Search

Neuroimaging studies point to neurostructural abnormalities in youth with anxiety disorders. Yet, findings are based on small-scale studies, often with small effect sizes, and have limited generalizability and clinical relevance. These issues have prompted a paradigm shift in the field towards highly powered (i.e., big data) individual-level inferences, which are data-driven, transdiagnostic, and neurobiologically informed. Here, we built and validated neurostructural machine learning (ML) models for individual-level inferences based on the largest-ever multi-site neuroimaging sample of youth with anxiety disorders (age: 10-25 years, N=3,343 individuals from 32 global sites), as compiled by three ENIGMA Anxiety Working Groups: Panic Disorder (PD), Generalized Anxiety Disorder (GAD), and Social Anxiety Disorder (SAD). ML classifiers were trained on MRI-derived regional measures of cortical thickness, surface area, and subcortical volumes to classify patients and healthy controls (HC) for each anxiety disorder separately and across disorders (transdiagnostic classification). Modest, yet robust, classification performance was achieved for PD vs. HC (AUC=0.62), but other disorder-specific and transdiagnostic classifications were not significantly different from chance. However, above chance-level transdiagnostic classifications were obtained in exploratory subgroup analyses of male patients vs. male HC, unmedicated patients vs. HC, and patients with low anxiety severity vs. HC (AUC 0.59-0.63). The above chance-level classifications were based on plausible and specific neuroanatomical features in fronto-striato-limbic and temporo-parietal regions. This study provides a realistic estimate of classification performance in a large, ecologically valid, multi-site sample of youth with anxiety disorders, and may as such serve as a benchmark.

Gamma‐hydroxybutyrate acid (GHB) is a recreational drug with a high addictive potential. Severe side effects such as GHB‐induced coma are common and linked to increased emergency room attendances. Task‐based functional‐imaging studies have revealed an association between the regular use of GHB and multiple GHB‐induced comas, and altered neurocognitive function. However the effects of multiple GHB‐induced comas and regular GHB‐use on intrinsic brain connectivity during rest remain unknown. The study population consisted of 23 GHB‐users with ≥4 GHB‐induced comas (GHB‐Coma), 22 GHB‐users who never experienced a GHB‐induced coma (GHB‐NoComa) and 24 polydrug users who never used GHB (No‐GHB). Resting‐state scans were collected to assess resting‐state functional‐connectivity within and between the default mode network (DMN), the bilateral central executive network (CEN) and the salience network (SN). The GHB‐NoComa group showed decreased rsFC of the right CEN with a region in the anterior cingulate cortex (p FWE = 0.048) and decreased rsFC between the right CEN and the DMN (p FWE = 0.048) when compared with the No‐GHB group. These results suggest that regular GHB‐use is associated with decreased rsFC within the right CEN and between the right CEN and the DMN. The presence of multiple GHB‐induced comas is not associated with (additional) alterations in rsFC.

Treatment of psychiatric disorders relies on subjective measures of symptoms to establish diagnoses and lacks an objective way to determine which treatments might work best for an individual patient. To improve the current state-of-the-art and to be able to help a growing number of patients with mental health disorders more efficiently, the discovery of biomarkers predictive of treatment outcome and prognosis is needed. In addition, the application of machine learning methods provides an improvement over the standard group-level analysis approach since it allows for individualized predictions. Machine learning models can also be tested for their generalization capabilities to new patients which would quantify their potential for clinical applicability. In this thesis, these approaches were combined and investigated across a set of different neuropsychiatric disorders. The investigated applications included the prediction of disease course in patients with anxiety disorders, early detection of behavioural frontotemporal dementia in at-risk individuals using structural magnetic resonance imaging (MRI), prediction of deep-brain stimulation treatment-outcome in patients with therapy-resistant obsessive compulsive disorder using structural MRI and prediction of treatment-response for adult and youth patients with posttraumatic stress disorder using resting-state functional MRI scans. Across all studies this thesis showed that machine learning methods combined with neuroimaging data can be utilized to identify biomarkers predictive of future clinical outcomes in neuropsychiatric disorders. Promising as it seems, this can only be the first step for the inclusion of these new approaches into clinical practice as further studies utilizing larger sample sizes are necessary to validate the discovered biomarkers.

Highlights • Preoperative MRI was associated with 12-months DBS treatment outcome in OCD patients. • Larger nucleus accumbens volume was associated with larger clinical improvement. • Machine learning analysis was not successful in predicting clinical improvement., Background Deep brain stimulation (DBS) is a new treatment option for patients with therapy-resistant obsessive–compulsive disorder (OCD). Approximately 60% of patients benefit from DBS, which might be improved if a biomarker could identify patients who are likely to respond. Therefore, we evaluated the use of preoperative structural magnetic resonance imaging (MRI) in predicting treatment outcome for OCD patients on the group- and individual-level. Methods In this retrospective study, we analyzed preoperative MRI data of a large cohort of patients who received DBS for OCD (n = 57). We used voxel-based morphometry to investigate whether grey matter (GM) or white matter (WM) volume surrounding the DBS electrode (nucleus accumbens (NAc), anterior thalamic radiation), and whole-brain GM/WM volume were associated with OCD severity and response status at 12-month follow-up. In addition, we performed machine learning analyses to predict treatment outcome at an individual-level and evaluated its performance using cross-validation. Results Larger preoperative left NAc volume was associated with lower OCD severity at 12-month follow-up (pFWE

Background: Posttraumatic stress disorder (PTSD) is associated with dysregulated neural, cortisol, and cardiac stress reactivity and recovery. This understanding is predominantly based on studies in adults applying emotional-cognitive and trauma-related stimuli inducing negative emotions or perceived threat. Despite large numbers of adolescents with PTSD, few studies are available on neurobiological stress reactivity in this population. Moreover, no previous studies investigated neural reactivity to social-evaluative stress. Objective: To investigate functional brain connectivity, cortisol and cardiac reactivity to acute social-evaluative stress, and additional cortisol measures in trauma-exposed adolescents with and without high PTSD symptoms. Method: A speech preparation task to induce acute social-evaluative stress elicited by anticipatory threat, was used in a subsample of the Amsterdam Born Child and their Development (ABCD) birth cohort, consisting of trauma-exposed adolescents with (n = 20) and without (n = 29) high PTSD symptoms. Psychophysiological interaction analyses were performed to assess group differences in functional connectivity of the hippocampus, mPFC and amygdala during social-evaluative stress and recovery, measured by fMRI. Additionally, perceived stress, heart rate and cortisol stress reactivity and recovery, cortisol awakening response and day curve were compared. Results: The stressor evoked significant changes in heart rate and perceived stress, but not cortisol. The PTSD symptom and control groups differed in functional connectivity between the hippocampus and cerebellum, middle and inferior frontal gyrus, and the mPFC and inferior frontal gyrus during social-evaluative stress versus baseline. Mostly, the same patterns were found during recovery versus baseline. We observed no significant group differences in amygdala connectivity, and cortisol and cardiac measures. Conclusions: Our findings suggest threat processing in response to social-evaluative stress is disrupted in adolescents with PTSD symptoms. Our findings are mainly but not entirely in line with findings in adults with PTSD, which denotes the importance to investigate adolescents with PTSD as a separate population., HIGHLIGHTS Adolescents with PTSD symptoms showed different functional brain connectivity for the hippocampus and medial prefrontal cortex during social-evaluative stress. Amygdala connectivity was not different. Functional connectivity findings seem mostly similar to adults with PTSD.

ObjectiveNo diagnostic biomarkers are available for obsessive-compulsive disorder (OCD). Magnetic resonance imaging (MRI) studies have provided evidence for structural abnormalities in distinct brain regions, but effect sizes are small and have limited clinical relevance. To investigate whether individual patients can be distinguished from healthy controls, we performed multivariate analysis of structural neuroimaging data from the ENIGMA-OCD consortium.MethodWe included 46 data sets with neuroimaging and clinical data from adult (≥18 years) and pediatric (1images from 2,304 OCD patients and 2,068 healthy controls were analyzed using standardized processing to extract regional measures of cortical thickness, surface area and subcortical volume. Machine learning classification performance was tested using cross-validation, and possible effects of clinical variables were investigated by stratification.ResultsClassification performance for OCD versus controls using the complete sample with different classifiers and cross-validation strategies was poor (AUC—0.57 (standard deviation (SD)=0.02;Pcorr=0.19) to 0.62 (SD=0.03;Pcorrcorr>.99) to 0.54 (SD=0.08;Pcorr>.99)). In contrast, good classification performance (>0.8 AUC) was achieved within subgroups of patients split according to their medication status.ConclusionsParcellated structural MRI data do not enable good distinction between patients with OCD and controls. However, classifying subgroups of patients based on medication status enables good identification at the individual subject level. This underlines the need for longitudinal studies on the short- and long-term effects of medication on brain structure.

Randomized controlled trials have shown efficacy of trauma-focused psychotherapies in youth with posttraumatic stress disorder (PTSD), but little is known about the relationship between treatment response and alternations in brain structures associated with PTSD. In this study, we longitudinally examined the association between treatment response and pre-to posttreatment changes in structural magnetic resonance imaging (MRI) scans using a voxel-based morphometry approach. We analyzed MRI scans of 35 patients (ages 8–18 years, 21 female) with PTSD (80%) or partial PTSD (20%) before and after eight weekly sessions of trauma-focused psychotherapy. PTSD severity was assessed longitudinally using the Clinician-Administered PTSD scale for Children and Adolescents to divide participants into responders and non-responders. Group by time interaction analysis showed significant differences in grey-matter volume in the bilateral insula due to volume reductions over time in non-responders compared to responders. Despite the significant group by time interaction, there were no significant group differences at baseline or follow-up. As typical development is associated with insula volume increase, these longitudinal MRI findings suggest that treatment non-response is associated with atypical neurodevelopment of the insula, which may underlie persistence of PTSD in youth. The absence of structural MRI changes in treatment responders, while in need of replication, suggest that successful trauma-focused psychotherapy may not directly normalize brain abnormalities associated with PTSD.

Resting-state functional magnetic resonance imaging (rs-fMRI) data are 4-dimensional volumes (3-space + 1-time) that have been posited to reflect the underlying mechanisms of information exchange between brain regions, thus making it an attractive modality to develop diagnostic biomarkers of brain dysfunction. The enormous success of deep learning in computer vision has sparked recent interest in applying deep learning in neuroimaging. But the dimensionality of rs-fMRI data is too high (~20 M), making it difficult to meaningfully process the data in its raw form for deep learning experiments. It is currently not clear how the data should be engineered to optimally extract the time information, and whether combining different representations of time could provide better results. In this paper, we explored various transformations that retain the full spatial resolution by summarizing the temporal dimension of the rs-fMRI data, therefore making it possible to train a full three-dimensional convolutional neural network (3D-CNN) even on a moderately sized [~2,000 from Autism Brain Imaging Data Exchange (ABIDE)-I and II] data set. These transformations summarize the activity in each voxel of the rs-fMRI or that of the voxel and its neighbors to a single number. For each brain volume, we calculated regional homogeneity, the amplitude of low-frequency fluctuations, the fractional amplitude of low-frequency fluctuations, degree centrality, eigenvector centrality, local functional connectivity density, entropy, voxel-mirrored homotopic connectivity, and auto-correlation lag. We trained the 3D-CNN on a publically available autism dataset to classify the rs-fMRI images as being from individuals with autism spectrum disorder (ASD) or from healthy controls (CON) at an individual level. We attained results competitive on this task for a combined ABIDE-I and II datasets of ~66%. When all summary measures were combined the result was still only as good as that of the best single measure which was regional homogeneity (ReHo). In addition, we also applied the support vector machine (SVM) algorithm on the same dataset and achieved comparable results, suggesting that 3D-CNNs could not learn additional information from these temporal transformations that were more useful to differentiate ASD from CON.

Contains fulltext : 216292.pdf (Publisher’s version ) (Open Access) -Gambling disorder (GD) is characterized by an inability to stop or control gambling behaviour and is often accompanied by gambling-related cognitive distortions. Task-based functional Magnetic Resonance Imaging (fMRI) studies have revealed abnormal responses within the prefrontal and insular cortex, and mesolimbic reward regions. Studies examining resting-state functional connectivity in GD, although limited in number, have so far applied seed-based analysis approaches which revealed altered brain functioning. Here, we applied data-driven Independent Components Analysis to resting-state multi-echo fMRI data. Networks of interest were selected by spatially correlating them to independently derived network templates. Using dual regression, we compared connectivity strength between 20 GD patients and 20 healthy controls within 4 well-known networks (the ventral attention, limbic, frontoparietal control, and default mode network) and an additional basal ganglia component. Compared to controls, GD patients showed increased integration of the right middle insula within the ventral attention network, an area suggested to play an important role in addiction-related drive. Moreover, our findings indicate that gambling-related cognitive distortions - a hallmark of GD - were positively related to stronger integration of the amygdala, medial prefrontal cortex and insula within various resting-state networks. 17 p.

ObjectiveTrauma-focused psychotherapy is the first-line treatment for posttraumatic stress disorder (PTSD) but 30-50% of patients do not benefit sufficiently. We investigated whether structural and resting-state functional magnetic resonance imaging (MRI/rs-fMRI) data could distinguish between treatment responders and non-responders on the group and individual level.MethodsForty-four male veterans with PTSD underwent baseline scanning followed by trauma-focused psychotherapy. Voxel-wise gray matter volumes were extracted from the structural MRI data and resting-state networks (RSNs) were calculated from rs-fMRI data using independent component analysis. Data were used to detect differences between responders and non-responders on the group level using permutation testing, and the single-subject level using Gaussian process classification with cross-validation.ResultsA RSN centered on the bilateral superior frontal gyrus differed between responders and non-responder groups (PFWE < 0.05) while a RSN centered on the pre-supplementary motor area distinguished between responders and non-responders on an individual-level with 81.4% accuracy (P < 0.001, 84.8% sensitivity, 78% specificity and AUC of 0.93). No significant single-subject classification or group differences were observed for gray matter volume.ConclusionsThis proof-of-concept study demonstrates the feasibility of using rs-fMRI to develop neuroimaging biomarkers for treatment response, which could enable personalized treatment of patients with PTSD.