Diagnostic Neuroimaging and Laboratory Tests

Abstract

The current diagnosis for psychiatric disorders largely relies on clinical symptoms, and therefore the accuracy of diagnosis is often affected by subjective factors. Psychological radiology is a new discipline at the intersection of radiology and psychiatry. It reveals anatomical and functional brain changes in patients with psychiatric disorders by using radiological techniques. Psychoradiology plays an important role in clinical diagnosis, evaluation of treatment response and prognosis, and prediction of illness risk. This chapter discusses the role of psychoradiology in image data acquisition and quality control, diagnostic neuroimaging involving magnetic resonance imaging, emission computed tomography, quantitive eletroencephalography, and magnetoencephalograghy, and laboratory tests for psychiatric disorders.

Introduction

Psychiatric disorders are heterogeneous and have integrations of varied alterations in emotion, cognition, motor activity, and social function, which has increased the challenges in delivering an accurate diagnosis and effective treatment. The primary diagnosis of psychiatric disorder is based on its symptom dependent on the experience and subjective judgment of clinicians. Detection of subtle clinical abnormalities in the early course of psychiatric disorders requires skilled doctors who are highly specialized in mental health services. There is a pressing need to identify objective biomarkers to assist the individual diagnosis, classification of different bio-subtypes, and prediction of disease prognosis. Psychoradiology is an emerging field that applies medical imaging technologies to the analysis of mental health, neurophysiology, and psychiatric disorders [1]. A prominent advantage of psychoradiology is that it relies on imaging data analysis rather than visual inspection of images [2]. Magnetic resonance imaging (MRI), together with other medical imaging techniques such as positron emission tomography and electroencephalogram, is the mainstay of psychoradiology. Using psychoradiological methods, structural and functional cerebral alterations have been identified in psychiatric disorders. With the application of artificial intelligence, psychoradiology may not only advance the understanding of pathological mechanisms of psychiatric disorders but also translate the observations into clinical practice, the findings of which have shown promise as neuroimaging biomarkers for classification, subtyping, and prediction [300,400,5]. The biomarkers found by psychoradiology also showed promises in biologically redefining psychiatric disorders and in optimizing drug treatment selection, thus paving the way to personalized and precision medicines for patients with psychiatric disorders.

In addition to imaging biomarkers, the dysfunction of immune-inflammatory process is also associated with the pathogenesis of psychiatric disorders, suggesting that combined neuroimaging and laboratory biomarkers may be helpful to make connections between molecular basis, neural pathophysiology, and behavioral symptoms in psychiatric disorders, and may boost diagnostic sensitivity and specificity.

In this chapter, we summarize various approaches to identifying biomarkers in psychoradiology. Specifically, we will discuss image data acquisition and quality control, the diagnostic neuroimaging and laboratory tests for psychiatric disorders, and the status of psychoradiology in clinical practice with a specified focus on its translational potentiality. The published clinical applications for different MRI modalities were introduced according to their role in disorder classification, subtyping, and prediction. The use of innovative biomarkers, beyond clinical symptoms, may be a way forward to improve differential diagnoses and individualized patient cares for psychiatric disorders. With the emerging multicenter studies and the development of advanced statistical analyses, considerable interdisciplinary collaborations that involve radiologists, psychiatrists, psychologists, and computer scientists could foster an optimized psychoradiological examinations flow for patients with psychiatric disorders and set up prospective clinical studies to identify and validate the accuracy and reliability of previous findings.

The application of emission computed tomography (ECT) using radioactive tracing technique has considerably advanced understanding of the pathophysiology of psychiatric disorders. ECT’s unique superiorities and developments in the study of psychiatric disorders were described in the third section of this chapter with a focus on imaging of regional cerebral blood flow, neurotransmitter, and neuroreceptor. These advances, together with improvements in scanner resolution, enable ECT to play an increasingly important role in clarifying causes and pathological mechanisms of psychiatric disorders based on molecular level.

As an important complement to fMRI and ECT techniques, eletroencephalography (EEG) are becoming more and more popular because of its low cost and excellent temporal resolution. The fourth section first described the neural basis and measurement of quantitive EEG and then focused on recent advances arising from event-related potential (ERP) studies in identifying objective and reliable biomarkers for the diagnosis of mental disorders. It is promising that considerable physiological evidence accumulated by EEG research and multi-center databases can be used to develop the “gold standard” of neuroelectrophysiological markers for clinical practice, and combine machine learning and multi-modality imaging indicators to further enhance the reliability and effectiveness of biomarkers for mental disorders. Another brain functional imaging tool, magnetoencephalography (MEG) featured with high spatial and temporal resolutions offers a unique view of psychiatry diagnosis. The new insights that MEG studies have provided into searching for biomarkers of psychiatric disorders were discussed in the fifth section. Developing new devices possessing lower prices and higher performance will greatly promote the clinical research and application of MEG.

Laboratory tests is also an effective and indispensable approach to identifying biomarkers of psychiatric disorders. We focally discussed the value of genetic testing and plasma-targeted metabonomic method in diagnosing mental disorders and predicting treatment response and/or adverse events in the last part of this chapter.

Data Acquisition and Quality Control

Anatomical MR Acquisition and Quality Control

T1-Weighted Anatomical MR Acquisition

Three-dimensional fast inverse recovery gradient-echo (GE) sequences, with field of view (FOV) covering the entire head, are useful for brain structure segmentation and cortical surface reconstruction. They are commonly acquired in the sagittal plane with 1-mm isotropic voxel size. Moreover, fat suppression is usually employed to reduce signal from bone marrow and scalp fat. The resolution can be further increased to 0.8 mm for most 3 T clinical scanners by using a 32-channel head coil. It is also recommended to acquire a pair of phase-encoding reversed spin-echo scans to construct a field map for readout distortion correction.

T2-Weighted Anatomical MR Acquisition

It is also recommended to acquire high-resolution anatomical T2 weighted images for psychiatric research. By calculating the T1w/T2w ratio, a myelin content map may be generated across the cortical surface and used to distinguish many architectonic areas non-invasively [6]. High-resolution T2 scans are commonly acquired with a slab excited 3D fast spin echo sequence with the same FOV and resolution of the T1w scan.

Quality Control

Artifacts in anatomical images may bias the outcome of structure segmentation and further statistical inference [7]. Despite head coverage and intensity inhomogeneity, quality control of anatomical images is mainly focused on the visual inspection and evaluation of motion artifacts, such as blurring and ringing (Fig. 8.1). Large motion artifacts have been shown to affect automated segmentation and parcellation techniques implemented in FreeSurfer [7, 8]. However, there is currently no established software package or criteria to determine the threshold of motion artifact. Based on our own experience, the recommended method is to send the data to the automatic segmentation process and then overlay the segmentation results on the original image for visual inspection. It is recommended to exclude volumes from further processing if motion artifacts significantly affect the segmentation.

Fig. 8.1

T1 weighted image with excellent quality (a) and with mild (b), moderate (c), and severe (d) motion artifact. White arrows indicate the ring and blur effect

Diffusion MR Acquisition and Quality Control

Diffusion MR Acquisition

Diffusion MR can measure subtle movement of water molecules within their local tissue environment. Diffusion MR is able to provide two types of metrics as potential biomarkers for psychiatric disorders. Voxel-wise metrics of diffusion properties, like fractional anisotropy (FA), mean diffusivity (MD) and NODDI measures, reflect the local status of tissue microstructure. Global metrics based on deterministic or probabilistic tractography reflect structural connectivity between pairs of brain regions. To fit a simple tensor model, at least one diffusion-weighted shell (b = 1000) and 30 diffusion-encoding directions should be acquired. For more complicated diffusion models or structural connectome estimation, two or more diffusion-weighted shells and 60 diffusion-encoding directions for each shell are recommended. Isotropic voxel size is required for any analysis based on fiber tractography. In order to generate appropriate fieldmap to perform EPI specific distortion correction, two acquisition strategies can be chosen depending on the patient’s tolerance. In the first option, all diffusion volumes are acquired with identical phase encoding direction and one additional b0 with opposite phase encoding direction is acquired to estimate the inhomogeneity fieldmap. In the other option, two volumes are captured with the opposite phase encoding direction for each b-value and diffusion direction. This allows the combination of the two volumes of each diffusion direction and shells into a single volume to increase SNR.

Quality Control

The quality of diffusion-weighted images is evaluated from the following aspects: (1) Geometric distortion. Distortion is unavoidable for EPI sequence but should be within acceptable limits. For images with opposite phase encoding directions, the distortion should also be opposite (Fig. 8.2a); (2) Head motion. Head motion in diffusion MRI is mainly reflected by signal loss in certain slices of a volume, which can easily be detected through visual inspection (Fig. 8.2b); and (3) Signal-to-noise ratio (SNR). The SNR can only be calculated quantitatively and should be consistent within a range for a given b value. If large fluctuation happens, the radio frequency emission and receiver parts of the imaging system should be checked.

Fig. 8.2

Diffusion volumes with opposite phase encoding directions and the distortion free volume after correction (a). Head motion in diffusion MRI is mainly manifested by partial or complete signal loss in certain slices (indicate by white rectangle) of a volume (b)

Geometric distortion. Distortion is unavoidable for EPI sequence but should be within acceptable limits. For images with opposite phase encoding directions, the distortion should also be opposite (Fig. 8.2a) Head motion. Head motion in diffusion MRI is mainly manifested by signal loss in certain slices of a volume, which can easily be detected through visual inspection (Fig. 8.2b) Signal to noise ratio (SNR). The SNR can only be calculated quantitatively and should be constant within a range for a given b value, if large fluctuation happens, the radio frequency emission and receiver parts of the imaging system should be checked (Fig. 8.3).

Fig. 8.3

Average signal intensity of diffusion weighted images at b = 1000 and b = 2000

Functional MR Acquisition and Quality Control

Functional MR Acquisition

Resting-state functional MRI (“rfMRI”) measures the changes in signal intensity caused by blood oxygenation levels associated with cerebral intrinsic activity. It can provide a valuable estimation of functional connectivity between pairs of brain regions. The rfMRI images can be acquired by a gradient echo formation followed by an EPI readout. The spatial resolution is around 3 mm and the temporal resolution is about 2 s for conventional single-shot GE-EPI sequence at 3 T. More than 5 min of scan duration is recommended for stable and effective signal series. The spatial and temporal resolution can be further improved by adopting the multi-band technique, which permits sub-second whole brain coverage at about 2 mm isotropic resolution. This is achieved by acquiring multiple slices in the time of a single EPI echo train with multiple receiver coils and a multi-band excitation pulse sequence. A pair of phase-encoding reversed spin-echo scans with identical geometry as rfMRI volumes are also needed to generate a field map for distortion correction.

Quality Control

The quality control points of fMRI are similar to those of diffusion MR, focusing on image distortion, signal-to-noise ratio, and motion artifacts. Since the TR of fMRI is much smaller than that of diffusion MR, motion artifacts on fMRI are mainly represented by head displacement rather than signal loss. There are several open source software available for quantitative QC, such as fsl, fMRIPrep, etc. In general, data with absolute head displacement greater than 2 mm or absolute rotation angle greater than 5° should be excluded in further analysis (Fig. 8.4).

Fig. 8.4

Quantitative plot of head motion during a scan session. Motions in translation, rotation, and displacement are showed, respectively

Diagnostic Neuroimaging Tests

Classification and Subtyping

Schizophrenia

Although regional brain deficits have been well documented in schizophrenia (SZ), one important question that remains largely unanswered is whether those deficits revealed by MRI could be used as individual biomarkers to discriminate patients from healthy controls. In a study, 163 drug-naïve first-episode SZ (FES) patients and 163 demographically matched healthy controls were recruited. Support Vector Machine was used to explore the potential utility of neuroanatomic measurements in the differentiation of individual patients and healthy controls [9]. The accuracy of the classification was 85.0% (specificity 87.0%, sensitivity 83.0%) for surface area and 81.8% (specificity 85.0%, sensitivity 76.9%) for cortical thickness. Regions contributing to classification accuracy mainly included the default mode, central executive, salience, and visual networks (Fig. 8.5, Table 8.1). These findings suggest that the patterns of illness-related gray matter changes have potential as biomarkers for identifying structural brain alterations in individuals with SZ.

Fig. 8.5

Brain regions (labeled by blue color) contributing to diagnostic classification based on surface area or cortical thickness. L left hemisphere, R right hemisphere

Table 8.1 Brain regions contributing to discrimination between first episode schizophrenia and healthy control groups based on surface area or cortical thickness

Further, SZ is a complex and heterogeneous syndrome. Whether quantitative imaging biomarkers can identify discrete subgroups of patients that might be used to foster personalized medicine remains unclear. In a study, structural MR images of 163 FES and 133 chronical patients with mid-course SZ and a total of 403 healthy controls were recruited [10]. Morphometric measures (cortical thickness, surface area and subcortical structures) were extracted for each subject and then the optimized subtyping results were obtained with non-supervised cluster analysis [11]. Three subgroups of patients defined by distinct patterns of regional cortical and subcortical morphology were identified in FES (Fig. 8.6)[12]. A similar three subgroup pattern was identified in the independent dataset of patients from the multi-site research consortium (Fig. 8.6) [12]. Pooled data also displayed 3-subgroup solution and different morphological alterations (Figs. 8.6 and 8.7) [12]. Similarity of classification patterns between these two patient cohorts suggests that the 3-group typology is relatively stable across the course of illness. Cognitive functions were worse in subgroup 1 with mid-course SZ than those in subgroup 3, supporting clinical differences in the MRI-defined illness subtypes. Regardless of clinical presentation and stage of illness, anatomic MR subgrouping biomarkers can separate neurobiologically distinct subgroups of SZ patients based on brain anatomy features.

Fig. 8.6

Clustering results illustrated by dissimilarity matrices for treatment naïve first episode schizophrenia patients (FES) (a), mid-course schizophrenia (SCZ) patients from Bipolar and Schizophrenia Network for Intermediate Phenotypes (B-SNIP) study (b) and the combined sample (c) [12]

Fig. 8.7

Cortical/subcortical structural alterations in the three subgroups of pooled patients compared with healthy controls (cold color: decreased cortical measurement in patients; asterisk indicates significant alterations for group comparison on subcortical morphology)

Major Depressive Disorder (MDD)

Neuroimaging models can also be used for differential diagnosis and patient subtyping. For instance, we applied Support-Vector-Machine (SVM) to distinguish patients with refractory depressive disorder (RDD) and non-refractory depressive disorder (NDD) and found different classification accuracies with gray matter (69.57%) and white matter (65.22%) (Fig. 8.8) [13]. Our results may serve as an initial step towards the use of biological markers to inform clinical treatment. Another recent study based on resting-state functional connectivity data grouped 458 depression patients into four depression biotypes that were differentially responsive to transcranial magnetic stimulation (Fig. 8.9) [14]. Such models provide typologies for diagnosis and treatment that complement existing typologies based on clinical symptoms.

Fig. 8.8

Most important regions discriminating between RDD and NDD. Gray matter regions (red dots) which showed the highest prognostic value included right middle frontal (1), left inferior parietal (2), right middle temporal (3), right supramarginal (4), left paracentral (5), left middle temporal (6), right paracentral (7), left middle frontal (8), right superior frontal (9), and right superior frontal (10) cortex. White matter regions (blue dots) which showed the highest prognostic value included left lingual (1), left supramarginal (2), left inferior parietal (3), right middle temporal (4), left middle frontal (5), right middle frontal (6), and right superior occipital (7), right inferior frontal (8), right supramarginal (9), and left superior occipital (10) white matter. L left, NDD non-refractory depressive disorder, R right, RDD refractory depressive disorder

Fig. 8.9

Neuroanatomical distribution of dysfunctional connectivity features for four depression biotypes. Nodes are colored to indicate the biotype with the most abnormal connectivity features

Obsessive Compulsive Disorder (OCD)

There are many different clinical manifestations between youths and adults with obsessive-compulsive disorder, including gender distribution, comorbidity, and treatment options. However, the differences in neural correlates between these two populations remain elusive. One coordinate-based meta-analytic investigation provided neuroimaging evidence regarding this issue (Fig. 8.10) [15]. This study demonstrated that, apart from the consistent prefrontal-striatal model in OCD, abnormalities of the visual cortex might be associated with pathology in OCD youths, while deficits in the limbic-cerebellar circuit may play an important role in the pathophysiology of OCD adults. These findings suggest that OCD youths and OCD adults might belong to different clinical subtypes.

Fig. 8.10

The meta-analytic maps of gray matter (GM) changes in patients with obsessive-compulsive disorder (OCD) compared with the healthy controls (HCS). (a) Regions of GM alterations in OCD-youths compared to HCS. (b) Regions of GM alterations in OCD-adults compared to HCS

Post-Traumatic Stress Disorder (PTSD)

PTSD is a traumatic- and stressor-related disorder characterized by re-experiencing, arousal, avoidance symptoms, and negative cognitions and emotion [16]. In recent decades, neuroimaging studies have demonstrated structural and functional brain changes in PTSD that might be associated with the identification and subtyping of PTSD. Based on whole brain neuroanatomy and the support vector machine, a neuroimaging study discriminated adult PTSD patients from non-PTSD controls with 91% accuracy (Fig. 8.11) [17]. Using relevance vector regression, successful prediction of individual PCL scores has been achieved using resting-state functional MRI data in a large group of adult survivors with and without PTSD [18].

Fig. 8.11

Grey matter regions that showed the highest discriminative value for the comparison between PTSD patients and non-PTSD controls. Regions were identified by setting the threshold to the top 30% of the maximum absolute weight vector score. Red indicates higher values in the group of survivors with PTSD, while blue indicates higher values for survivors without PTSD. PTSD posttraumatic stress disorder; non-PTSD, trauma-exposed controls without PTSD

DSM-5 includes an additional subtype of PTSD: PTSD with prominent dissociative symptoms (PTSD+DS), denoting a particular class of patients that exhibit symptoms of depersonalization/derealization [16]. Imaging-based parameters could provide an objective marker for subtyping PTSD. Using Multiclass Gaussian Process Classification, a recent functional neuroimaging study has successfully predicted the diagnosis of PTSD, PTSD + DS, and healthy individuals using individual amygdala functional connectivity maps. The classification accuracy was 83.67% for PTSD + DS patients and 85.37% for PTSD patients (Fig. 8.12) [19].

Fig. 8.12

Whole-brain corrected t test for mALFF resting-state activation, comparing PTSD patients, PTSD + DS patients, and healthy controls. Red circles represent increased mALFF, and blue circles represent decreased mALFF. PTSD posttraumatic stress disorder, PTSD + DS dissociative subtype posttraumatic stress disorder patients, mALFF mean amplitude of low-frequency fluctuations

Attention-Deficit/Hyperactivity Disorder (ADHD)

Attention-deficit/hyperactivity disorder (ADHD) is one of the most common childhood-onset neurodevelopmental disorders characterized by age-inappropriate inattention, hyperactivity, and impulsivity. The diagnosis and subtyping of ADHD are based on clinical and behavioral evaluations for a long time [20]. However, ADHD is a highly heterogeneous disorder with various symptoms and a high psychiatric comorbidity rate [21]. Therefore, the subjective clinical assessments could not discriminate patients with ADHD from typically developing controls precisely.

Imaging-based parameters could provide an objective marker for diagnosing and subtyping ADHD. A recent study used radiomics and machine learning analyses to identify imaging features (including gray matter properties and white matter properties) relevant to ADHD and constructed random forest classifiers based on these features (Table 8.2 and Fig. 8.13). The accuracies for discriminating ADHD patients from controls and ADHD inattentive patients from combined inattentive and hyperactive subtypes were 73.7% and 80.1%, respectively [22].

Table 8.2 Significant features for discriminating ADHD and TDC, ADHD-I, and ADHD-C subtypes

Fig. 8.13

The workflow of radiomics analysis. First, radiomics parameters were extracted from T1-weighted and diffusion-tensor images. Then, the features relevant to diagnosing ADHD were selected by using the all-relevant feature selection approach. Finally, the random forest model was constructed for group and subtype discrimination. DTI diffusion tensor image, T1WI T1-weighted image

Predicting and Monitoring

Predictions of Illness Onset, Relapse, and Long-Term Prognosis

Schizophrenia

The MRI studies suggested that SZ is associated with progressive structural and functional alternations in the prefrontal, temporal, and thalamus (Fig. 8.14), making it possible to predict and monitor brain changes over the different stages of SZ. These regions might involve alterations in the cortico-cerebellar-thalamic-cortical loop (Fig. 8.15). Selective evidence was shown below:

• (a) In individuals with ultra-high risk for psychosis, structural abnormalities of the thalamocortical circuit may be a potential marker for transition to psychosis [23].

• (b) In first-episode SZ, a meta-analysis [24] reviewed 11 structural studies in patients in the first 12 months of illness. Alterations in medial temporal and prefrontal cortical regions, and in the networks that connect them with subcortical areas, are potential neuroanatomical markers for poor symptomatic and functional outcomes.

• (c) Another article [25] systematically reviewed resting-sate functional MRI studies and found SZ-related abnormalities in the cortico-cerebellar-thalamic-cortical networks. However, they did not observe the unique relationships between distinct cognitive domains and specific network abnormalities, suggesting that the disturbances of functional connectivity in the circuitry may contribute to a general cognitive deficit in SZ.

• (d) One longitudinal MRI study [26] found the relapse duration was related to both total cerebral volume and regional (e.g., frontal volume) brain measures, while the number of relapses was uncorrelated to brain measures.

• (e) Studies focusing on patients with chronic SZ showed that gray matter decreased to a greater extent in the thalamus, frontal, and temporal areas, especially in poor-outcome patients [27].

Fig. 8.14

Structural and functional alternations in schizophrenia involving in cortical and subcortical regions. (a) superior view and (b) sagittal view. SFGdor. L left superior frontal gyrus, dorsolateral, ORBmid. L orbital part of left middle frontal gyrus, IFGoperc. L left inferior frontal gyrus, pars opercularis, IFGtriang. L left inferior frontal gyrus, pars triangularis, ORBsup. L orbital part of left superior frontal gyrus, ORBinf. L orbital part of left inferior frontal gyrus, TPOmid. L left temporal pole, middle temporal gyrus, TPOsup. L left temporal pole, superior temporal gyrus, STG. L left superior temporal gyrus, THA. L left thalamus, PCUN. L left precuneus, PCUN. R right precuneus, PCG. L left posterior cingulate gyrus, HIP. L left hippocampus, HIP. R right hippocampus, ITG.L left inferior temporal gyrus, FFG. L left fusiform gyrus, MFG. L left middle frontal gyrus, ACG. R right anterior cingulate gyrus, MTG. L left middle temporal gyrus, PoCG. R right postcentral gyrus

Fig. 8.15

The demonstration of Cortico-Cerebellar-Thalamic-Cortical Circuit

Major Depressive Disorder (MDD)

Previous MRI studies have proposed that neurobiological alterations may contribute to illness onset, relapse, and prognosis of major depressive disorder (MDD). Among the affected regions and networks, the deficits of the default mode network (DMN) have most consistently been identified in structural and functional MRI studies in medication-free depressed patients (Fig. 8.16) [28, 29]. Family risk studies have shown that the function of the DMN may be a resilience factor to MDD [30]. Earlier fMRI findings suggest a dissociation within the DMN, where abnormal functional connectivity within the anterior subnetwork of DMN may be a biomarker for relapse [31]. Spies et al. found that DMN activity changes can be successfully used to predict early treatment response [32].

Fig. 8.16

Medication-free patients with major depressive disorder showed increased cortical thickness in default mode network (DMN) compared to healthy controls. Regions of increased cortical thickness (warm color) in medication-free patients with major depressive disorder than healthy controls. ACC anterior cingulate cortex, L left, PCC posterior cingulate cortex, vmPFC ventromedial prefrontal cortex, R right

Obsessive Compulsive Disorder (OCD)

Recent advances in resting-state functional magnetic resonance imaging (rs-fMRI) have proposed that the pathophysiology of obsessive-compulsive disorder (OCD) encompasses a specific cortico-striato-limbic network. However, the vast majority of these studies published so far have been based on average differences between groups. Whether functional neuroimaging could be used to inform the clinical assessment of individual OCD patients remains unclear. One study applied Support Vector Machine (SVM) to distinguish drug-naive OCD patients from healthy control subjects (HCS) based on four common rs-fMRI parameters including the amplitude of low-frequency fluctuation (ALFF), the fractional amplitude of low-frequency fluctuation (fALFF), regional homogeneity (ReHo), and functional connectivity strength (FCS) [33]. This study showed that the four common rs-fMRI parameters exhibited significant differences in predicting the diagnosis of OCD (Fig. 8.17). In particular, the ALFF showed the highest accuracy, sensitivity and specificity, suggesting its potential in clinical practice to help identify OCD at the individual level.

Fig. 8.17

Receiver operating characteristic (ROC) curves for the assessment of SVM performance using different resting-state functional MRI parameters to distinguish OCD patients from HCS. ALFF amplitude of low frequency fluctuation, fALFF fractional amplitude of low frequency fluctuation, FCS functional connectivity strength, HCS healthy control subjects, MRI magnetic resonance imaging, OCD obsessive-compulsive disorder, ReHo regional homogeneity, SVM support vector machine

Post-Traumatic Stress Disorder (PTSD)

The identification of posttraumatic stress disorder (PTSD) among natural disaster survivors is remarkably challenging, and there are no reliable objective features available to assist clinical diagnosis. One study used 108 brain nodes based on the Schaefer atlas (Fig. 8.18) and their connections to represent each subject’s functional connectivity profile [34]. Multivariate pattern analysis with a relevance vector machine was used to distinguish PTSD patients from trauma-exposed healthy controls (TEHCs). A promising diagnostic accuracy of 89.2% in distinguishing PTSD patients from TEHCs was obtained. The most predictive connections for PTSD were between the default mode network, visual network, somatomotor network, limbic network, and dorsal attention network, which may provide potential features to facilitate clinical diagnosis.

Fig. 8.18

The demonstration of 108 brain nodes associated with eight networks based on the Schaefer atlas. Each node denotes a brain region and further mapped onto the cortical surface using the BrainNet Viewer package (www.nitrc.org/projects/bnv). Associations of these nodes with specific brain networks are shown in different color. L left, R right, VIS visual network, SMN somatomotor network, DAN dorsal attention network, VAN ventral attention network, LIM limbic network, FPC frontoparietal control network, DMN default mode network, Subcor subcortical regions

A sizable minority of veterans who initially experience prominent symptoms of PTSD eventually experience remission [35]. However, the underlying neural correlates of remission are still unclear. One study explored the differences in structural covariance networks between remitted PTSD, current PTSD, and TEHCs [36]. Cortical thickness was assessed for 148 cortical regions using the aparc.a2009s template and interregional partial correlations of cortical thickness across subjects were computed in each group. The centrality of the right subcallosal gyrus, left frontal pole, and right superior frontal sulcus may play a role in remission, current symptoms, and PTSD history, respectively.

Attention-Deficit/Hyperactivity Disorder (ADHD)

Approximately 60% of childhood ADHD patients demonstrate symptom persistence in adulthood [37], 20–30% retain full syndrome, and the rest show either partial or full remission [38, 39]. Simply relying on demographics, clinical information, and neuropsychological data cannot accurately predict the outcome of an ADHD child [40]. Therefore, recent studies sought to explore objective measures of the brain and revealed their potential predictive roles in clinical outcomes. For example, one longitudinal study investigated 205 clinically followed participants with ADHD since childhood by using functional MRI and magnetoencephalography (MEG), and found that persistent inattentive symptoms were associated with atypical connectivity within the default mode network (DMN) and between the DMN and networks pertinent to attention and cognitive control [41] (Fig. 8.19).

Fig. 8.19

(a) Group differences in the functional networks associated with inattention among the persistent, remitted ADHD and never-affected controls. The heat map shows the strength of intra- and internetwork connectivity for each group. Hotter colors indicate stronger connectivity. (b) Network connectivity patterns associated with inattention identified in MEG. The thickness of the lines correspond to the strength of the connectivity. DMN default mode network, FPN frontoparietal network, DAN dorsal attention network, VAN ventral attention network, SMN somatomotor, VN visual network, ADHD attention-deficit/hyperactivity disorder

Predicting and Monitoring Treatment Response

Schizophrenia

The topological organization of the brain network is related to response to antipsychotic treatment in SZ. Both never-treated and antipsychotic-treated SZ patients showed altered nodal centralities in left pre−/postcentral gyri relative to controls; however, decreased nodal centralities in right amygdala/hippocampus and bilateral putamen/caudate were detected in never-treated patients relative to antipsychotic-treated patients and controls, supporting the effect of treatment (Fig. 8.20) [42]. Furthermore, different antipsychotics can act differently in structural networks. Patients treated with risperidone showed higher nodal degree than patients treated with clozapine in the right thalamus, cuneus, and precentral gyrus (Fig. 8.21) [43]; the decreased pattern of FA-weighted white matter network was more prominent in clozapine-treated patients, as indicated by significantly lower values than in risperidone-treated patients, which primarily involved subcortical, prefrontal, and also some occipital regions. The findings above provided a potential method for the selection of antipsychotics (Fig. 8.22) [43].

Fig. 8.20

Group differences in functional topological properties among never-treated schizophrenia patients, chronically treated schizophrenia patients, and healthy controls. Both never-treated schizophrenia patients and chronically treated schizophrenia patient groups showed lower nodal efficiency of left pre−/post-central gyri within functional network; only never-treated patients showed decreased nodal efficiency of bilateral putamen/caudate and right amygdala/hippocampus. PUT putamen, CAU caudate, AMYG amygdala, HIP hippocampus, PRE/POCG pre−/post-central gyri, L left, R right

Fig. 8.21

Group differences in structural topological properties between clozapine-treated and risperidone-treated schizophrenia patients. The direct comparison between the clozapine-treated and risperidone-treated group revealed significant differences in right thalamus, right cuneus, and right precentral gyrus, with higher nodal degrees in the risperidone group. THA thalamus, CUN cuneus, PRECG, pre-central gyrus, R right

Fig. 8.22

Group differences in FA-weighted white matter network between clozapine-treated and risperidone-treated schizophrenia patients. The Post-hoc tests revealed lower values in risperidone-treated patients than clozapine-treated patients with FA-weighted white matter network, which primarily involved subcortical, prefrontal, and also some occipital regions. L left, R right, MFG middle frontal gyrus, IFGtri inferior frontal gyrus, pars triangularis, ORBinf orbital part of inferior frontal gyrus, ORBmid orbital part of middle frontal gyrus, SOG superior occipital gyrus, HIP hippocampus, THA thalamus, CUN cuneus, PRECG pre-central gyrus

Major Depressive Disorder (MDD)

Since depression is associated with substantial morbidity, mortality, and family burden, it is particularly pressing to k for potential predictors for treatment response [40,41,46]. A broad array of studies have focused on neuroimaging biomarkers in MDD after first-line treatments (i.e., evidence-based psychotherapy and pharmacotherapy) and interventions reserved for highly treatment-resistant patients, such as stimulation treatments or unconventional medications [44].

Volumetric reductions in the hippocampus are well-replicated findings in MDD patients versus healthy controls, where studies have identified reduced volume of the hippocampus as a structural imaging predictor for poorer response to antidepressant medication [43,44,45,46,51] (Fig. 8.23). Treatment with antidepressant medication [47, 52] has been reported to be associated with increases in hippocampal volume. A group of studies have investigated the relations between resting-state functional-MRI and treatment response in MDD, and found the associations between treatment (ie, antidepressant medications and transcranial magnetic stimulation) response and functional connectivity regarding frontal and limbic brain regions, subcallosal cingulate cortex, visual recognition circuits, default mode network, and cognitive control network (Fig. 8.24) [53]. The prediction reached an accuracy of 75% (74% positive predictive value, 76% negative predictive value) using task-based connectivity measures between the right superior anterior temporal lobe and the subgenual cingulate cortex [54]. Studies reported consistent findings of altered DTI measures (ie, fractional anisotropy, mean diffusivity and radial diffusivity) in frontolimbic circuits in remitted MDD patients compared with non-remitters.

Fig. 8.23

Reduced volume of the hippocampus as predictors of poor response to antidepressant medication

Fig. 8.24

Resting-state functional networks associated with treatment response in major depressive disorder (MDD). MFG.R right middle frontal gyrus, SFGdor. L left superior frontal gyrus, dorsolateral, IFGtriang.L left inferior frontal gyrus, triangular part, ACG.L left anterior cingulate and paracingulate gyri, PCG.R right posterior cingulate gyrus, HIP.L left hippocampus, PHG.R right parahippocampal gyrus, AMYG.L left amygdala, SPG.L left superior parietal gyrus, PCUN.R right precuneus, STG.L left superior temporal gyrus, MTG.R right middle temporal gyrus

Obsessive Compulsive Disorder (OCD)

Identifying clinically predictive biomarkers that differentiate responders from nonresponders is important for improving treatment efficacy for obsessive-compulsive disorder (OCD). Neuroimaging techniques hold the promise to predict treatment responses in OCD. One review summarized neuroimaging findings about treatment-refractory obsessive-compulsive disorder and imaging markers that predict response to neurotherapeutic interventions [55]. This review indicated that the orbitofrontal cortex (OFC) is a key region for the prediction of conventional psychiatric interventions including pharmacotherapy and psychotherapy, while the anterior cingulate cortex (ACC) is a core neuroanatomical structure to predict the response of invasive neurotherapeutic interventions such as deep brain stimulation (DBS) (Fig. 8.25).

Fig. 8.25

Brain anatomical regions predicting the treatment response for conventional psychiatric interventions (red) and invasive neurotherapeutic interventions (blue) in the obsessive-compulsive disorder. ACC anterior cingulate cortex, OFC orbitofrontal cortex

Post-Traumatic Stress Disorder (PTSD)

Trauma-focused cognitive behavior therapy (TF-CBT) is a recommended treatment strategy for PTSD [56]. Despite the promise of TF-CBT, approximately one-third of patients do not respond to this approach [57]. Therefore, there is a need to examine biomarkers for identifying patients who are not responsive to TF-CBT. One study aimed to identify neural markers of treatment response to TF-CBT when participants are reappraising aversive material and to test the hypotheses regarding neural activation associated with emotional reactivity [58]. The neural circuitry of negative emotion processing (including the anterior cingulate cortex, bilateral amygdala, insula, and hippocampus) was defined using the Automated Anatomical Labeling (AAL) atlas (Fig. 8.26). Lower connectivity of the left amygdala with the anterior cingulate cortex and right insula, and of the left hippocampus with the right amygdala were associated with symptom improvement (Fig. 8.27).

Fig. 8.26

The demonstration of 90 brain nodes based on Automated Anatomical Labeling atlas. Each node denotes a brain region and further mapped onto the cortical surface using the BrainNet Viewer package (www.nitrc.org/projects/bnv). L left, R right

Fig. 8.27

Functional connectivity during emotion reactivity correlated with improvement in post-traumatic stress disorder (PTSD) symptoms. The brain map depicts connectivity between the negative affect networks that is significantly correlated with a change in symptom improvement. sgACC subgenual anterior cingulate cortex, pgACC pregenual anterior cingulate cortex, AMYG amygdala, INS insula, Hip hippocampus

Attention-Deficit/Hyperactivity Disorder (ADHD)

Psychostimulants are generally considered a first-line pharmacologic treatment for ADHD. Among them, methylphenidate (MPH) is currently the most used drug with a high response rate of approximately 70% [59]. Although stimulants are successful in reducing core symptoms of ADHD, early assessment of treatment response is also crucial, which may help check whether the currently used stimulant is effective and if an alternative treatment is needed. Previous structural MRI and PET/SPECT studies revealed psychoradiological characteristics could be potential predictive biomarkers of treatment response in ADHD [60, 61] (Table 8.3 and Fig. 8.28). Few functional MRI studies have been utilized to investigate the relationship between treatment response and brain function. One d-based functional connectivity study found that a good response to MPH was associated with lower functional connectivity between striatal subregions and frontal cortices in ADHD children [62] (Fig. 8.29).

Table 8.3 Predictive biomarkers of good response to MPH in ADHD

Fig. 8.28

Structural predictive biomarkers of good response to methylphenidate (MPH) in ADHD. Increase volume of caudate, nucleus accumbens, cerebellum, corpus callosum, and prefrontal cortex could predict good response to MPH in ADHD children

Fig. 8.29

Significantly decreased striatal functional connectivity in good responders to methylphenidate compared with poor responders of ADHD. The red spheres indicate d regions; the blue spheres indicate effect regions; the blue lines indicate functional connectivity between ds and effect regions. MFG middle frontal gyrus, SFG superior frontal gyrus, ACC anterior cingulate cortex, hippo hippocampus, L left, R right, B bilateral

Diagnostic Neuroimaging of ECT

For the first time, single photon emission computed tomography (SPECT) and positron emission tomography (PET) have achieved the goal of studying brain function in vivo with useful clinical information. Emission computed tomography (ECT) can observe changes in cerebral blood flow, cerebral oxygen consumption, cerebral glucose metabolism, brain protein synthesis, and brain neurotransmitters and receptors, which opens up new approaches for the study of psychiatric diseases and provides a new method for the study of neurobiology (Fig. 8.30).

Fig. 8.30

FDG-PET/CT imaging in a healthy person. A indicates the anterior position

^99mTC-ECD (99mTc-ethylcysteinate dimer) imaging of SZ patients showed hypoperfusion in frontal and temporal lobes (Fig. 8.31). One study using this method for SZ patients has found reduced regional cerebral blood flow (rCBF) perfusion in early-onset SZ group including the precentral and inferior frontal gyri, and reduced cerebral blood flow perfusion in the bilateral postcentral gyrus in late-onset SZ patients [63]. These results revealed a significant difference in brain perfusion between early-onset SZ and late-onset SZ and suggesting that the characteristic changes in regional cerebral blood flow are related to onset-age in SZ.

Fig. 8.31

^99mTC-ECD SPECT/CT imaging in a healthy person (a) and a representative schizophrenia patient (b). Patients with schizophrenia showed increased rCBF perfusion in the right temporal lobe

Dopamine (DA) receptor and transporter dysfunction play an important role in the pathophysiology of neuropsychiatric disorders including SZ. Compared with other psychiatry diseases, the function of DA in SZ is impaired in the substantia nigra striatum and the mesocortical system; the availability of DA is increased in the striatum cortex, and the availability of DA is decreased in the mesocortical pathway [64]. These results indicate various modes of neurotransmitter action in SZ.

The physiological or pathological condition of regional cerebral blood flow can be objectively described through real-time observation of the functional activity of the living brain by using SPECT, to the pathological basis of depression and the related brain function (Fig. 8.32). Most of the research results showed that depressed patients had abnormal blood perfusion in the frontal and temporal lobes, among which the anterior and ventral cingulate cortex of the knee of the corpus callosum were the most frequently reported [65]. Neuroreceptor imaging is also a hot topic in depression research. Studies have shown that compared with the placebo group, depressive symptoms in patients treated with Gonadotropin-releasing hormone agonist (GnRHa) are correlated with the increase of 5-HT (5-hydroxytryptamine) transporter binding in the cortical brain [66].

Fig. 8.32

^99mTC-ECD SPECT/CT imaging in a healthy person (a) and a representative depression patient (b). Red arrow indicated reduced rCBF in the left frontal lobe in depressed patients

To sum up, SPECT and PET cerebral imaging can observe the local energy metabolism, oxygen metabolism, brain blood perfusion, and the change of various neurotransmitters and receptors using various physiological tracer, which allows brain function research from multiple levels and multiple perspectives (Fig. 8.33). Until now, SPECT and PET have been widely used in psychiatric research of SZ, depression, anxiety disorder, personality disorder, and mania [67]. It is believed that with the improvement and popularization of SPECT and PET technology, the research of SPECT and PET in psychiatry will be further deepened, and the etiology and pathophysiology of psychiatric diseases can be better understood.

Fig. 8.33

PET imaging using the norepinephrine transporter (NET)-selective (S,S)-O-[11C] methylreboxetine in healthy controls (a) and attention-deficit/hyperactivity disorder (ADHD) patients (b). DVR distribution volume ratio

Diagnostic Neuroimaging of Quantitive EEG

With the advanced technology development of electrical engineering, computer, and signal processing in the 1980s, EEG and event-related brain potentials (ERPs) were becoming popular in both basic and clinical studies. EEG is a direct measurement of neuronal activity with a sampling frequency as high as dozens of kHz, resulting in extremely high temporal resolution. Compared to other neuroimaging techniques, which are used to investigate brain function (e.g., PET and fMRI), EEG is much cheaper, since the equipment is relatively cheaper and the disposable supplies required for experimenting are less. Moreover, in contrast to other electrophysiological techniques, the EEG technique is noninvasive. Thus, EEG recordings can be collected from subjects easily and repeatedly. In addition, EEG and ERP techniques can provide excellent temporal resolution. For these reasons, EEG recordings are becoming more and more popular as an important complement to PET and fMRI techniques.

Mental disorder is a syndrome characterized by clinically significant abnormalities in cognitive function, emotion regulation, and social behavior, which reflects an individual’s obstacles in the development of psychological, biological, or social functions. A mass scientific epidemic survey found that there are 970 million people with mental disorders in the world. From 2010 to 2030, the cumulative loss of global economic output caused by mental illness is expected to reach US$16 trillion, equivalent to 25% of global GDP in 2010. However, the pathological mechanism of mental disorders is complex and the symptoms are diverse, which brings many difficulties to basic research and clinical practice in this field. In addition, traditional clinical diagnosis methods are also affected by the subjective factors of clinicians. It is difficult to illustrate the pathological mechanism of mental disorders only from the view of an objective evaluation of the symptoms of mental disorders. Finding objective biomarkers of mental disorders is an important way to solve these problems.

Biomarkers are objective indicators that can be repeatedly measured and reflect the medical status of patients. They can help clinicians, epidemiologists, and scientists to study various diseases and determine treatment goals by confirming the disease and tracking the course of the disease, as well as evaluating the efficacy of drugs. Currently, the most concerning mental disorder biomarkers mainly come from non-invasive neuroelectrophysiology or imaging techniques such as EEG and MRI. Researchers use the above tools to find the neural structure and functional state associated with specific mental disorders to find biologically based objective indicators or biomarkers. Compared with MRI, EEG equipment is cheaper and is easier to generalize and can sensitively and effectively evaluate the various stages of cognitive processing. The high time resolution also makes EEG widely used in cognitive neuroscience, and this type of research has accumulated considerable physiological evidence for mental disorders.

Neural Basis and Measurement of Quantitive EEG

EEG is defined as the electrical activity, which is generated by the firing of neurons within the human brain, and normally recorded at the brain scalp [68]. It is commonly recognized that the EEG originates from summed synchronized synaptic activities in populations of cortical neurons, with the main contribution from pyramidal cells. The information transmitted and processed in these neurons is realized by electrochemical signaling, via specialized connections called synapses. The relationship between the cortical processing within the brain and the electrical activity measured at the scalp is not formatted in a straightforward manner. The pyramidal neurons, which are the main composition of the cerebral cortex and close to the scalp, are highly polarized with the major orientation perpendicular to the cortical surface. Therefore, the cortical pyramidal neurons are the primary generators of the scalp EEG.

To obtain high quality EEG signals, the EEG measurement system should consist of several necessary elements, including electrodes with conductive media, amplifiers with filters, an analog-to-digital (A/D) converter, and a recording device. Specifically, the microvolt signals recorded from electrodes at the scalp were transformed by amplifiers into signals within a proper range of voltage. Then, signals were transformed by a converter from an analog format to a digital one and finally stored via the recording device. In 1958, a standard protocol for EEG electrode placement was developed by the American Electroencephalographic Society [69, 70]. It is called the International 10–20 System, which is widely used for defining electrode placement. This system contains standardized locations of 75 electrodes on the scalp (Fig. 8.34). The electrodes are placed at 10% and 20% points along lines of latitude and longitude, and this is the reason why the system is called the 10–20 system.

Fig. 8.34

International 10–20 System for electrode placement

The recordings of the spontaneous electrical activity, called spontaneous EEG activity, exhibit certain characteristic waveforms that dominate in a wide range of frequencies and are normally applied in various clinical diagnoses. The spontaneous EEG activity is normally classified according to the frequency bands. Empirically, spontaneous EEG activity is divided into five frequency bands (Fig. 8.35): delta (δ, <4 Hz), theta (θ, 4–8 Hz), alpha (α, 8–13 Hz), beta (β, 13–30 Hz), and gamma (γ, >30 Hz). The rhythmic activity at each of these frequency bands characterizes by a certain distribution over the scalp and a certain biological significance.

Fig. 8.35

Rhythmic EEG activity patterns. The rhythmic EEG activity is often divided into bands by frequency. Empirically, spontaneous EEG activity is divided into five frequency bands: delta (δ, < 4 Hz), theta (θ, 4–8 Hz), alpha (α, 8–13 Hz), beta (β, 13–30 Hz), and gamma (γ, > 30 Hz)

When a stimulus associated with a specific sensory, cognitive, or motor event occurs, the neural responses disturb and can be extracted from the spontaneous EEG activity using a simple averaging technique. Specifically, the averaged responses are called ERPs to denote the fact that they are electrical potentials and related to the target events. Some researchers showed that ERPs appear to be an additional activity and independent of the ongoing spontaneous EEG activity [68]. In most cases, the magnitude of ERPs is often several factors smaller than the magnitude of the background EEG activity. Therefore, the identification of ERPs relies on signal processing methods for enhancing the SNR by means of the across-trial averaging in the time domain. And the averaged ERPs often comprise monophasic deflections which are characterized by their polarity, latency, amplitude, and scalp distribution.

Common Auditory and Visual ERP Components

The presentation of an auditory stimulus evokes a set of brainstem auditory evoked potentials (BAEPs), which reflects the flow of information from the cochlea through the brainstem and into the thalamus (Fig. 8.36). These auditory brainstem responses are typically labeled with Roman numerals (waves I–VI) and probably represent the activation of acoustic nerve (wave I), cochlear nuclear nuclei (wave II), superior olives (wave III), lateral lemniscus tracts and nuclei (wave IV), and inferior colliculi (wave V). Wave VI presumably arises from the medial geniculate body but is not considered clinically useful. BAEPs are highly automatic and relatively unaffected by sleep and anesthesia. Therefore, BAEPs can be used to assess the integrity of the auditory pathways, especially for newborns and infants. It is an extraordinary tool for the early identification of hearing impairment or hearing loss, aiding in early intervention.

Fig. 8.36

Typical sequence of auditory evoked components. The waveform elicited by an auditory stimulus is shown over a period of time to demonstrate the auditory brainstem responses (waves I–VI), the mid-latency responses, and the long-latency responses

The BAEPs are followed by the mid-latency responses potentials (such as P1, N1, and P2), which probably arise at least in part from the medial geniculate nucleus and the primary auditory cortex, and the long-latency responses potentials (such as N2) [71]. The auditory N1 wave with a peak latency between 50 and 150 ms is one of the most prominent components during an auditory task. The auditory N1 is sensitive to attention. It has several distinct subcomponents that sum together to form the N1 peak, including a set of specific subcomponents that associated with the physical and temporal aspects of the stimulus and general state of the subject and a set of subcomponents. When deviant stimuli are occasionally presented within the repetitive train, a larger N2 amplitude is evoked in response to the deviants. This effect can be divided into three subcomponents—N2a, N2b, and N2c. The N2a is an automatic effect that occurs mainly for auditory mismatches, even if they are task-irrelevant, which is more commonly known as the auditory mismatch negativity (MMN). Specifically, MMN is generated by a relatively automatic response to an auditory stimulus that differs from the preceding stimuli, that is, an infrequent change in homogeneous repetitive stimuli even in the absence of attention.

Visual evoked potentials include P1 and N1 (Fig. 8.37). Typically, the first major visual ERP component is the P1 wave. Visual P1 is maximal at lateral occipital electrode sites and onsets 60–90 ms poststimulus with a peak between 100 and 130 ms. The P1 wave is sensitive to variations in stimulus parameters and can be modulated by selective attention as well as by the individual’s state of arousal [72]. It is believed that the P1 component is mainly generated in extrastriate visual areas. Following the P1 wave is the visual N1 wave, which is highly refractory. Like auditory N1, there are several visual N1 subcomponents. The earliest N1 subcomponent peaks at 100–150 ms poststimulus at anterior electrode sites, and there appear to be at least two posterior N1 components that peak at 150–200 ms poststimulus, arising from parietal cortex and lateral occipital cortex, respectively. It has shown that all three N1 subcomponents are influenced by spatial attention.

Fig. 8.37

Typical visual evoked components, including P1, N1, P2, N2, and P3 components

Diagnostic Performance of ERP Measures

EEG measures the electric potential generated by the synchronous firing of neurons by placing electrodes on multiple positions on the scalp. It can detect the electric potential generated by the cerebral cortex and some subcortical areas, thereby providing the brain’s neural activity in milliseconds. In the EEG research of mental disorders, ERPs are mostly measured as outcome variables. ERP is calculated from the activity of EEG in some data periods and makes full use of the inherent high time resolution advantages of EEG technology. The selected data period is time-locked to some event of interest, such as an external stimulus or the subject’s response, and the time-locked data period is averaged to produce a separate ERP waveform. In addition, EEG equipment is inexpensive and it is easy to operate, and its cost is much lower than that of MRI and MEG. It can analyze the electrophysiological characteristics of the brain economically and effectively. This also makes EEG expected to be widely used in the basic research of neuroscience and the clinical practice of individual medicine. In recent years, in order to more accurately assess the cognitive function characteristics of patients with mental disorders and high-risk groups, researchers have combined EEG technology and meta-analysis to carry out neuro-electrophysiological evidence-based medicine research on mental disorders. It also provides more objective and reliable biomarkers for the establishment of a new diagnostic framework for mental disorders.

The neuro-electrophysiological research of mental disorders has made considerable achievements in identifying objective biomarkers of SZ, anxiety, and obsessive-compulsive disorder. The cognitive deficits of patients and high-risk populations provide reliable empirical data. However, the existing evidence-based medicine research on mental disorders has different perspectives, and most studies only explore the components of a single type of ERP in a certain mental disorder. And the abnormal conditions in the EEG frequency band also make it difficult to form a consistent view of cognitive deficits in patients with mental disorders in theoretical construction and clinical practice. The time domain and frequency domain characteristics of EEG and the cognitive activities reflected in EEG-like research results not only help to form a systematic understanding of the cognitive impairment of specific mental disorders, but also help to compare the cognitive deficits of various mental disorders horizontally, and then establish a new mental disorder diagnostic framework and disease clustering. Recent neuro-electrophysiological research integrating evidence-based medicine has provided an important way for this and has achieved considerable research results. However, the existing research results are still controversial and difficult to reach a consensus.

The ERP Components as Potential Endophenotypes

An endophenotype is defined as an internal phenotype that is discoverable using biochemical or microscopic tests. Endophenotypes form the causal links between genetic influences and overt phenotypic expression and are therefore key in the understanding of the underlying biological mechanisms of disease risk and expression. Unlike biological markers, which may not be heritable, criteria for an endophenotype include heritability of the marker as well as a higher prevalence of the marker in non-affected family members relative to the general population. The strategy for validating endophenotypes for mental disorders first involves identifying deficits in patients, followed by exploring evidence of heritability in unaffected relatives. For example, SZ is a complex disease that involves a combination of numerous genetic and environmental factors. It has been consistently demonstrated that patients with SZ exhibit deficits in sensory and cognitive processing. Many studies have aimed to further explore these deficits using EEG to measure (ERPs. Three specific components, the P50, P300, and mismatch negativity (MMN) have been shown to reliably differ between patients with SZ and healthy controls in response to auditory stimuli (Fig. 8.38).

Fig. 8.38

Example component waveforms. (a) Example auditory-evoked response from a healthy control (left) and patient with schizophrenia (right). Arrows mark the location of the P50 wave for the conditioning stimulus and identical test stimulus. T/C indicates the test-to-conditioning ratio for each subject. The P50 response to the second stimulus is attenuated for the healthy control subject, but not for patient with schizophrenia. (b) Grand average waveforms for 38 healthy control subjects (top) and 52 patients with schizophrenia (bottom) in response to an infrequent auditory stimulus to which participants made a button press. As marked by arrows, patients with schizophrenia exhibit smaller P300 amplitudes than healthy controls. (c) Grand average mismatch negativity (MMN) response to an auditory pitchdeviant stimulus for 20 healthy control subjects (top) and 19 patients with schizophrenia (bottom). As marked by arrows, patients with schizophrenia exhibit smaller MMN amplitudes than healthy controls. (From Turetsky et al. [89]; with permission)

The P50 ERP component is a positive deflection occurring approximately 50 ms after stimulus onset and generally shows a decrease in amplitude to repeated stimuli, termed P50 suppression. Specifically, when two identical stimuli are presented successively, a decrease in P50 amplitude to the second stimulus is generally found. P50 is mostly thought to reflect a sensory gating mechanism or filtering of redundant stimuli. The P300 component is a positive deflection occurring between 250 and 500 ms after stimulus onset and is thought to reflect attentional processes. This component is often recorded using an oddball paradigm, in which the subject is presented with frequent and infrequent stimuli and asked to respond to the infrequent stimuli. An increase in amplitude to the infrequent relative to the frequent stimuli is generally found. It has been demonstrated that P300 amplitude is related to the number of attentional resources devoted to the task, whereas P300 latency indexes stimulus classification speed. The oddball paradigm also induces the MMN component that occurs in response to deviant stimuli. This response is elicited by auditory tones differing in a variety of perceptual features, such as frequency or duration. The MMN generally occurs 150–250 ms after the onset of the deviant stimulus is maximal over frontocentral scalp locations and is related to the degree of deviance. Unlike the P300, the MMN is elicited even in the absence of attention. The MMN reflects automatic auditory processing, perceptual discrimination ability, and sensory memory.

To sum up, the SZ patients showed a change of perception from early to late defects in processing, cognitive control, semantic processing, affect, and social cognition. Many studies have demonstrated robust deficits in P50 suppression, as reflected by larger P50 ratios for patients with SZ relative to healthy controls. Attention deficit/hyperactivity disorder has cognitive control defects from early to late, anxiety and obsessive-compulsive disorder have early cognitive control defects, and autism spectrum disorders have early perceptual processing and social cognitive deficits. Deficits in MMN generation are robust in patients with most of the mental disorders. In addition, the decrease in P300 amplitude and the increase in latency, which index attention allocation and processing speed, span multiple diagnostic types, indicating that this component may reflect the generality of cognitive deficits of mental disorders, and may be related to high-level cross-diagnostic factors. In the future, existing data sets and multi-center databases can be used to develop the “gold standard” of neuroelectrophysiological markers for clinical practice, and integrate machine learning and multi-modality. Imaging indicators to further enhance the reliability and effectiveness of biomarkers for mental disorders.

Diagnostic Neuroimaging of Magnetoencephalography

As a noninvasive functional imaging tool, magnetoencephalography (MEG) is featured by a fine combination of high spatial and temporal resolutions, offering a unique view in psychiatry diagnosis. Although MEG and EEG share the same origin of postsynaptic potentials, there are important differences due to the physics of detected signals (Table 8.4 for a detailed comparison between EEG and MEG). Since the magnetic fields are hardly affected by the skull or other brain tissues, MEG has better sensitivity in event-related responses and brain oscillations detection compared with EEG.

Table 8.4 Comparison between EEG and MEG

Apart from clinically detecting abnormal neuronal activities such as epilepsy, MEG has also been used in the search for biomarkers in psychiatry. It has been found that autism spectrum disorders (ASD) patients show a significantly delayed MEG auditory response around 100 ms (M100) in right-hemispheres, indicating impaired auditory processing in its early stage. Detailed MEG results also revealed a frequency-dependent latency shared by ASD patients (Fig. 8.39) [73]. As for SZ, the MEG recordings of wide-band neural oscillation provide the detection of gamma band or even high gamma band abnormalities apart from 40-Hz frequency range commonly used with EEG (Fig. 8.40) [74]. Also, by investigating the alteration of the resting-state network across brain regions and rhythms, MEG could potentially act as research tool for psychiatric patients such as Alzheimer’s disease (AD) [75]. Finally, recent studies also show the possibility of combing brain stimulation such as transcranial electrical stimulation (tES) with MEG [76], providing concurrent monitoring and evaluation of the entrainment of brain oscillations.

Fig. 8.39

The right hemisphere M100 response latencies in ASD patients and typical development (TD) subjects. It was found ASD patients shared a delayed of M100 response across frequencies, while the latency delay of 500 Hz was the most evident. (Error bars represent standard error of the mean, while “*” shows significant t-test result with p < 0.05 for Group Frequency ANOVA test.) (From Roberts et al. [73]; with permission)

Fig. 8.40

MEG topographies of schizophrenia patients (left panels) show extent differences with controls (middle panels) in response to upright Mooney faces both in lower (25–60 Hz) and higher (60–140 Hz) gamma band. The right panels show the statistical difference where red denotes stronger activation for controls compared to patients, whereas blue represents stronger activation in patients relative to controls (FDR corrected, q < 0.05) (From Grützner et al. [74]; Creative Commons Attribution license [CC-BY])

However, the research of MEG in the diagnosis of psychiatric diseases is still in its early stage with limited studies due to the high costs of commercial systems based on superconducting quantum interference devices (SQUID). Recent progress in optically-pumped magnetometers (OPM) has proposed a new generation of non-cryogenic MEG with a much lower price [77]. The other important advantage of OPM over SQUID is the reduced distance between sensor and scalp (Fig. 8.41), which brings several times increase in signal magnitude (Fig. 8.42) [78, 79]. The OPM–MEG could also be developed as a wearable device that offers much more flexibility for clinical diagnosis and other applications [77].

Fig. 8.41

The reconstructed SQUID (left) and OPM (right) sensor array positions relative to the head. By getting rid of the Dewar in the SQUID system, the OPM sensors can get much closer to the scalp, with the distance between sensitive point and scalp less than 1 mm. SQUID superconducting quantum interference devices, OPM superconducting quantum interference devices, MEG magnetoencephalography

Fig. 8.42

Comparison of the magnitudes between OPM and SQUID in detecting evoked responses. (a) Measurements of the median nerve stimulation responses (in fT) of both OPM (black line) and SQUID (read line). (b) The OPM response shares the similar amplitude and profile as SQUID time course multiplied by four. SQUID superconducting quantum interference devices, OPM superconducting quantum interference devices (From Boto et al. [79]; Attribution 4.0 International [CC BY 4.0])

Diagnostic Laboratory Tests

Genetic testing may be useful in various clinical settings. Genome -wide association studies (GWAS) evolved as a key tool to identify genetic risk variants related to complex diseases. For example, rare or de novo copy number variants and SNPs are common in individuals with SZ, autism spectrum disorder, and other psychiatric disorders. Microdeletion 22qll.l syndrome is typically caused by a recurrent 3 MB deletion of 40 genes, including TBX1. 20 to 50% of patients with this deletion develop ASD [80], and the deletion is also found in approximately 1% of people with SZ, bipolar disorder, and idiopathic Parkinson disease [81, 82]. And approximately 2.5% of SZ patients carry one of the associated copy number variants [83]. Thus, the identification of de novo CNVs could be useful in the management of severe psychiatric disorders.

Besides genetic testing, the plasma-targeted metabonomic method has also been used for the diagnosis of mental disorders. Pan et al. simultaneously quantified the levels of 19 plasma metabolites involved in the GABAergic, catecholaminergic, and serotonergic neurotransmitter systems in 50 first-episode, antidepressant drug-naïve MDD subjects by using a GC-MS coupled with LC-MS/MS-based targeted metabolomics approach, they found a panel of four candidate plasma metabolite biomarkers (GABA, dopamine, tyramine, kynurenine) could distinguish MDD subjects from health controls with an AUC of 0.968 and 0.953 in the training and testing set, respectively [84].

The pharmacological treatment of psychiatric disorders has been severely hampered by large inter-individual variations in terms of drug response and/or side effects, often leading to painful, frustrating, and inefficient trial-and-error-based changes in treatment regimens. Such variations are to a large extent due to genetic factors, with an estimated heritability of 0.6–0.8 [85]. Thus, numerous studies attempted to detect gene variants associated with individual differences in drug responsiveness or side effect. Recent GWASs have suggested that the proportion of variance in antidepressant response explained by common genetic variation is as high as 42% [86], and pharmacogenetic (ie, evaluating a single gene) and pharmacogenomic (ie, evaluating multiple genes) testing have shown promise for informing the selection of antidepressants [87, 88]. Pharmacogenetic and pharmacogenomic testing have been developed to predict which medication may yield the best treatment response and/or the lowest risk of adverse events for a given individual.