Review Article| Volume 30, ISSUE 1, P1-13, February 2020

Download started.

Ok

Clinical Strategies and Technical Challenges in Psychoradiology

  • Author Footnotes
    1 F. Li and D. Wu contributed equally to this work.
    Fei Li
    Footnotes
    1 F. Li and D. Wu contributed equally to this work.
    Affiliations
    Huaxi MR Research Center (HMRRC), Department of Radiology, West China Hospital of Sichuan University, No. 37 Guo Xue Lane, Chengdu 610041, China

    Psychoradiology Research Unit of Chinese Academy of Medical Sciences, West China Hospital of Sichuan University, No. 37 Guo Xue Lane, Chengdu 610041, China
    Search for articles by this author
  • Author Footnotes
    1 F. Li and D. Wu contributed equally to this work.
    Dongsheng Wu
    Footnotes
    1 F. Li and D. Wu contributed equally to this work.
    Affiliations
    Huaxi MR Research Center (HMRRC), Department of Radiology, West China Hospital of Sichuan University, No. 37 Guo Xue Lane, Chengdu 610041, China

    Psychoradiology Research Unit of Chinese Academy of Medical Sciences, West China Hospital of Sichuan University, No. 37 Guo Xue Lane, Chengdu 610041, China
    Search for articles by this author
  • Su Lui
    Correspondence
    Corresponding author. Huaxi MR Research Center (HMRRC), Department of Radiology, West China Hospital of Sichuan University, No. 37 Guo Xue Lane, Chengdu 610041, China.
    Affiliations
    Huaxi MR Research Center (HMRRC), Department of Radiology, West China Hospital of Sichuan University, No. 37 Guo Xue Lane, Chengdu 610041, China

    Psychoradiology Research Unit of Chinese Academy of Medical Sciences, West China Hospital of Sichuan University, No. 37 Guo Xue Lane, Chengdu 610041, China
    Search for articles by this author
  • Qiyong Gong
    Affiliations
    Huaxi MR Research Center (HMRRC), Department of Radiology, West China Hospital of Sichuan University, No. 37 Guo Xue Lane, Chengdu 610041, China

    Psychoradiology Research Unit of Chinese Academy of Medical Sciences, West China Hospital of Sichuan University, No. 37 Guo Xue Lane, Chengdu 610041, China
    Search for articles by this author
  • John A. Sweeney
    Affiliations
    Department of Psychiatry and Behavioral Neuroscience, University of Cincinnati, Suite 3200, 260 Stetson Street, Cincinnati, OH 45219, USA
    Search for articles by this author
  • Author Footnotes
    1 F. Li and D. Wu contributed equally to this work.
Open AccessPublished:November 07, 2019DOI:https://doi.org/10.1016/j.nic.2019.09.001

      Keywords

      Key points

      • The biological heterogeneity of psychiatric syndromes and neurobiological mechanisms underpinning radiological abnormalities need to be addressed and further investigated.
      • Proper examination procedures, including optimal image acquisition, rigorous image quality control, standardized image processing, and individualized analysis, are essential for psychoradiology.
      • Psychoradiology has the potential to play an important role in clinical diagnosis, evaluation of treatment response and prognosis, and illness risk prediction for patients with psychiatric disorders.
      • Clinical challenges remain for improving psychoradiological utility and validity.

      Introduction

      Psychoradiology is an emerging discipline at the intersection between radiology and psychiatry. It applies radiological technologies to unveil patterns of anatomic and functional brain changes in patients with psychiatric disorders in vivo. It holds promise for playing a key role in clinical diagnosis, evaluation of treatment response, prediction of prognosis, and illness risk prediction for patients with psychiatric disorders.
      • Lui S.
      • Zhou X.J.
      • Sweeney J.A.
      • et al.
      Psychoradiology: the frontier of neuroimaging in psychiatry.
      Although psychoradiology was formally described by Lui and colleagues in 2016,
      • Lui S.
      • Zhou X.J.
      • Sweeney J.A.
      • et al.
      Psychoradiology: the frontier of neuroimaging in psychiatry.
      the idea of developing imaging biomarkers for psychiatric disorders dates back to 1976, when the first study revealed an enlarged ventricular size in patients with schizophrenia.
      • Johnstone E.C.
      • Crow T.J.
      • Frith C.D.
      • et al.
      Cerebral ventricular size and cognitive impairment in chronic schizophrenia.
      Gong and colleagues and other investigators have since developed the psychoradiological hypothesis of mental disorders, theorizing that brain structural and functional connectivity alterations lead to clinical symptoms and syndromes.
      • Lui S.
      • Zhou X.J.
      • Sweeney J.A.
      • et al.
      Psychoradiology: the frontier of neuroimaging in psychiatry.
      • Gong Q.
      • Lui S.
      • Sweeney J.A.
      A selective review of cerebral abnormalities in patients with first-episode schizophrenia before and after treatment.
      • Tregellas J.
      Connecting brain structure and function in schizophrenia.
      • Lui S.
      • Deng W.
      • Huang X.
      • et al.
      Association of cerebral deficits with clinical symptoms in antipsychotic-naive first-episode schizophrenia: an optimized voxel-based morphometry and resting state functional connectivity study.
      Well-replicated radiological observations
      • Klausner J.D.
      • Sweeney J.A.
      • Deck M.D.
      • et al.
      Clinical correlates of cerebral ventricular enlargement in schizophrenia. Further evidence for frontal lobe disease.
      have played a major role in the shift from seeing serious mental illnesses in psychological terms as problems of adaptation to life circumstances to the current view that they represent brain disorders.
      • Sarpal D.K.
      • Lencz T.
      • Malhotra A.K.
      In support of neuroimaging biomarkers of treatment response in first-episode schizophrenia.
      • Port J.D.
      Diagnosis of attention deficit hyperactivity disorder by using MR imaging and radiomics: a potential tool for clinicians.
      • Gong Q.
      Response to Sarpal et al.: importance of neuroimaging biomarkers for treatment development and clinical practice.
      This effort has accelerated to a great degree within the past 20 years due to the rapid and extensive development of magnetic resonance imaging (MRI), molecular imaging, and other diagnostic imaging techniques. In particular, new MRI technologies, such as high-resolution structural MRI, perfusion mapping, magnetic resonance spectroscopy (MRS), diffusion tensor imaging (DTI), and blood oxygenation level–dependent (BOLD) functional MRI (fMRI), have given rise to an increasing body of scientific literature that elucidates how various psychiatric syndromes are associated with alterations in brain structures and function across time and with treatment.
      Multiple imaging biomarkers have been identified in different psychiatric disorders, among which some have shown potential clinical utility for subtyping, prediction, and evaluation. Although the clinical use of psychoradiology is already in sight, there are still issues and challenges that need to be addressed before wide-scale clinical application. In particular, the clinical strategies for examining patients, analyzing images, and using findings to help clinical work need to be better validated and optimized.

      Strategy for examinations

      Like other imaging examinations, the first step of psychoradiology is to choose proper techniques for managing patients. Unlike tumor, stroke, and most neurologic diseases, brain abnormalities in psychiatric disorders are subtle and often involve functional alterations that contribute to cognitive and emotional disturbances. As a result, psychoradiology approaches require multimodal imaging techniques, especially high spatial resolution structural MRI, DTI, fMRI, perfusion-weighted imaging, MRS, electroencephalography, and positron emission tomography (PET). Multimodal imaging techniques requires clinical balance between the number of the techniques needed for particular patients or disorders and their cost. As in neurologic disorders, high-resolution T1-weighted imaging (T1WI) is used for detecting anatomic gray matter abnormality, DTI for white matter deficits, resting-state fMRI (rs-fMRI) for brain dysfunction identification, and MRS for neurometabolic information.
      A second question pertains to which patients need psychoradiological examination. Given the current state of knowledge, anyone with a suspected serious mental illness may benefit from an imaging examination, not only for ruling out other diseases, such as inflammation or tumor, but also as a baseline for subtyping and following treatment response. The clinical high-risk population, individuals with strong familial liability or prodromal manifestations of illness, also may benefit from an imaging examination to objectively assess risk and guide initiation of preventive interventions. Another issue is whether sedation or even anesthesia is needed for evaluating patients. A large majority of patients can cooperate with examination, sometimes with the help of mental health staff and relatives. In cases of some children or acutely ill manic patients, however, sedation may be needed, although sedating medications can affect imaging features, especially fMRI.
      • Starbuck V.N.
      • Kay G.G.
      • Platenberg R.C.
      • et al.
      Functional magnetic resonance imaging reflects changes in brain functioning with sedation.
      The safety of the patient needs to be an important consideration as in other imaging examinations. To reduce patient distress and increase cooperation, it can be advantageous to have a psychiatrist, psychologist, or relative accompany patients to scan sessions and, if needed, be with patients during an examination. A quiet ready room is helpful for preparing patients for examination and a mock scanner may be helpful to prepare patients. Magnetic resonance (MR)-compatible monitoring devices, such as an eye tracking system, are useful not only for monitoring patient safety but also for monitoring head position and collecting data about eye movements that have an impact on fMRI data.
      • Sweeney J.A.
      • Levy D.
      • Harris M.S.
      Commentary: eye movement research with clinical populations.
      • Rosano C.
      • Krisky C.M.
      • Welling J.S.
      • et al.
      Pursuit and saccadic eye movement subregions in human frontal eye field: a high-resolution fMRI investigation.
      • Sweeney J.A.
      • Luna B.
      • Keedy S.K.
      • et al.
      fMRI studies of eye movement control: investigating the interaction of cognitive and sensorimotor brain systems.
      Usually, the time for 1 MR examination is best limited to 40 minutes to 60 to maintain patient comfort and safety and optimize acquired image quality.
      Image quality control in psychoradiology is stricter than in other radiology disciplines because both structural and functional quantitative analyses are needed. In addition to general quality control, the control of the signal-to-noise ratio (SNR),
      • Yu S.
      • Dai G.
      • Wang Z.
      • et al.
      A consistency evaluation of signal-to-noise ratio in the quality assessment of human brain magnetic resonance images.
      the contrast-to-noise ratio (CNR), and image uniformity are required.
      • Jara J.L.
      • Saeed N.P.
      • Panerai R.B.
      • et al.
      Increasing the contrast-to-noise ratio of MRI signals for regional assessment of dynamic cerebral autoregulation.
      For example, when acquiring T1WI, the image interpolation function should be turned off, the acceleration factor should not exceed 2, and T2-weighted images with the same resolution should be acquired for better accuracy of brain surface reconstruction.
      • Misaki M.
      • Savitz J.
      • Zotev V.
      • et al.
      Contrast enhancement by combining T1- and T2-weighted structural brain MR Images.
      For DTI, the acceleration factors of 2 to 3 might be optimal, and the use of cardiac gating would be helpful in minimizing the tissue pulsation secondary to the cardiac cycle.
      • Yang E.
      • Nucifora P.G.
      • Melhem E.R.
      Diffusion MR imaging: basic principles.
      In rs-fMRI acquisition, an electrocardiogram needs to be acquired simultaneously for the correction of cardiac cycle effects on BOLD signals. The incorporation of a gradient-echo acquisition also can be useful in generating phase and magnitude images simultaneously that can be used for correcting or visualizing distortions caused by susceptibility and inhomogeneous fields.
      • Conklin C.J.
      • Faro S.H.
      • Mohamed F.B.
      Technical considerations for functional magnetic resonance imaging analysis.
      • Zaca D.
      • Agarwal S.
      • Gujar S.K.
      • et al.
      Special considerations/technical limitations of blood-oxygen-level-dependent functional magnetic resonance imaging.
      The acquisition of MRS data requires precise anatomic location of regions of interest (ROIs). For psychoradiology, the main focus shifts from visual inspection of images to quantitative analysis, leading to many of these requirements for image acquisition.

      Strategy for image analysis

      Unlike traditional qualitative diagnostic procedures in clinical radiology, psychoradiology is performed in a quantitative way. Thus, postprocessing is necessary after image acquisition. Although tools for such analysis are rapidly evolving, there is no guideline for the integrated analysis of multimodel imaging, which remains a challenge for the clinical application of psychoradiology. In this issue, some detailed strategies for data analysis are presented in the following chapters.
      In brief, for T1WI, there are 2 common methods for structural MRI postprocessing, voxel-based analysis (VBA) and surface-based morphologic (SBM) analysis. Voxel-based morphometry (VBM) analysis can be conducted via Statistical Parametric Mapping (SPM) (http://www.fil.ion.ucl.ac.uk/spm/), and the most commonly used method is diffeomorphic anatomical registration through exponentiated Lie algebra (DARTEL).
      • Good C.D.
      • Johnsrude I.S.
      • Ashburner J.
      • et al.
      A voxel-based morphometric study of ageing in 465 normal adult human brains.
      • Wright I.C.
      • McGuire P.K.
      • Poline J.B.
      • et al.
      A voxel-based method for the statistical analysis of gray and white matter density applied to schizophrenia.
      • Ashburner J.
      A fast diffeomorphic image registration algorithm.
      Basic procedures include segmentation of gray/white matter and cerebrospinal fluid, creation of DARTEL templates (6 templates through an 18 iteration process), registration of all individual deformations to the DARTEL template (the sixth template is the clearest for standard Montreal Neurological Institute space), and smoothing (usually with a full-width at half-maximum [FWHM] gaussian kernel of 8 mm to 12 mm).
      • Radua J.
      • Canales-Rodriguez E.J.
      • Pomarol-Clotet E.
      • et al.
      Validity of modulation and optimal settings for advanced voxel-based morphometry.
      The smoothed image allows for intergroup comparisons of gray matter density and gray matter volume; the image after modulation represents gray matter volume, and the unmodulated image represents gray matter density. In subsequent statistical analyses, the whole-brain volume of each subject is acquired as a covariate for comparison between groups to eliminate the effects of individual differences in total brain volume, and for comparison of individual patient values with normative data. SBM analysis is usually performed using FreeSurfer (http://surfer.nmr.mgh.harvard.edu) or CIVET (developed by the McConnell Brain Imaging Centre). The standard procedures of image preprocessing with FreeSurfer include head motion correction, removal of nonbrain tissue,
      • Segonne F.
      • Dale A.M.
      • Busa E.
      • et al.
      A hybrid approach to the skull stripping problem in MRI.
      • Fischl B.
      • Salat D.H.
      • Busa E.
      • et al.
      Whole brain segmentation: automated labeling of neuroanatomical structures in the human brain.
      • Fischl B.
      • van der Kouwe A.
      • Destrieux C.
      • et al.
      Automatically parcellating the human cerebral cortex.
      automated transformation to standard Talairach space, segmentation of the subcortical white matter and deep gray matter volumetric structures, intensity normalization,
      • Fischl B.
      • Dale A.M.
      Measuring the thickness of the human cerebral cortex from magnetic resonance images.
      tessellation of gray matter and white matter boundary, automated topology correction,
      • Fischl B.
      • van der Kouwe A.
      • Destrieux C.
      • et al.
      Automatically parcellating the human cerebral cortex.
      • Segonne F.
      • Pacheco J.
      • Fischl B.
      Geometrically accurate topology-correction of cortical surfaces using nonseparating loops.
      and surface deformation following intensity gradients to optimally place the gray/white matter and gray matter/cerebrospinal fluid borders where the greatest shift in intensity defines the transition to the other tissue class.
      • Jovicich J.
      • Czanner S.
      • Greve D.
      • et al.
      Reliability in multi-site structural MRI studies: effects of gradient non-linearity correction on phantom and human data.
      FreeSurfer automatically applies different FWHM (ie, 10 mm, 15 mm, 20 mm, or 25 mm) gaussian smoothing kernels for later statistical analysis. In statistical analysis, between-group comparisons have been performed with the general linear model method, which can include features, such as sex and age, as covariates.
      • Fischl B.
      • Sereno M.I.
      • Tootell R.B.
      • et al.
      High-resolution intersubject averaging and a coordinate system for the cortical surface.
      Cortical thickness, surface area, sulcal features (the number, depth, and frequency) and brain asymmetry can also be quantified.
      In DTI, by using tract-based spatial statistics (TBSS) and VBA, parameters, including fractional anisotropy (FA), mean diffusivity, packing density, myelination, and axon diameter, can be quantified to assess changes of the physical properties of white matter bundles. VBA registers all data of the subject (including gray matter and cerebrospinal fluid) into a standard space and performs statistical analyses on each voxel of the brain; then, it locates brain regions with altered parameters. The image data preprocessing and statistical analysis of VBA can be achieved by SPM version 2 and above. TBSS examines the whole brain without prespecifying tracts of interest for estimating localized change in FA by constructing average FA fiber skeleton maps of all subjects first and then registering all FA images of from patients on it to identify altered white matter tracts. TBSS combines the strengths of both VBAs with those of tractography-based analyses, thus avoiding the problem of inaccurate positioning caused by inaccurate image registration and smoothing. It can quantitatively and objectively evaluate white matter structure. Additionally, by using tractography, graph theory analyses can be performed to comprehensively evaluate white matter connectivity in the brain.
      • Owen J.P.
      • Ziv E.
      • Bukshpun P.
      • et al.
      Test-retest reliability of computational network measurements derived from the structural connectome of the human brain.
      • van den Heuvel M.P.
      • Sporns O.
      Rich-club organization of the human connectome.
      • Hu M.L.
      • Zong X.F.
      • Mann J.J.
      • et al.
      A review of the functional and anatomical default mode network in Schizophrenia.
      Rs-fMRI focuses on spontaneous changes in the BOLD signal in a resting state or task-negative state. In addition to regular data preprocessing, slice time correction, motion correction, spatial normalization, and spatial smoothing are required. The time series needs to be bandpass filtered to assess a particular frequency band of interest (eg, 0.01–0.08 Hz) to eliminate influences of low-frequency signal drift and high-frequency noise caused by respiration and heartbeat and to extract low-frequency signal oscillations that reflect spontaneous activity of the brain.
      • Barry R.L.
      • Williams J.M.
      • Klassen L.M.
      • et al.
      Evaluation of preprocessing steps to compensate for magnetic field distortions due to body movements in BOLD fMRI.
      Postprocessing typically is carried out in 1 of 2 ways: (1) the synchronization analysis of low-frequency oscillations between different brain regions for functional connectivity analyses,
      • Bullmore E.
      • Sporns O.
      Complex brain networks: graph theoretical analysis of structural and functional systems.
      which is used to identify connectivity alterations in neural networks, such as in the default mode network,
      • Hu M.L.
      • Zong X.F.
      • Mann J.J.
      • et al.
      A review of the functional and anatomical default mode network in Schizophrenia.
      • Whitfield-Gabrieli S.
      • Ford J.M.
      Default mode network activity and connectivity in psychopathology.
      and (2) the regional characteristic analysis of the low-frequency oscillations. The amplitude of low-frequency fluctuation, fractional amplitude of low-frequency fluctuation, and regional homogeneity are commonly used as indicators to investigate the characteristics of low frequency oscillations in local brain regions.
      MRS noninvasively detects and quantifies neurometabolites, such as N-acetylaspartate (NAA) (a putative marker of neuronal viability), choline (Cho) (a marker for membrane integrity and phospholipid metabolism), creatine (Cr) (a marker for energy metabolism), myoinositol (mI) (an astroglial marker), γ-aminobutyric acid (an inhibitory neurotransmitter in mammalian brain), and glutamate or/and glutamine (related to excitatory neurotransmission). One common finding in depression is the reduction of NAA in temporal and hippocampal regions. Other findings included reductions in NAA/Cr, NAA/Cho, and NAA/(Cr plus Cho). For example, the NAA/Cr ratio in prefrontal cortex of individuals with depression has been shown to be significantly lower than that of healthy individuals, with this deficit being greater in moderate than mild depression. Detailed information is provided in John D. Port's article, “Magnetic Resonance Spectroscopy for Psychiatry: Progress in the Last Decade,” in this issue. These quantitative results may be useful for clinical work.

      Strategy for clinical application

      Although traditional MRI is widely used to detect tumors or inflammation in patients with psychiatric disorders, the subtle brain abnormalities associated with psychiatric disorders have been noninvasively identified with advanced radiological technologies. Numerous basic, preclinical, and clinical studies have revealed a series of imaging biomarkers of brain structural and functional abnormalities. These studies have greatly promoted understanding of the pathologic mechanisms of abnormal brain structure and function in psychiatric disorders. The clinical value of imaging biomarkers for the most common psychiatric disorders, such as major depressive disorder, schizophrenia, posttraumatic stress disorder, and autism spectrum disorder, are presented in the later chapters of Section Three. In this article, strategies for clinical use are summarized. Generally, psychoradiological biomarkers considered in isolation have not yet been validated for making patient care decisions for individual patients; however, they can visually show psychiatrists and patients subtle structural and functional changes and in the near future help assist doctors in differential diagnosis, treatment planning, and prediction of illness course.

      Diagnosis and Subtyping

      Psychiatric disorders traditionally have been diagnosed and classified on the basis of broad syndromes defined by patients’ and parents’ reports and behavioral observations rather than on the basis of their underlying neurobiological substrates. As a result, psychiatric syndromes are heterogeneous and they biologically overlap,
      • Clementz B.A.
      • Sweeney J.A.
      • Hamm J.P.
      • et al.
      Identification of distinct psychosis biotypes using brain-based biomarkers.
      limiting the success of developing imaging and other biological biomarkers. Because this situation limits feasibility of identifying imaging biomarkers for different psychiatric syndromes, the field has moved in different directions, primarily toward developing objective psychoradiological biomarkers, such as structural features (alterations in gray matter/white matter volume, cortical morphometric features including thickness and surface area, and diffusion properties of white matter tracts) and functional features (activity and connectivity), which can facilitate early diagnosis and subtype patients into more biologically homogeneous groups. The US National Institute of Mental Health proposed the Research Domain Criteria project, with the aim to in part develop psychoradiological biomarkers for psychiatric disorders based on different dimensions of observable behaviors and neurobiological measures that are correlated to specific cognitive constructs across different brain systems.
      For example, Sun and colleagues
      • Sun H.
      • Lui S.
      • Yao L.
      • et al.
      Two patterns of white matter abnormalities in medication-naive patients with first-episode schizophrenia revealed by diffusion tensor imaging and cluster analysis.
      made 1 of the first efforts to subtype psychiatric disorders based on imaging features, and 2 distinct schizophrenia subtypes were identified using DTI. Using fMRI in a multisite sample, it was shown that patients with major depressive disorder can be divided into 4 neurophysiological subtypes (biotypes) by distinct patterns of dysfunctional connectivity in limbic and frontostriatal networks. Clustering patients on this basis led to high sensitivity and specificity (82%–93%) in the development of diagnostic classifiers for depression subtypes with multisite validation and out-of-sample replication data sets.
      • Drysdale A.T.
      • Grosenick L.
      • Downar J.
      • et al.
      Resting-state connectivity biomarkers define neurophysiological subtypes of depression.
      Similar studies have revealed new biotypes in schizophrenia, attention-deficit/hyperactivity disorder (ADHD), bipolar disorder, and other psychiatric disorders,
      • Clementz B.A.
      • Sweeney J.A.
      • Hamm J.P.
      • et al.
      Identification of distinct psychosis biotypes using brain-based biomarkers.
      • Insel T.R.
      • Cuthbert B.N.
      Medicine. Brain disorders? Precisely.
      • Karalunas S.L.
      • Fair D.
      • Musser E.D.
      • et al.
      Subtyping attention-deficit/hyperactivity disorder using temperament dimensions: toward biologically based nosologic criteria.
      • Tamminga C.A.
      • Pearlson G.
      • Keshavan M.
      • et al.
      Bipolar and schizophrenia network for intermediate phenotypes: outcomes across the psychosis continuum.
      which highlights the potential application for psychoradiology to help subtype patients with psychiatric disorders based on objective imaging markers rather than observation of behavior and symptom profiles and for using these classifications to better individualize patient care in parallel with development of optimal treatment strategies for the identified subgroups.

      Guiding Therapeutic Intervention and Interventional Psychoradiology

      Psychoradiological biomarkers can be of importance in guiding treatment of patients with psychiatric disorders by helping clinicians make difficult differential diagnoses and select optimal treatment procedures and targets for subgroups of patients with particular neurobiological abnormalities. Psychoradiological biomarkers also may help predict psychiatric disorders in at-risk individuals so that primary prevention approaches can be implemented for those who most need them, so patients likely to be refractory to first-line treatments can be identified before beginning treatment, and so that the fundamental understanding of causal neural mechanisms of illness can be identified to spur development of novel treatments.
      • Doehrmann O.
      • Ghosh S.S.
      • Polli F.E.
      • et al.
      Predicting treatment response in social anxiety disorder from functional magnetic resonance imaging.
      • Hahn T.
      • Kircher T.
      • Straube B.
      • et al.
      Predicting treatment response to cognitive behavioral therapy in panic disorder with agoraphobia by integrating local neural information.
      • Ma N.
      • Li L.
      • Shu N.
      • et al.
      White matter abnormalities in first-episode, treatment-naive young adults with major depressive disorder.
      • Martinez D.
      • Carpenter K.M.
      • Liu F.
      • et al.
      Imaging dopamine transmission in cocaine dependence: link between neurochemistry and response to treatment.
      • Whitfield-Gabrieli S.
      • Ghosh S.S.
      • Nieto-Castanon A.
      • et al.
      Brain connectomics predict response to treatment in social anxiety disorder.
      For example, studies have identified a distributed pattern of brain activity reflected in fMRI responses during fear conditioning, which can discriminate patients with panic disorder who responded to cognitive behavioral therapy from those who did not with 82% accuracy.
      • Ma N.
      • Li L.
      • Shu N.
      • et al.
      White matter abnormalities in first-episode, treatment-naive young adults with major depressive disorder.
      Using resting-state functional connectivity analyses, other studies discovered that disrupted functional connectivity mainly in thalamocortical circuits is correlated to refractory depression, whereas more widespread decreased functional connectivity in the limbic-striatal-pallidal-thalamic circuit is associated with nonrefractory depression.
      • Doehrmann O.
      • Ghosh S.S.
      • Polli F.E.
      • et al.
      Predicting treatment response in social anxiety disorder from functional magnetic resonance imaging.
      Imaging biomarkers also could provide valuable information about treatment targets.
      • Gong Q.
      • Lui S.
      • Sweeney J.A.
      A selective review of cerebral abnormalities in patients with first-episode schizophrenia before and after treatment.
      • Li F.
      • Lui S.
      • Yao L.
      • et al.
      Longitudinal changes in resting-state cerebral activity in patients with first-episode schizophrenia: a 1-year follow-up functional MR imaging study.
      For example, regions including prefrontal cortex and striatum receive robust dopaminergic projections, which are believed to be implicated in the pathogenesis of schizophrenia. Hypofunction of the medial prefrontal cortex as well as hyperactivity of the hippocampus and striatum, in patients with schizophrenia may in time provide psychoradiological biomarkers for the targets of treatment.
      • Gong Q.
      • Lui S.
      • Sweeney J.A.
      A selective review of cerebral abnormalities in patients with first-episode schizophrenia before and after treatment.
      These findings demonstrate promising new evidence that psychoradiological biomarkers may provide valuable information in monitoring and predicting treatment response, which could be of great importance in detecting patients at an early stage who require adjunctive medical and psychosocial therapies, because they may not respond to first-line treatments, and those likely to recover without any intervention, by optimizing timing, intensity, and form of therapeutic intervention.

      Prediction of Illness Onset

      Predicting the onset of illness for a high-risk person is another important role of psychoradiology. Psychoradiological studies have suggested that the brain’s structure and function are different between high-risk individuals who subsequently develop psychosis and individuals who do not.
      • Addington J.
      • Heinssen R.
      Prediction and prevention of psychosis in youth at clinical high risk.
      • Beardslee W.R.
      • Brent D.A.
      • Weersing V.R.
      • et al.
      Prevention of depression in at-risk adolescents: longer-term effects.
      • Cao H.
      • Chen O.Y.
      • Chung Y.
      • et al.
      Cerebello-thalamo-cortical hyperconnectivity as a state-independent functional neural signature for psychosis prediction and characterization.
      • Chung Y.
      • Addington J.
      • Bearden C.E.
      • et al.
      Use of machine learning to determine deviance in neuroanatomical maturity associated with future psychosis in youths at clinically high risk.
      For example, using multiparadigm fMRI data to investigate network-level changes in functional connectome of the human brain, Cao and colleagues
      • Cao H.
      • Chen O.Y.
      • Chung Y.
      • et al.
      Cerebello-thalamo-cortical hyperconnectivity as a state-independent functional neural signature for psychosis prediction and characterization.
      found an individual-specific “trait” abnormality in brain architecture characterized as increased connectivity in the cerebello-thalamo-cortical circuitry in individuals at clinical high risk for psychosis. This is a pattern that is significantly more pronounced among those who develop psychosis than those who do not among high-risk individuals. This abnormality is significantly associated with thought disorder and predictive of time to conversion.
      These findings highlight the potential for psychoradiology to help identify those at high risk for developing psychosis to predict who will later convert to a disease state in advance of its onset. This knowledge can indicate a need for those at greatest risk for early preventive pharmacologic and psychological interventions, while sparing those with lower conversion risk of unnecessary exposure to treatment side effects. These advances could help optimize allocation of clinical resources in mental health care systems.
      The workflow pipeline of psychoradiology in clinical practice is summarized in Fig. 1.
      Figure thumbnail gr1
      Fig. 1Workflow pipeline of psychoradiology in clinical practice. AFNI, analyses of functional neuroimages; DPABI, data processing and analysis for brain imaging; EEG, electroencephalography; FSL, FMRIB software library; GRE, gradient echo; PWI, perfusion-weighted imaging; REST, resting-state fMRI data analysis toolkit.

      Future challenges

      Although psychoradiology has great promise as a clinical discipline aiding in the diagnosis and treatment of psychiatric patients, there are still many issues and challenges in this growing field that need to be addressed before routine clinical application.

      Problems in Clinical Translation

      First, MRI-based brain volume measurements can be influenced by various technical parameters, especially when reliability and reproducibility are critical for clinical translation in psychoradiology. For instance, the number of head coil channels,
      • Krueger G.
      • Granziera C.
      • Jack Jr., C.R.
      • et al.
      Effects of MRI scan acceleration on brain volume measurement consistency.
      inconsistent subject positioning,
      • Caramanos Z.
      • Fonov V.S.
      • Francis S.J.
      • et al.
      Gradient distortions in MRI: characterizing and correcting for their effects on SIENA-generated measures of brain volume change.
      inconsistent image contrast,
      • Preboske G.M.
      • Gunter J.L.
      • Ward C.P.
      • et al.
      Common MRI acquisition non-idealities significantly impact the output of the boundary shift integral method of measuring brain atrophy on serial MRI.
      and differences in MR scanner vendors, and field strength
      • Zhang F.
      • Chen G.
      • He M.
      • et al.
      Altered white matter microarchitecture in amyotrophic lateral sclerosis: a voxel-based meta-analysis of diffusion tensor imaging.
      can have an impact on on interpretation of quantitative image characteristics. Heterogeneity of MRI scanning parameters across studies and sites in voxel size, number of diffusion directions, and slice thickness may have resulted in decreased reliability of functional and structural MRI studies in psychiatric disorders.
      • Zhang F.
      • Chen G.
      • He M.
      • et al.
      Altered white matter microarchitecture in amyotrophic lateral sclerosis: a voxel-based meta-analysis of diffusion tensor imaging.
      • Chen G.
      • Hu X.
      • Li L.
      • et al.
      Disorganization of white matter architecture in major depressive disorder: a meta-analysis of diffusion tensor imaging with tract-based spatial statistics.
      • Church J.A.
      • Petersen S.E.
      • Schlaggar B.L.
      The "Task B problem" and other considerations in developmental functional neuroimaging.
      These differences are difficult to eliminate by statistical means. Homogenization of technical considerations becomes more crucial with the introduction of multiple MRI scanners, in a situation where multicenter trials or cross-site application of a diagnostic or predictive algorithm are used. Kruggel and colleagues
      • Kruggel F.
      • Turner J.
      • Muftuler L.T.
      • et al.
      Impact of scanner hardware and imaging protocol on image quality and compartment volume precision in the ADNI cohort.
      compared different scanner platforms of 1.5T and 3.0T and found that different levels of image quality, regarding SNR, CNR, and combined information of the joint histogram limited the consistency of brain volume measurements. Hence, data from different scanning protocols and platforms must be carefully considered to avoid confounding the true effects of interest with variability among scanning platforms. Also, this issue highlights the need to establish optimal acquisition parameters for specific clinical applications.
      An early study
      • Manoach D.S.
      • Halpern E.F.
      • Kramer T.S.
      • et al.
      Test-retest reliability of a functional MRI working memory paradigm in normal and schizophrenic subjects.
      in schizophrenia with fMRI suggested that factors of variation (both artifactual and intrinsic) could be controlled to improve test-retest reliability. Zhao and colleagues
      • Zhao N.
      • Yuan L.X.
      • Jia X.Z.
      • et al.
      Intra- and inter-scanner reliability of voxel-wise whole-brain analytic metrics for resting state fMRI.
      analyzed data from 21 subjects who underwent 2 scans within 2 weeks on a 3T MRI scanner from General Electric (Milwaukee, United States) and the third visit on a 3T MRI scanner from Siemens approximately 8 months later for assessment of intrascanner and interscanner reliability of rs-fMRI based on voxelwise whole-brain analytical metrics. The rs-fMRI results
      • Zhao N.
      • Yuan L.X.
      • Jia X.Z.
      • et al.
      Intra- and inter-scanner reliability of voxel-wise whole-brain analytic metrics for resting state fMRI.
      • Jovicich J.
      • Czanner S.
      • Han X.
      • et al.
      MRI-derived measurements of human subcortical, ventricular and intracranial brain volumes: reliability effects of scan sessions, acquisition sequences, data analyses, scanner upgrade, scanner vendors and field strengths.
      indicated that the data were reliable within the same scanner, whereas inter-canner reliability was a more significant factor when data were compared across platforms and field strength. Data, including MRI-derived measurements in a longitudinal morphometric study of human brains,
      • Zhao N.
      • Yuan L.X.
      • Jia X.Z.
      • et al.
      Intra- and inter-scanner reliability of voxel-wise whole-brain analytic metrics for resting state fMRI.
      • Lee H.
      • Nakamura K.
      • Narayanan S.
      • et al.
      Estimating and accounting for the effect of MRI scanner changes on longitudinal whole-brain volume change measurements.
      were trustworthy at a single 1.5T site, even with different sequences or after upgrading systems; nevertheless, the reliability should be carefully considered when images are collected across vendors and/or field strength.
      • Jovicich J.
      • Czanner S.
      • Han X.
      • et al.
      MRI-derived measurements of human subcortical, ventricular and intracranial brain volumes: reliability effects of scan sessions, acquisition sequences, data analyses, scanner upgrade, scanner vendors and field strengths.
      Changes in scanner hardware can lead to the introduction of different bias effects in the brain analyses, whereas intervendor changes generally exerted greater effects compared with intravendor scanner changes.
      • Kruggel F.
      • Turner J.
      • Muftuler L.T.
      • et al.
      Impact of scanner hardware and imaging protocol on image quality and compartment volume precision in the ADNI cohort.
      In the context of quantitative analysis to detect relatively subtle brain alterations, consideration of the impact of technical factors in multisite research and in developing widely useful normative data are particularly important.
      The acquisition of structural MRI and fMRI data is costly, especially regarding the time involved in postprocessing. Furthermore, the quality of MRI data is affected by many factors, including head motion, and physiologic factors, such as heart beating and breathing, are critical biological confounds. This is especially with fMRI, where cardiopulmonary function, age and sex effects, and anatomic variability have an impact on data interpretation.
      • Lee H.
      • Nakamura K.
      • Narayanan S.
      • et al.
      Estimating and accounting for the effect of MRI scanner changes on longitudinal whole-brain volume change measurements.
      • Khalili-Mahani N.
      • Rombouts S.A.R.B.
      • van Osch M.J.P.
      • et al.
      Biomarkers, designs, and interpretations of resting-state fMRI in translational pharmacological research: a review of state-of-the-Art, challenges, and opportunities for studying brain chemistry.
      Thus, individual patient variability and protocol/scanner factors are critical for clinical application.
      Therefore, it is imperative to establish standardized data acquisition and image quality control solutions. To address heterogeneity between different centers/sequences, a standardized MRI sequence is needed that can generate images with similar properties concerning SNR, CNR, voxel size, and slice thickness, regardless of the scanner platform and manufacturer. Homogeneous data acquisition and analysis can be provided by a specific protocol across sites, like that developed by the Alzheimer’s Disease Neuroimaging Initiative consortium.
      • Bocchetta M.
      • Boccardi M.
      • Ganzola R.
      • et al.
      Harmonized benchmark labels of the hippocampus on magnetic resonance: the EADC-ADNI project.
      Even with these efforts, heterogeneities based on site still exist for complicated reasons.
      • Pearlson G.
      Multisite collaborations and large databases in psychiatric neuroimaging: advantages, problems, and challenges.
      Multitask learning has been deployed to simultaneously learn the features of site-shared and site-specific features extracted from multicenter MRI data of brain morphology. Neuroimaging studies have demonstrated the advantages of multitask learning to decode brain alterations and for classifying disease.
      • Ma Q.
      • Zhang T.
      • Zanetti M.V.
      • et al.
      Classification of multi-site MR images in the presence of heterogeneity using multi-task learning.
      • Wang X.
      • Zhang T.
      • Chaim T.M.
      • et al.
      Classification of MRI under the presence of disease heterogeneity using multi-task learning: application to bipolar disorder.
      Second, it is necessary to develop stable and efficient semiautomated computational approaches for image analysis. Studies of brain morphometry
      • Stevens J.R.
      An anatomy of Schizophrenia?.
      • Rozycki M.
      • Satterthwaite T.D.
      • Koutsouleris N.
      • et al.
      Multisite machine learning analysis provides a robust structural imaging signature of schizophrenia detectable across diverse patient populations and within individuals.
      initially used morphometric measurements obtained from brain regions with manual delineation of ROIs. This ROI-based method sometimes encountered difficulties under certain conditions for the delineation of unambiguous structures, such as the hippocampi or the ventricles. VBAs and surface-based analyses can be used to identify whole-brain changes
      • Scarpazza C.
      • De Simone M.
      Voxel-based morphometry: current perspectives.
      ; these analyses are automated, relatively easy-to-use, time-efficient tools and have been widely used in psychiatric research.
      • Salvador R.
      • Radua J.
      • Canales-Rodriguez E.J.
      • et al.
      Evaluation of machine learning algorithms and structural features for optimal MRI-based diagnostic prediction in psychosis.
      • Steele V.R.
      • Rao V.
      • Calhoun V.D.
      • et al.
      Machine learning of structural magnetic resonance imaging predicts psychopathic traits in adolescent offenders.
      • Zhang J.
      • Liu W.
      • Zhang J.
      • et al.
      Distinguishing adolescents with conduct disorder from typically developing youngsters based on pattern classification of brain structural MRI.
      Such approaches can also shorten the duration of the evaluation pipeline to speed availability of feedback to referring physicians. The recent development of psychoradiological tools to detect the individual-specific biomarkers in patients with psychotic disorders has been extremely exciting.
      • Wang D.
      • Li M.
      • Wang M.
      • et al.
      Individual-specific functional connectivity markers track dimensional and categorical features of psychotic illness.
      • Huang X.
      • Gong Q.
      • Sweeney J.A.
      • et al.
      Progress in psychoradiology, the clinical application of psychiatric neuroimaging.
      Although reliable, fully automated, standardized methods can improve implementation across different sites with a unified software platform, an individualized approach can be implemented as suited for specific purposes.
      • Abi-Dargham A.
      • Horga G.
      The search for imaging biomarkers in psychiatric disorders.
      Standardized pipelines for preprocessing MRI data are becoming increasingly sophisticated, such as improved sequences linked to the Human Connectome Project.
      • Glasser M.F.
      • Sotiropoulos S.N.
      • Wilson J.A.
      • et al.
      The minimal preprocessing pipelines for the Human Connectome Project.
      Third, the development of fast multimodal imaging facilities is important. To discover robust neuroimaging biomarkers for diagnosis and patient stratification, multicenter and multimodal studies are becoming popular, thereby increasing sample size and providing detailed imaging features of psychosis patients.
      • Kempton M.J.
      • McGuire P.
      How can neuroimaging facilitate the diagnosis and stratification of patients with psychosis?.
      Multimodal imaging, which can combine structural MRI, DTI, fMRI, and even PET data together, can scan subjects with a variety of sequences to acquire structural, functional, and metabolic data of the brain in a single session.
      • Smieskova R.
      • Allen P.
      • Simon A.
      • et al.
      Different duration of at-risk mental state associated with neurofunctional abnormalities. A multimodal imaging study.
      • Fusar-Poli P.
      • Howes O.D.
      • Allen P.
      • et al.
      Abnormal frontostriatal interactions in people with prodromal signs of psychosis: a multimodal imaging study.
      It is a challenge for psychoradiologists to interpret these combined data sets acquired from variable combinations of different imaging modalities and methodologies, but machine learning approaches offer promise for usefully organizing multimodal data.
      A method was specifically developed to analyze multimodal imaging data by Radua and colleagues.
      • Radua J.
      • Borgwardt S.
      • Crescini A.
      • et al.
      Multimodal meta-analysis of structural and functional brain changes in first episode psychosis and the effects of antipsychotic medication.
      This technique is a voxel-wise multimodal meta-analysis applied to anatomic and fMRI examinations in first-episode psychosis patients. This meta-analysis identified both structural and functional abnormalities in the brain as well as heterogeneity between studies. To speed up the reconstruction of images, Xiang and colleagues
      • Xiang L.
      • Chen Y.
      • Chang W.
      • et al.
      Deep leaning based multi-modal fusion for fast MR reconstruction.
      fused multimodal MR acquisitions through deep learning. This deep learning approach reconstructed a 3-dimensional T2WI volume from the T1WI data and undersampled T2WI images. This approach was applied to data sets acquired by different MR vendors, and it showed excellent transferring capability. These approaches open the way for faster and more efficient multimodal image acquisition in clinical settings.

      Pathophysiology of Imaging Signs

      In addition to the challenges of particular imaging techniques, the unclear pathophysiology of imaging biomarkers is another challenge when explaining the meaning of brain imaging findings. Until this is resolved, psychoradiology will remain an empirical or actuarial field.
      For example, in vivo neuroimaging studies
      • Jauhar S.
      • McCutcheon R.
      • Borgan F.
      • et al.
      The relationship between cortical glutamate and striatal dopamine in first-episode psychosis: a cross-sectional multimodal PET and magnetic resonance spectroscopy imaging study.
      • Howes O.
      • McCutcheon R.
      • Stone J.
      Glutamate and dopamine in schizophrenia: an update for the 21st century.
      have supported the hypotheses that cortical glutamate dysfunction has an impact on subcortical dopamine synthesis capacity, which is a leading theory of the pathogenesis of schizophrenia. It is unclear, however, how alterations in in vivo cortical glutamate and dopamine levels and function result in the pathophysiology of psychosis. PET studies have shown that multiple neurochemical systems and molecular mechanisms can play roles in the pathophysiology of psychosis in the early course of the syndrome.
      • Schifani C.
      • Hafizi S.
      • Da Silva T.
      • et al.
      Using molecular imaging to understand early schizophrenia-related psychosis neurochemistry: a review of human studies.
      • Salavati B.
      • Rajji T.K.
      • Price R.
      • et al.
      Imaging-based neurochemistry in schizophrenia: a systematic review and implications for dysfunctional long-term potentiation.
      In terms of correlations between brain activation and behavior using task-based fMRI (studying brain activity as a particular cognitive, motor or emotion task is performed) to study individual differences, it is crucial that behavioral paradigms be optimized for clinical application. These studies can be effective in challenging brain circuitry of clinical interest and relating brain alterations to clinical features of interest, but add an additional layer of methods validation.
      Region-specific structural changes in the rat cortex
      • Vernon A.C.
      • Crum W.R.
      • Lerch J.P.
      • et al.
      Reduced cortical volume and elevated astrocyte density in rats chronically treated with antipsychotic drugs-linking magnetic resonance imaging findings to cellular pathology.
      • Vernon A.C.
      • Natesan S.
      • Modo M.
      • et al.
      Effect of chronic antipsychotic treatment on brain structure: a serial magnetic resonance imaging study with ex vivo and postmortem confirmation.
      after chronic antipsychotic treatment raise the issue that psychopharmacological treatments themselves can alter brain anatomy and function and thus have an impact on the use of algorithms for diagnosis and prediction. Postmortem structural studies in schizophrenia
      • Dorph-Petersen K.A.
      • Lewis D.A.
      Postmortem structural studies of the thalamus in schizophrenia.
      found volume and cell number reduction in the pulvinar with stereological studies of the thalamus. This finding is in accordance with the view that thalamocortical dysfunction might play a role in schizophrenia because of the function of the thalamus as a key node in whole-brain neuronal circuits. This type of finding highlights the importance of network-level analysis of brain systems in studies of psychiatric illness. Structural imaging studies
      • Erp T.G.M.V.
      • Hibar D.P.
      • Rasmussen J.M.
      • et al.
      Subcortical brain volume abnormalities in 2028 individuals with schizophrenia and 2540 healthy controls via the ENIGMA consortium.
      • Shepherd A.M.
      • Laurens K.R.
      • Matheson S.L.
      • et al.
      Systematic meta-review and quality assessment of the structural brain alterations in schizophrenia.
      provide evidence of volume reductions in bilateral thalami in schizophrenia that support the thalamocortical dysfunction hypothesis in patients with schizophrenia. Although understanding of the pathogenesis of psychosis remains limited, these initial results highlight the potential value of psychoradiology in better understanding the pathogenesis of serious mental illness and the translation of these findings into clinical practice.

      Heterogeneity of Psychiatric Disorders

      Psychiatric disorders are now diagnostically classified as broad syndromes defined based on patient complaints and behavioral observations.
      • Lui S.
      • Zhou X.J.
      • Sweeney J.A.
      • et al.
      Psychoradiology: the frontier of neuroimaging in psychiatry.
      The overlap of these syndromes in terms of illness presentation, genetics, neurobiology, and treatment response profiles remains large. This highlights the importance of psychoradiology not only for patient care but also as an important field working toward nosologic reorganization for diagnosis of serious mental illnesses based on biological features. In this latter effort, in vivo brain imaging can play a crucial role, to the great advantage of efforts to identify biologically discrete patient subgroups requiring specific therapies based on the nature of their brain disturbances.
      Individuals with symptoms indicating that they are at risk for serious mental illness based on psychological assessment may share a phenotypic expression but one that results from different underlying brain abnormalities as suggested by recent large-scale multisite collaborative studies. For instance, autism spectrum disorder and schizophrenia share clinical features, including social withdrawal, theory of mind deficits, and sensory abnormalities.
      • Couture S.M.
      • Penn D.L.
      • Losh M.
      • et al.
      Comparison of social cognitive functioning in schizophrenia and high functioning autism: more convergence than divergence.
      • Hommer R.E.
      • Swedo S.E.
      Schizophrenia and autism-related disorders.
      • Pina-Camacho L.
      • Parellada M.
      • Kyriakopoulos M.
      Autism spectrum disorder and schizophrenia: boundaries and uncertainties.
      It is not uncommon that several mental disorders show similar anatomic and functional deficits in brain networks including schizophrenia and bipolar disorder.
      • Ivleva E.I.
      • Clementz B.A.
      • Dutcher A.M.
      • et al.
      Brain structure biomarkers in the psychosis biotypes: findings from the bipolar-schizophrenia network for intermediate phenotypes.
      Psychiatric disorders share similar neurocognitive deficits
      • Hill S.K.
      • Reilly J.L.
      • Keefe R.S.
      • et al.
      Neuropsychological impairments in schizophrenia and psychotic bipolar disorder: findings from the Bipolar-Schizophrenia Network on Intermediate Phenotypes (B-SNIP) study.
      and overlapping features of emotional disorders.
      • Lui S.
      • Zhou X.J.
      • Sweeney J.A.
      • et al.
      Psychoradiology: the frontier of neuroimaging in psychiatry.
      Wang and colleagues
      • Wang J.B.
      • Zheng L.J.
      • Cao Q.J.
      • et al.
      Inconsistency in abnormal brain activity across cohorts of ADHD-200 in children with attention deficit hyperactivity disorder.
      analyzed the multisite data set of patients with ADHD (ie, data from the ADHD-200
      • Consortium H.
      The ADHD-200 consortium: a model to advance the translational potential of neuroimaging in clinical neuroscience.
      ) with rs-fMRI based on 3 whole-brain VBA methods. Abnormal activity was found in some brain regions with the pooled data set; however, these results were highly heterogeneous across cohorts and within the same research center. The results from the study of multisite functional connectivity classification of autism
      • Skatun K.C.
      • Kaufmann T.
      • Doan N.T.
      • et al.
      Consistent functional connectivity alterations in schizophrenia spectrum disorder: a multisite study.
      also indicated a poorer degree of accuracy of whole-brain functional connectivity in favor of heterogeneous features of connectivity disturbances with a particular spatial distribution in specific brain regions.
      The clinical and imaging heterogeneity of psychiatric syndromes resulted in compromised specificity when assigning psychiatric disorders using machine learning approaches with MRI data. This has led many in psychiatric research to step back and conclude that the next step forward needs to be collecting large samples in multisite projects with dense phenotyping (including but not restricted to MRI) to identify biologically discrete patient groups, which can then be studied separately or stratified in clinical trials and genetic research. In this way, the clinical relevance of novel patient subgrouping approaches developed using MRI data can be evaluated. This is especially important as decades of patient subgrouping efforts based on psychological and behavioral features has produced limited gains. Until this issue of within diagnosis heterogeneity is resolved, the ability to identify MR biomarkers for diagnosis or guiding treatment decisions remains limited. The best way for psychoradiology to address this issue is to use psychoradiology and machine learning with prolonged imaging times, standardized acquisition strategies, advanced classification methodology, and large sample sizes to identify distinct patterns of brain abnormalities that run across and are not specific to particular psychiatric syndromes, and to define patterns of brain alterations and their relation to cognitive, affective, and behavioral clinical manifestations.
      • Lui S.
      • Zhou X.J.
      • Sweeney J.A.
      • et al.
      Psychoradiology: the frontier of neuroimaging in psychiatry.
      This effort will require considerable interdisciplinary collaboration, from radiologists and physicists optimizing acquisition protocols, to statisticians and programmers developing rapid and automated measurement of brain features of interest, and to adiologists, psychiatrists, and psychologists working to translate new approaches into clinically useful contributions for improving psychiatric patient care.

      Summary

      Psychoradiology is a young and evolving field. Research to date is already showing the utility of MRI data in psychiatry for facilitating clinical diagnosis, evaluation of treatment response and prognosis, identifying patient subgroups, and illness risk prediction. Further advances to translate these observations into clinical practice will require proper examination and validation of image acquisition and processing methods, rigorous image quality control, and standardized semiautomated image analysis. Addressing complex issues, such as the biological heterogeneity of psychiatric syndromes and unclear neurobiological mechanisms underpinning radiological abnormalities, is a challenge that needs to be resolved but one for which psychoradiology can make important contributions. With the advance of multimodal imaging and more efforts in standardization of image acquisition and analysis, psychoradiology has become a promising tool for the future of the clinical care of patients with psychiatric disorders.

      Acknowledgments

      This study was supported by the National Natural Science Foundation of China (Grant Nos. 81671664, 81621003, 81761128023 and 81820108018). The authors acknowledge the support from Chang Jiang Scholars of China (Award Nos. T2014190, IRT16R52 and Q2015154), National Program for Support of Top-notch Young Professionals (National Program for Special Support of Eminent Professionals, Award No. W02070140) and the Functional and Molecular Imaging Key Laboratory of Sichuan Province (FMIKLSP, Grant: 2019JDS0044).

      References

        • Lui S.
        • Zhou X.J.
        • Sweeney J.A.
        • et al.
        Psychoradiology: the frontier of neuroimaging in psychiatry.
        Radiology. 2016; 281: 357-372
        • Johnstone E.C.
        • Crow T.J.
        • Frith C.D.
        • et al.
        Cerebral ventricular size and cognitive impairment in chronic schizophrenia.
        Lancet. 1976; 2: 924-926
        • Gong Q.
        • Lui S.
        • Sweeney J.A.
        A selective review of cerebral abnormalities in patients with first-episode schizophrenia before and after treatment.
        Am J Psychiatry. 2016; 173: 232-243
        • Tregellas J.
        Connecting brain structure and function in schizophrenia.
        Am J Psychiatry. 2009; 166: 134-136
        • Lui S.
        • Deng W.
        • Huang X.
        • et al.
        Association of cerebral deficits with clinical symptoms in antipsychotic-naive first-episode schizophrenia: an optimized voxel-based morphometry and resting state functional connectivity study.
        Am J Psychiatry. 2009; 166: 196-205
        • Klausner J.D.
        • Sweeney J.A.
        • Deck M.D.
        • et al.
        Clinical correlates of cerebral ventricular enlargement in schizophrenia. Further evidence for frontal lobe disease.
        J Nerv Ment Dis. 1992; 180: 407-412
        • Sarpal D.K.
        • Lencz T.
        • Malhotra A.K.
        In support of neuroimaging biomarkers of treatment response in first-episode schizophrenia.
        Am J Psychiatry. 2016; 173: 732-733
        • Port J.D.
        Diagnosis of attention deficit hyperactivity disorder by using MR imaging and radiomics: a potential tool for clinicians.
        Radiology. 2018; 287: 631-632
        • Gong Q.
        Response to Sarpal et al.: importance of neuroimaging biomarkers for treatment development and clinical practice.
        Am J Psychiatry. 2016; 173: 733-734
        • Starbuck V.N.
        • Kay G.G.
        • Platenberg R.C.
        • et al.
        Functional magnetic resonance imaging reflects changes in brain functioning with sedation.
        Hum Psychopharmacol. 2000; 15: 613-618
        • Sweeney J.A.
        • Levy D.
        • Harris M.S.
        Commentary: eye movement research with clinical populations.
        Prog Brain Res. 2002; 140: 507-522
        • Rosano C.
        • Krisky C.M.
        • Welling J.S.
        • et al.
        Pursuit and saccadic eye movement subregions in human frontal eye field: a high-resolution fMRI investigation.
        Cereb Cortex. 2002; 12: 107-115
        • Sweeney J.A.
        • Luna B.
        • Keedy S.K.
        • et al.
        fMRI studies of eye movement control: investigating the interaction of cognitive and sensorimotor brain systems.
        Neuroimage. 2007; 36: T54-T60
        • Yu S.
        • Dai G.
        • Wang Z.
        • et al.
        A consistency evaluation of signal-to-noise ratio in the quality assessment of human brain magnetic resonance images.
        BMC Med Imaging. 2018; 18: 17
        • Jara J.L.
        • Saeed N.P.
        • Panerai R.B.
        • et al.
        Increasing the contrast-to-noise ratio of MRI signals for regional assessment of dynamic cerebral autoregulation.
        Acta Neurochir Suppl. 2018; 126: 153-157
        • Misaki M.
        • Savitz J.
        • Zotev V.
        • et al.
        Contrast enhancement by combining T1- and T2-weighted structural brain MR Images.
        Magn Reson Med. 2015; 74: 1609-1620
        • Yang E.
        • Nucifora P.G.
        • Melhem E.R.
        Diffusion MR imaging: basic principles.
        Neuroimaging Clin N Am. 2011; 21 (vii): 1-25
        • Conklin C.J.
        • Faro S.H.
        • Mohamed F.B.
        Technical considerations for functional magnetic resonance imaging analysis.
        Neuroimaging Clin N Am. 2014; 24: 695-704
        • Zaca D.
        • Agarwal S.
        • Gujar S.K.
        • et al.
        Special considerations/technical limitations of blood-oxygen-level-dependent functional magnetic resonance imaging.
        Neuroimaging Clin N Am. 2014; 24: 705-715
        • Good C.D.
        • Johnsrude I.S.
        • Ashburner J.
        • et al.
        A voxel-based morphometric study of ageing in 465 normal adult human brains.
        Neuroimage. 2001; 14: 21-36
        • Wright I.C.
        • McGuire P.K.
        • Poline J.B.
        • et al.
        A voxel-based method for the statistical analysis of gray and white matter density applied to schizophrenia.
        Neuroimage. 1995; 2: 244-252
        • Ashburner J.
        A fast diffeomorphic image registration algorithm.
        Neuroimage. 2007; 38: 95-113
        • Radua J.
        • Canales-Rodriguez E.J.
        • Pomarol-Clotet E.
        • et al.
        Validity of modulation and optimal settings for advanced voxel-based morphometry.
        Neuroimage. 2014; 86: 81-90
        • Segonne F.
        • Dale A.M.
        • Busa E.
        • et al.
        A hybrid approach to the skull stripping problem in MRI.
        Neuroimage. 2004; 22: 1060-1075
        • Fischl B.
        • Salat D.H.
        • Busa E.
        • et al.
        Whole brain segmentation: automated labeling of neuroanatomical structures in the human brain.
        Neuron. 2002; 33: 341-355
        • Fischl B.
        • van der Kouwe A.
        • Destrieux C.
        • et al.
        Automatically parcellating the human cerebral cortex.
        Cereb Cortex. 2004; 14: 11-22
        • Fischl B.
        • Dale A.M.
        Measuring the thickness of the human cerebral cortex from magnetic resonance images.
        Proc Natl Acad Sci U S A. 2000; 97: 11050-11055
        • Segonne F.
        • Pacheco J.
        • Fischl B.
        Geometrically accurate topology-correction of cortical surfaces using nonseparating loops.
        IEEE Trans Med Imaging. 2007; 26: 518-529
        • Jovicich J.
        • Czanner S.
        • Greve D.
        • et al.
        Reliability in multi-site structural MRI studies: effects of gradient non-linearity correction on phantom and human data.
        Neuroimage. 2006; 30: 436-443
        • Fischl B.
        • Sereno M.I.
        • Tootell R.B.
        • et al.
        High-resolution intersubject averaging and a coordinate system for the cortical surface.
        Hum Brain Mapp. 1999; 8: 272-284
        • Owen J.P.
        • Ziv E.
        • Bukshpun P.
        • et al.
        Test-retest reliability of computational network measurements derived from the structural connectome of the human brain.
        Brain Connect. 2013; 3: 160-176
        • van den Heuvel M.P.
        • Sporns O.
        Rich-club organization of the human connectome.
        J Neurosci. 2011; 31: 15775-15786
        • Hu M.L.
        • Zong X.F.
        • Mann J.J.
        • et al.
        A review of the functional and anatomical default mode network in Schizophrenia.
        Neurosci Bull. 2017; 33: 73-84
        • Barry R.L.
        • Williams J.M.
        • Klassen L.M.
        • et al.
        Evaluation of preprocessing steps to compensate for magnetic field distortions due to body movements in BOLD fMRI.
        Magn Reson Imaging. 2010; 28: 235-244
        • Bullmore E.
        • Sporns O.
        Complex brain networks: graph theoretical analysis of structural and functional systems.
        Nat Rev Neurosci. 2009; 10: 186-198
        • Whitfield-Gabrieli S.
        • Ford J.M.
        Default mode network activity and connectivity in psychopathology.
        Annu Rev Clin Psychol. 2012; 8: 49-76
        • Clementz B.A.
        • Sweeney J.A.
        • Hamm J.P.
        • et al.
        Identification of distinct psychosis biotypes using brain-based biomarkers.
        Am J Psychiatry. 2016; 173: 373-384
        • Sun H.
        • Lui S.
        • Yao L.
        • et al.
        Two patterns of white matter abnormalities in medication-naive patients with first-episode schizophrenia revealed by diffusion tensor imaging and cluster analysis.
        JAMA Psychiatry. 2015; 72: 678-686
        • Drysdale A.T.
        • Grosenick L.
        • Downar J.
        • et al.
        Resting-state connectivity biomarkers define neurophysiological subtypes of depression.
        Nat Med. 2017; 23: 28-38
        • Insel T.R.
        • Cuthbert B.N.
        Medicine. Brain disorders? Precisely.
        Science. 2015; 348: 499-500
        • Karalunas S.L.
        • Fair D.
        • Musser E.D.
        • et al.
        Subtyping attention-deficit/hyperactivity disorder using temperament dimensions: toward biologically based nosologic criteria.
        JAMA Psychiatry. 2014; 71: 1015-1024
        • Tamminga C.A.
        • Pearlson G.
        • Keshavan M.
        • et al.
        Bipolar and schizophrenia network for intermediate phenotypes: outcomes across the psychosis continuum.
        Schizophr Bull. 2014; 40: S131-S137
        • Doehrmann O.
        • Ghosh S.S.
        • Polli F.E.
        • et al.
        Predicting treatment response in social anxiety disorder from functional magnetic resonance imaging.
        JAMA Psychiatry. 2013; 70: 87-97
        • Hahn T.
        • Kircher T.
        • Straube B.
        • et al.
        Predicting treatment response to cognitive behavioral therapy in panic disorder with agoraphobia by integrating local neural information.
        JAMA Psychiatry. 2015; 72: 68-74
        • Ma N.
        • Li L.
        • Shu N.
        • et al.
        White matter abnormalities in first-episode, treatment-naive young adults with major depressive disorder.
        Am J Psychiatry. 2007; 164: 823-826
        • Martinez D.
        • Carpenter K.M.
        • Liu F.
        • et al.
        Imaging dopamine transmission in cocaine dependence: link between neurochemistry and response to treatment.
        Am J Psychiatry. 2011; 168: 634-641
        • Whitfield-Gabrieli S.
        • Ghosh S.S.
        • Nieto-Castanon A.
        • et al.
        Brain connectomics predict response to treatment in social anxiety disorder.
        Mol Psychiatry. 2016; 21: 680-685
        • Li F.
        • Lui S.
        • Yao L.
        • et al.
        Longitudinal changes in resting-state cerebral activity in patients with first-episode schizophrenia: a 1-year follow-up functional MR imaging study.
        Radiology. 2016; 279: 867-875
        • Addington J.
        • Heinssen R.
        Prediction and prevention of psychosis in youth at clinical high risk.
        Annu Rev Clin Psychol. 2012; 8: 269-289
        • Beardslee W.R.
        • Brent D.A.
        • Weersing V.R.
        • et al.
        Prevention of depression in at-risk adolescents: longer-term effects.
        JAMA Psychiatry. 2013; 70: 1161-1170
        • Cao H.
        • Chen O.Y.
        • Chung Y.
        • et al.
        Cerebello-thalamo-cortical hyperconnectivity as a state-independent functional neural signature for psychosis prediction and characterization.
        Nat Commun. 2018; 9: 3836
        • Chung Y.
        • Addington J.
        • Bearden C.E.
        • et al.
        Use of machine learning to determine deviance in neuroanatomical maturity associated with future psychosis in youths at clinically high risk.
        JAMA Psychiatry. 2018; 75: 960-968
        • Krueger G.
        • Granziera C.
        • Jack Jr., C.R.
        • et al.
        Effects of MRI scan acceleration on brain volume measurement consistency.
        J Magn Reson Imaging. 2012; 36: 1234-1240
        • Caramanos Z.
        • Fonov V.S.
        • Francis S.J.
        • et al.
        Gradient distortions in MRI: characterizing and correcting for their effects on SIENA-generated measures of brain volume change.
        Neuroimage. 2010; 49: 1601-1611
        • Preboske G.M.
        • Gunter J.L.
        • Ward C.P.
        • et al.
        Common MRI acquisition non-idealities significantly impact the output of the boundary shift integral method of measuring brain atrophy on serial MRI.
        Neuroimage. 2006; 30: 1196-1202
        • Zhang F.
        • Chen G.
        • He M.
        • et al.
        Altered white matter microarchitecture in amyotrophic lateral sclerosis: a voxel-based meta-analysis of diffusion tensor imaging.
        Neuroimage Clin. 2018; 19: 122-129
        • Chen G.
        • Hu X.
        • Li L.
        • et al.
        Disorganization of white matter architecture in major depressive disorder: a meta-analysis of diffusion tensor imaging with tract-based spatial statistics.
        Sci Rep. 2016; 6: 21825
        • Church J.A.
        • Petersen S.E.
        • Schlaggar B.L.
        The "Task B problem" and other considerations in developmental functional neuroimaging.
        Hum Brain Mapp. 2010; 31: 852-862
        • Kruggel F.
        • Turner J.
        • Muftuler L.T.
        • et al.
        Impact of scanner hardware and imaging protocol on image quality and compartment volume precision in the ADNI cohort.
        Neuroimage. 2010; 49: 2123-2133
        • Manoach D.S.
        • Halpern E.F.
        • Kramer T.S.
        • et al.
        Test-retest reliability of a functional MRI working memory paradigm in normal and schizophrenic subjects.
        Am J Psychiatry. 2001; 158: 955-958
        • Zhao N.
        • Yuan L.X.
        • Jia X.Z.
        • et al.
        Intra- and inter-scanner reliability of voxel-wise whole-brain analytic metrics for resting state fMRI.
        Front Neuroinform. 2018; 12: 54
        • Jovicich J.
        • Czanner S.
        • Han X.
        • et al.
        MRI-derived measurements of human subcortical, ventricular and intracranial brain volumes: reliability effects of scan sessions, acquisition sequences, data analyses, scanner upgrade, scanner vendors and field strengths.
        Neuroimage. 2009; 46: 177-192
        • Lee H.
        • Nakamura K.
        • Narayanan S.
        • et al.
        Estimating and accounting for the effect of MRI scanner changes on longitudinal whole-brain volume change measurements.
        Neuroimage. 2019; 184: 555-565
        • Khalili-Mahani N.
        • Rombouts S.A.R.B.
        • van Osch M.J.P.
        • et al.
        Biomarkers, designs, and interpretations of resting-state fMRI in translational pharmacological research: a review of state-of-the-Art, challenges, and opportunities for studying brain chemistry.
        Hum Brain Mapp. 2017; 38: 2276-2325
        • Bocchetta M.
        • Boccardi M.
        • Ganzola R.
        • et al.
        Harmonized benchmark labels of the hippocampus on magnetic resonance: the EADC-ADNI project.
        Alzheimers Dement. 2015; 11: 151-160.e5
        • Pearlson G.
        Multisite collaborations and large databases in psychiatric neuroimaging: advantages, problems, and challenges.
        Schizophr Bull. 2009; 35: 1-2
        • Ma Q.
        • Zhang T.
        • Zanetti M.V.
        • et al.
        Classification of multi-site MR images in the presence of heterogeneity using multi-task learning.
        Neuroimage Clin. 2018; 19: 476-486
        • Wang X.
        • Zhang T.
        • Chaim T.M.
        • et al.
        Classification of MRI under the presence of disease heterogeneity using multi-task learning: application to bipolar disorder.
        Med Image Comput Comput Assist Interv. 2015; 9349: 125-132
        • Stevens J.R.
        An anatomy of Schizophrenia?.
        Arch Gen Psychiatry. 1973; 29: 177
        • Rozycki M.
        • Satterthwaite T.D.
        • Koutsouleris N.
        • et al.
        Multisite machine learning analysis provides a robust structural imaging signature of schizophrenia detectable across diverse patient populations and within individuals.
        Schizophr Bull. 2018; 44: 1035-1044
        • Scarpazza C.
        • De Simone M.
        Voxel-based morphometry: current perspectives.
        Neurosci Neuroecon. 2016; 5: 19-35
        • Salvador R.
        • Radua J.
        • Canales-Rodriguez E.J.
        • et al.
        Evaluation of machine learning algorithms and structural features for optimal MRI-based diagnostic prediction in psychosis.
        PLoS One. 2017; 12: e0175683
        • Steele V.R.
        • Rao V.
        • Calhoun V.D.
        • et al.
        Machine learning of structural magnetic resonance imaging predicts psychopathic traits in adolescent offenders.
        Neuroimage. 2017; 145: 265-273
        • Zhang J.
        • Liu W.
        • Zhang J.
        • et al.
        Distinguishing adolescents with conduct disorder from typically developing youngsters based on pattern classification of brain structural MRI.
        Front Hum Neurosci. 2018; 12: 152
        • Wang D.
        • Li M.
        • Wang M.
        • et al.
        Individual-specific functional connectivity markers track dimensional and categorical features of psychotic illness.
        Mol Psychiatry. 2018; https://doi.org/10.1038/s41380-018-0276-1
        • Huang X.
        • Gong Q.
        • Sweeney J.A.
        • et al.
        Progress in psychoradiology, the clinical application of psychiatric neuroimaging.
        Br J Radiol. 2019; 92: 20181000
        • Abi-Dargham A.
        • Horga G.
        The search for imaging biomarkers in psychiatric disorders.
        Nat Med. 2016; 22: 1248-1255
        • Glasser M.F.
        • Sotiropoulos S.N.
        • Wilson J.A.
        • et al.
        The minimal preprocessing pipelines for the Human Connectome Project.
        Neuroimage. 2013; 80: 105-124
        • Kempton M.J.
        • McGuire P.
        How can neuroimaging facilitate the diagnosis and stratification of patients with psychosis?.
        Eur Neuropsychopharmacol. 2015; 25: 725-732
        • Smieskova R.
        • Allen P.
        • Simon A.
        • et al.
        Different duration of at-risk mental state associated with neurofunctional abnormalities. A multimodal imaging study.
        Hum Brain Mapp. 2012; 33: 2281-2294
        • Fusar-Poli P.
        • Howes O.D.
        • Allen P.
        • et al.
        Abnormal frontostriatal interactions in people with prodromal signs of psychosis: a multimodal imaging study.
        Arch Gen Psychiatry. 2010; 67: 683-691
        • Radua J.
        • Borgwardt S.
        • Crescini A.
        • et al.
        Multimodal meta-analysis of structural and functional brain changes in first episode psychosis and the effects of antipsychotic medication.
        Neurosci Biobehav Rev. 2012; 36: 2325-2333
        • Xiang L.
        • Chen Y.
        • Chang W.
        • et al.
        Deep leaning based multi-modal fusion for fast MR reconstruction.
        IEEE Trans Biomed Eng. 2019; 66: 2105-2114
        • Jauhar S.
        • McCutcheon R.
        • Borgan F.
        • et al.
        The relationship between cortical glutamate and striatal dopamine in first-episode psychosis: a cross-sectional multimodal PET and magnetic resonance spectroscopy imaging study.
        Lancet Psychiatry. 2018; 5: 816-823
        • Howes O.
        • McCutcheon R.
        • Stone J.
        Glutamate and dopamine in schizophrenia: an update for the 21st century.
        J Psychopharmacol. 2015; 29: 97-115
        • Schifani C.
        • Hafizi S.
        • Da Silva T.
        • et al.
        Using molecular imaging to understand early schizophrenia-related psychosis neurochemistry: a review of human studies.
        Int Rev Psychiatry. 2017; 29: 555-566
        • Salavati B.
        • Rajji T.K.
        • Price R.
        • et al.
        Imaging-based neurochemistry in schizophrenia: a systematic review and implications for dysfunctional long-term potentiation.
        Schizophr Bull. 2014; 41: 44-56
        • Vernon A.C.
        • Crum W.R.
        • Lerch J.P.
        • et al.
        Reduced cortical volume and elevated astrocyte density in rats chronically treated with antipsychotic drugs-linking magnetic resonance imaging findings to cellular pathology.
        Biol Psychiatry. 2014; 75: 982-990
        • Vernon A.C.
        • Natesan S.
        • Modo M.
        • et al.
        Effect of chronic antipsychotic treatment on brain structure: a serial magnetic resonance imaging study with ex vivo and postmortem confirmation.
        Biol Psychiatry. 2011; 69: 936-944
        • Dorph-Petersen K.A.
        • Lewis D.A.
        Postmortem structural studies of the thalamus in schizophrenia.
        Schizophr Res. 2017; 180: 28-35
        • Erp T.G.M.V.
        • Hibar D.P.
        • Rasmussen J.M.
        • et al.
        Subcortical brain volume abnormalities in 2028 individuals with schizophrenia and 2540 healthy controls via the ENIGMA consortium.
        Mol Psychiatry. 2016; 21: 547-553
        • Shepherd A.M.
        • Laurens K.R.
        • Matheson S.L.
        • et al.
        Systematic meta-review and quality assessment of the structural brain alterations in schizophrenia.
        Neurosci Biobehav Rev. 2012; 36: 1342-1356
        • Couture S.M.
        • Penn D.L.
        • Losh M.
        • et al.
        Comparison of social cognitive functioning in schizophrenia and high functioning autism: more convergence than divergence.
        Psychol Med. 2010; 40: 569-579
        • Hommer R.E.
        • Swedo S.E.
        Schizophrenia and autism-related disorders.
        Schizophr Bull. 2015; 41: 313-314
        • Pina-Camacho L.
        • Parellada M.
        • Kyriakopoulos M.
        Autism spectrum disorder and schizophrenia: boundaries and uncertainties.
        BJPsych Adv. 2016; 22: 316-324
        • Ivleva E.I.
        • Clementz B.A.
        • Dutcher A.M.
        • et al.
        Brain structure biomarkers in the psychosis biotypes: findings from the bipolar-schizophrenia network for intermediate phenotypes.
        Biol Psychiatry. 2017; 82: 26-39
        • Hill S.K.
        • Reilly J.L.
        • Keefe R.S.
        • et al.
        Neuropsychological impairments in schizophrenia and psychotic bipolar disorder: findings from the Bipolar-Schizophrenia Network on Intermediate Phenotypes (B-SNIP) study.
        Am J Psychiatry. 2013; 170: 1275-1284
        • Wang J.B.
        • Zheng L.J.
        • Cao Q.J.
        • et al.
        Inconsistency in abnormal brain activity across cohorts of ADHD-200 in children with attention deficit hyperactivity disorder.
        Front Neurosci. 2017; 11: 320
        • Consortium H.
        The ADHD-200 consortium: a model to advance the translational potential of neuroimaging in clinical neuroscience.
        Front Syst Neurosci. 2012; 6: 62
        • Skatun K.C.
        • Kaufmann T.
        • Doan N.T.
        • et al.
        Consistent functional connectivity alterations in schizophrenia spectrum disorder: a multisite study.
        Schizophr Bull. 2017; 43: 914-924