Annual research review: progress in using brain morphometry as a clinical tool for diagnosing psychiatric disorders

Machine learning based identification of structural brain alterations underlying suicide risk in adolescents

Suicide is the third leading cause of death for individuals between 15 and 19 years of age. The high suicide mortality rate and limited prior success in identifying neuroimaging biomarkers indicate that it is crucial to improve the accuracy of clinical neural signatures underlying suicide risk. The current study implements machine-learning (ML) algorithms to examine structural brain alterations in adolescents that can discriminate individuals with suicide risk from typically developing (TD) adolescents at the individual level. Structural MRI data were collected from 79 adolescents who demonstrated clinical levels of suicide risk and 79 demographically matched TD adolescents. Region-specific cortical/subcortical volume (CV/SCV) was evaluated following whole-brain parcellation into 1000 cortical and 12 subcortical regions. CV/SCV parameters were used as inputs for feature selection and three ML algorithms (i.e., support vector machine [SVM], K-nearest neighbors, and ensemble) to classify adolescents at suicide risk from TD adolescents. The highest classification accuracy of 74.79% (with sensitivity = 75.90%, specificity = 74.07%, and area under the receiver operating characteristic curve = 87.18%) was obtained for CV/SCV data using the SVM classifier. Identified bilateral regions that contributed to the classification mainly included reduced CV within the frontal and temporal cortices but increased volume within the cuneus/precuneus for adolescents at suicide risk relative to TD adolescents. The current data demonstrate an unbiased region-specific ML framework to effectively assess the structural biomarkers of suicide risk. Future studies with larger sample sizes and the inclusion of clinical controls and independent validation data sets are needed to confirm our findings.

Co-Occurrence of Tic Disorders and Attention-Deficit/Hyperactivity Disorder-Does It Reflect a Common Neurobiological Background?

BACKGROUND

The co-existence of tic disorders and attention-deficit/hyperactivity disorder (TD + ADHD) has proven to be highly important in daily clinical practice. The factor ADHD is not only associated with further comorbidities, but also has a long-term negative psychosocial effect, while the factor TD is usually less disturbing for the major part of the patients. It remains unclear how far this is related to a different neurobiological background of the associated disorders or whether TD + ADHD reflects a common one.

OBJECTIVE

This review provides an update on the neurobiological background of TD + ADHD in order to better understand and treat this clinical problem, while clarifying whether an additive model of TD + ADHD holds true and should be used as a basis for further clinical recommendations.

METHOD

A comprehensive research of the literature was conducted and analyzed, including existing clinical guidelines for both TD and ADHD. Besides genetical and environmental risk factors, brain structure and functions, neurophysiological processes and neurotransmitter systems were reviewed.

RESULTS

Only a limited number of empirical studies on the neurobiological background of TD and ADHD have taken the peculiarity of co-existing TD + ADHD into consideration, and even less studies have used a 2 × 2 factorial design in order to disentangle the impact/effects of the factors of TD versus those of ADHD. Nevertheless, the assumption that TD + ADHD can best be seen as an additive model at all levels of investigation was strengthened, although some overlap of more general, disorder non-specific aspects seem to exist.

CONCLUSION

Beyond stress-related transdiagnostic aspects, separate specific disturbances in certain neuronal circuits may lead to disorder-related symptoms inducing TD + ADHD in an additive way. Hence, within a classificatory categorical framework, the dimensional aspects of multilevel diagnostic-profiling seem to be a helpful precondition for personalized decisions on counselling and disorder-specific treatment in TD + ADHD.

Editorial: Biomarkers in precision medicine for mental illnesses

Precision medicine and biomarker development have become the prevailing paradigm for mental health research. Despite its conceptual elegance and dominance as a research framework, precision medicine has a very limited track record of demonstrable success thus far for mental illnesses, due in varying degrees to the complexity of both the brain and the pathophysiology of mental illnesses, which limits our ability to develop, replicate, and validate biomarkers for use in enhancing clinical care for mental illnesses, especially in high-risk and complex clinical populations. Research and funding priorities should integrate biomarker development and precision medicine interventions that target the robust behavioral, environmental, and social determinants that we know are important for population-based mental health.

Three dimensional convolutional neural network-based classification of conduct disorder with structural MRI

Conduct disorder (CD) is a common child and adolescent psychiatric disorder with various representative symptoms, and may cause long-term burden to patients and society. Recently, an increasing number of studies have used deep learning-based approaches, such as convolutional neural network (CNN), to analyze neuroimaging data and to identify biomarkers. In this study, we applied an optimized 3D AlexNet CNN model to automatically extract multi-layer high dimensional features of structural magnetic resonance imaging (sMRI), and to classify CD from healthy controls (HCs). We acquired high-resolution sMRI from 60 CD and 60 age- and gender-matched HCs. All subjects were male, and the age (mean ± std. dev) of participants in the CD and HC groups was 15.3 ± 1.0 and 15.5 ± 0.7, respectively. Five-fold cross validation (CV) was used to train and test this model. The receiver operating characteristic (ROC) curve for this model and that for support vector machine (SVM) model were compared. Feature visualization was performed to obtain intuition about the sMRI features learned by our AlexNet model. Our proposed AlexNet model achieved high classification performance with accuracy of 0.85, specificity of 0.82 and sensitivity of 0.87. The area under the ROC curve (AUC) of AlexNet was 0.86, significantly higher than that of SVM (AUC = 0.78; p = 0.046). The saliency maps for each convolutional layer highlighted the different brain regions in sMRI of CD, mainly including the frontal lobe, superior temporal gyrus, parietal lobe and occipital lobe. The classification results indicated that deep learning-based method is able to explore the hidden features from the sMRI of CD and might assist clinicians in the diagnosis of CD.

Distinguishing Adolescents With Conduct Disorder From Typically Developing Youngsters Based on Pattern Classification of Brain Structural MRI

Background: Conduct disorder (CD) is a mental disorder diagnosed in childhood or adolescence that presents antisocial behaviors, and is associated with structural alterations in brain. However, whether these structural alterations can distinguish CD from healthy controls (HCs) remains unknown. Here, we quantified these structural differences and explored the classification ability of these quantitative features based on machine learning (ML).

Materials and Methods: High-resolution 3D structural magnetic resonance imaging (sMRI) was acquired from 60 CD subjects and 60 age-matched HCs. Voxel-based morphometry (VBM) was used to assess the regional gray matter (GM) volume difference. The significantly different regional GM volumes were then extracted as features, and input into three ML classifiers: logistic regression, random forest and support vector machine (SVM). We trained and tested these ML models for classifying CD from HCs by using fivefold cross-validation (CV).

Results: Eight brain regions with abnormal GM volumes were detected, which mainly distributed in the frontal lobe, parietal lobe, anterior cingulate, cerebellum posterior lobe, lingual gyrus, and insula areas. We found that these ML models achieved comparable classification performance, with accuracy of 77.9 ∼ 80.4%, specificity of 73.3 ∼ 80.4%, sensitivity of 75.4 ∼ 87.5%, and area under the receiver operating characteristic curve (AUC) of 0.76 ∼ 0.80.

Conclusion: Based on sMRI and ML, the regional GM volumes may be used as potential imaging biomarkers for stable and accurate classification of CD.

Abnormal fronto-striatal activation as a marker of threshold and subthreshold Bulimia Nervosa

This study aimed to determine whether functional disturbances in fronto-striatal control circuits characterize adolescents with Bulimia Nervosa (BN) spectrum eating disorders regardless of clinical severity. FMRI was used to assess conflict-related brain activations during performance of a Simon task in two samples of adolescents with BN symptoms compared with healthy adolescents. The BN samples differed in the severity of their clinical presentation, illness duration and age. Multi-voxel pattern analyses (MVPAs) based on machine learning were used to determine whether patterns of fronto-striatal activation characterized adolescents with BN spectrum disorders regardless of clinical severity, and whether accurate classification of less symptomatic adolescents (subthreshold BN; SBN) could be achieved based on patterns of activation in adolescents who met DSM5 criteria for BN. MVPA classification analyses revealed that both BN and SBN adolescents could be accurately discriminated from healthy adolescents based on fronto-striatal activation. Notably, the patterns detected in more severely ill BN compared with healthy adolescents accurately discriminated less symptomatic SBN from healthy adolescents. Deficient activation of fronto-striatal circuits can characterize BN early in its course, when clinical presentations are less severe, perhaps pointing to circuit-based disturbances as useful biomarker or risk factor for the disorder, and a tool for understanding its developmental trajectory, as well as the development of early interventions.

A Comparison of Neuroimaging Abnormalities in Multiple Sclerosis, Major Depression and Chronic Fatigue Syndrome (Myalgic Encephalomyelitis): is There a Common Cause?

There is copious evidence of abnormalities in resting-state functional network connectivity states, grey and white matter pathology and impaired cerebral perfusion in patients afforded a diagnosis of multiple sclerosis, major depression or chronic fatigue syndrome (CFS) (myalgic encephalomyelitis). Systemic inflammation may well be a major element explaining such findings. Inter-patient and inter-illness variations in neuroimaging findings may arise at least in part from regional genetic, epigenetic and environmental variations in the functions of microglia and astrocytes. Regional differences in neuronal resistance to oxidative and inflammatory insults and in the performance of antioxidant defences in the central nervous system may also play a role. Importantly, replicated experimental findings suggest that the use of high-resolution SPECT imaging may have the capacity to differentiate patients afforded a diagnosis of CFS from those with a diagnosis of depression. Further research involving this form of neuroimaging appears warranted in an attempt to overcome the problem of aetiologically heterogeneous cohorts which probably explain conflicting findings produced by investigative teams active in this field. However, the ionising radiation and relative lack of sensitivity involved probably preclude its use as a routine diagnostic tool.

Reproducibility of Brain Morphometry from Short-Term Repeat Clinical MRI Examinations: A Retrospective Study

Purpose

To assess the inter session reproducibility of automatic segmented MRI-derived measures by FreeSurfer in a group of subjects with normal-appearing MR images.

Materials and Methods

After retrospectively reviewing a brain MRI database from our institute consisting of 14,758 adults, those subjects who had repeat scans and had no history of neurodegenerative disorders were selected for morphometry analysis using FreeSurfer. A total of 34 subjects were grouped by MRI scanner model. After automatic segmentation using FreeSurfer, label-wise comparison (involving area, thickness, and volume) was performed on all segmented results. An intraclass correlation coefficient was used to estimate the agreement between sessions. Wilcoxon signed rank test was used to assess the population mean rank differences across sessions. Mean-difference analysis was used to evaluate the difference intervals across scanners. Absolute percent difference was used to estimate the reproducibility errors across the MRI models. Kruskal-Wallis test was used to determine the across-scanner effect.

Results

The agreement in segmentation results for area, volume, and thickness measurements of all segmented anatomical labels was generally higher in Signa Excite and Verio models when compared with Sonata and TrioTim models. There were significant rank differences found across sessions in some labels of different measures. Smaller difference intervals in global volume measurements were noted on images acquired by Signa Excite and Verio models. For some brain regions, significant MRI model effects were observed on certain segmentation results.

Conclusions

Short-term scan-rescan reliability of automatic brain MRI morphometry is feasible in the clinical setting. However, since repeatability of software performance is contingent on the reproducibility of the scanner performance, the scanner performance must be calibrated before conducting such studies or before using such software for retrospective reviewing.

Structural MRI-Based Predictions in Patients with Treatment-Refractory Depression (TRD)

The application of machine learning techniques to psychiatric neuroimaging offers the possibility to identify robust, reliable and objective disease biomarkers both within and between contemporary syndromal diagnoses that could guide routine clinical practice. The use of quantitative methods to identify psychiatric biomarkers is consequently important, particularly with a view to making predictions relevant to individual patients, rather than at a group-level. Here, we describe predictions of treatment-refractory depression (TRD) diagnosis using structural T₁-weighted brain scans obtained from twenty adult participants with TRD and 21 never depressed controls. We report 85% accuracy of individual subject diagnostic prediction. Using an automated feature selection method, the major brain regions supporting this significant classification were in the caudate, insula, habenula and periventricular grey matter. It was not, however, possible to predict the degree of ‘treatment resistance’ in individual patients, at least as quantified by the Massachusetts General Hospital (MGH-S) clinical staging method; but the insula was again identified as a region of interest. Structural brain imaging data alone can be used to predict diagnostic status, but not MGH-S staging, with a high degree of accuracy in patients with TRD.

Editorial: Research Domain Criteria (RDoC): a new psychiatric nosology whose time has not yet come

In developing new ways of classifying mental disorders, RDoC is developing a new nosology, a new way of dividing nature at its seams. Given the NIMH influence on research agendas across the world, this scientific agenda will have important consequences for researchers and clinicians worldwide. Defining discrete neural systems and the behavioral and cognitive functions they subserve is scientifically important. Understanding how these systems relate to clinical problems, patient suffering, and improved treatments has immense potential practical value for clinical care worldwide. This Editorial places the RDoC framework in context and then sets out a series of conceptual, empirical, and developmental challenges for RDoC. Together these challenges suggest that RDoC is premature as a nosology and, as currently implemented, risks being reified and overly rigid in its application.

Editorial: The shape of the nosology to come in developmental psychopathology

The field has embarked on an effort to better integrate neurobiological and psychological dimensions of functioning with putative psychopathological syndromes. If successful, this effort aims to be a turning point as impactful as the change, a century ago, away from the study of symptom dimensions and toward the study of psychopathological syndromes. New statistical and neurobiological methods and findings hold considerable promise in this regard, and several papers in the present issue underscore these ongoing and important new directions. For this proposed direction to succeed, however, three guiding principles are necessary. First, the syndromal approach must continue to be viewed as provisional, and not reified. Second, in contrast, individual dimensions of neurobiology, psychology, personality, or symptoms should not be decontextualized but considered in relation to other traits and dimensions, syndromal configurations. Major clinical syndromes cannot be ignored. Third, following the Kraepelian insights of a century ago in addition to the more recent insights of developmental psychopathlogy, trait and dimension aspects of psychopathology need to be understood in their developmental context. Whether an integrated dimensional-categorical-developmental understanding of psychopathology can be extended to the entire nosology or only parts of it remains to be seen.

Annual research review: Current limitations and future directions in MRI studies of child- and adult-onset developmental psychopathologies

BACKGROUND

The widespread use of Magnetic Resonance Imaging (MRI) in the study of child- and adult-onset developmental psychopathologies has generated many investigations that have measured brain structure and function in vivo throughout development, often generating great excitement over our ability to visualize the living, developing brain using the attractive, even seductive images that these studies produce. Often lost in this excitement is the recognition that brain imaging generally, and MRI in particular, is simply a technology, one that does not fundamentally differ from any other technology, be it a blood test, a genotyping assay, a biochemical assay, or behavioral test. No technology alone can generate valid scientific findings. Rather, it is only technology coupled with a strong experimental design that can generate valid and reproducible findings that lead to new insights into the mechanisms of disease and therapeutic response.

METHODS

In this review we discuss selected studies to illustrate the most common and important limitations of MRI study designs as most commonly implemented thus far, as well as the misunderstanding that the interpretations of findings from those studies can create for our theories of developmental psychopathologies.

RESULTS

Common limitations of MRI study designs are in large part responsible thus far for the generally poor reproducibility of findings across studies, poor generalizability to the larger population, failure to identify developmental trajectories, inability to distinguish causes from effects of illness, and poor ability to infer causal mechanisms in most MRI studies of developmental psychopathologies. For each of these limitations in study design and the difficulties they entail for the interpretation of findings, we discuss various approaches that numerous laboratories are now taking to address those difficulties, which have in common the yoking of brain imaging technologies to studies with inherently stronger designs that permit more valid and more powerful causal inferences. Those study designs include epidemiological, longitudinal, high-risk, clinical trials, and multimodal imaging studies.

CONCLUSIONS

We highlight several studies that have yoked brain imaging technologies to these stronger designs to illustrate how doing so can aid our understanding of disease mechanisms and in the foreseeable future can improve clinical diagnosis, prevention, and treatment planning for developmental psychopathologies.

Congenital hypoplasia of the cerebellum: developmental causes and behavioral consequences

Over the last 60 years, the spotlight of research has periodically returned to the cerebellum as new techniques and insights have emerged. Because of its simple homogeneous structure, limited diversity of cell types and characteristic behavioral pathologies, the cerebellum is a natural home for studies of cell specification, patterning, and neuronal migration. However, recent evidence has extended the traditional range of perceived cerebellar function to include modulation of cognitive processes and implicated cerebellar hypoplasia and Purkinje neuron hypo-cellularity with autistic spectrum disorder. In the light of this emerging frontier, we review the key stages and genetic mechanisms behind cerebellum development. In particular, we discuss the role of the midbrain hindbrain isthmic organizer in the development of the cerebellar vermis and the specification and differentiation of Purkinje cells and granule neurons. These developmental processes are then considered in relation to recent insights into selected human developmental cerebellar defects: Joubert syndrome, Dandy–Walker malformation, and pontocerebellar hypoplasia. Finally, we review current research that opens up the possibility of using the mouse as a genetic model to study the role of the cerebellum in cognitive function.