeneity will be critical to achieve better ADS diagnosis performance. Machine learning methods will continue to be developed and are poised to help advance the field in this regard.

Localization and Diagnosis of Attention-Deficit/Hyperactivity Disorder

In this paper, a random-forest-based method was proposed for the classification and localization of Attention-Deficit/Hyperactivity Disorder (ADHD), a common neurodevelopmental disorder among children. Experimental data were magnetic resonance imaging (MRI) from the public case-control dataset of 3D images for ADHD-200. Each MRI image was a 3D-tensor of 121×145×121 size. All 3D matrices (MRI) were segmented into the slices from each of three orthogonal directions. Each slice from the same position of the same direction in the training set was converted into a vector, and all these vectors were composed into a designed matrix to train the random forest classification algorithm; then, the well-trained RF classifier was exploited to give a prediction label in correspondence direction and position. Diagnosis and location results can be obtained upon the intersection of these three prediction matrices. The performance of our proposed method was illustrated on the dataset from New York University (NYU), Kennedy Krieger Institute (KKI) and full datasets; the results show that the proposed methods can archive more accuracy identification in discrimination of ADHD, and can be extended to the other practices of diagnosis. Moreover, another suspected region was found at the first time.

Optimising a Simple Fully Convolutional Network for Accurate Brain Age Prediction in the PAC 2019 Challenge

Brain age prediction from brain MRI scans not only helps improve brain ageing modelling generally, but also provides benchmarks for predictive analysis methods. Brain-age delta, which is the difference between a subject's predicted age and true age, has become a meaningful biomarker for the health of the brain. Here, we report the details of our brain age prediction models and results in the Predictive Analysis Challenge 2019. The aim of the challenge was to use T1-weighted brain MRIs to predict a subject's age in multicentre datasets. We apply a lightweight deep convolutional neural network architecture, Simple Fully Convolutional Neural Network (SFCN), and combined several techniques including data augmentation, transfer learning, model ensemble, and bias correction for brain age prediction. The model achieved first place in both of the two objectives in the PAC 2019 brain age prediction challenge: Mean absolute error (MAE) = 2.90 years without bias removal (Second Place = 3.09 yrs; Third Place = 3.33 yrs), and MAE = 2.95 years with bias removal, leading by a large margin (Second Place = 3.80 yrs; Third Place = 3.92 yrs).

Neuroimaging-Based Deep Learning in Autism Spectrum Disorder and Attention-Deficit/Hyperactivity Disorder

Deep learning (DL) is a kind of machine learning technique that uses artificial intelligence to identify the characteristics of given data and efficiently analyze large amounts of information to perform tasks such as classification and prediction. In the field of neuroimaging of neurodevelopmental disorders, various biomarkers for diagnosis, classification, prognosis prediction, and treatment response prediction have been examined; however, they have not been efficiently combined to produce meaningful results. DL can be applied to overcome these limitations and produce clinically helpful results. Here, we review studies that combine neurodevelopmental disorder neuroimaging and DL techniques to explore the strengths, limitations, and future directions of this research area.

MALINI (Machine Learning in NeuroImaging): A MATLAB toolbox for aiding clinical diagnostics using resting-state fMRI data

Resting-state functional Magnetic Resonance Imaging (rs-fMRI) has been extensively used for diagnostic classification because it does not require task compliance and is easier to pool data from multiple imaging sites, thereby increasing the sample size. A MATLAB-based toolbox called Machine Learning in NeuroImaging (MALINI) for feature extraction and disease classification is presented. The MALINI toolbox extracts functional and effective connectivity features from preprocessed rs-fMRI data and performs classification between healthy and disease groups using any of 18 popular and widely used machine learning algorithms that are based on diverse principles. A consensus classifier combining the power of multiple classifiers is also presented. The utility of the toolbox is illustrated by accompanying data consisting of resting-state functional connectivity features from healthy controls and subjects with various brain-based disorders: autism spectrum disorder from autism brain imaging data exchange (ABIDE), Alzheimer's disease and mild cognitive impairment from Alzheimer's disease neuroimaging initiative (ADNI), attention deficit hyperactivity disorder from ADHD-200, and post-traumatic stress disorder and post-concussion syndrome acquired in-house. Results of classification performed on the above datasets can be obtained from the main article titled “Supervised machine learning for diagnostic classification from large-scale neuroimaging datasets” [1]. The data was divided into homogeneous and heterogeneous splits, such that 80% could be used for training, model building and cross-validation, while the remaining 20% of the data could be used as a hold-out independent test data for replication of the classification performance, to ensure the robustness of the classifiers to population variance in image acquisition site and age of the sample.

Deterioration from healthy to mild cognitive impairment and Alzheimer's disease mirrored in corresponding loss of centrality in directed brain networks

OBJECTIVE

It is important to identify brain-based biomarkers that progressively deteriorate from healthy to mild cognitive impairment (MCI) to Alzheimer's disease (AD). Cortical thickness, amyloid-ß deposition, and graph measures derived from functional connectivity (FC) networks obtained using functional MRI (fMRI) have been previously identified as potential biomarkers. Specifically, in the latter case, betweenness centrality (BC), a nodal graph measure quantifying information flow, is reduced in both AD and MCI. However, all such reports have utilized BC calculated from undirected networks that characterize synchronization rather than information flow, which is better characterized using directed networks.

METHODS

Therefore, we estimated BC from directed networks using Granger causality (GC) on resting-state fMRI data (N = 132) to compare the following populations (p < 0.05, FDR corrected for multiple comparisons): normal control (NC), early MCI (EMCI), late MCI (LMCI) and AD. We used an additional metric called middleman power (MP), which not only characterizes nodal information flow as in BC, but also measures nodal power critical for information flow in the entire network.

RESULTS

MP detected more brain regions than BC that progressively deteriorated from NC to EMCI to LMCI to AD, as well as exhibited significant associations with behavioral measures. Additionally, graph measures obtained from conventional FC networks could not identify a single node, underscoring the relevance of GC.

CONCLUSION

Our findings demonstrate the superiority of MP over BC as well as GC over FC in our case. MP obtained from GC networks could serve as a potential biomarker for progressive deterioration of MCI and AD.