Background: Recent advances in data-driven computational approaches have been helpful in devising tools to objectively diagnose psychiatric disorders. However, current machine learning studies... Show moreBackground: Recent advances in data-driven computational approaches have been helpful in devising tools to objectively diagnose psychiatric disorders. However, current machine learning studies limited to small homogeneous samples, different methodologies, and different imaging collection protocols, limit the ability to directly compare and generalize their results. Here we aimed to classify individuals with PTSD versus controls and assess the generalizability using a large heterogeneous brain datasets from the ENIGMA-PGC PTSD Working group. Methods: We analyzed brain MRI data from 3,477 structural-MRI; 2,495 resting state-fMRI; and 1,952 diffusion-MRI. First, we identified the brain features that best distinguish individuals with PTSD from controls using traditional machine learning methods. Second, we assessed the utility of the denoising variational autoencoder (DVAE) and evaluated its classification performance. Third, we assessed the generalizability and reproducibility of both models using leave-one-site-out cross-validation procedure for each modality. Results: We found lower performance in classifying PTSD vs. controls with data from over 20 sites (60 % test AUC for s-MRI, 59 % for rs-fMRI and 56 % for D-MRI), as compared to other studies run on single-site data. The performance increased when classifying PTSD from HC without trauma history in each modality (75 % AUC). The classification performance remained intact when applying the DVAE framework, which reduced the number of features. Finally, we found that the DVAE framework achieved better generalization to unseen datasets compared with the traditional machine learning frameworks, albeit performance was slightly above chance. Conclusion: These results have the potential to provide a baseline classification performance for PTSD when using large scale neuroimaging datasets. Our findings show that the control group used can heavily affect classification performance. The DVAE framework provided better generalizability for the multi-site data. This may be more significant in clinical practice since the neuroimaging-based diagnostic DVAE classification models are much less site-specific, rendering them more generalizable. Show less
Results of neuroimaging datasets aggregated from multiple sites may be biased by site-specific profiles in participants' demographic and clinical characteristics, as well as MRI acquisition... Show moreResults of neuroimaging datasets aggregated from multiple sites may be biased by site-specific profiles in participants' demographic and clinical characteristics, as well as MRI acquisition protocols and scanning platforms. We compared the impact of four different harmonization methods on results obtained from analyses of cortical thickness data: (1) linear mixed-effects model (LME) that models site-specific random intercepts (LME INT), (2) LME that models both site-specific random intercepts and age-related random slopes (LME INT+ SLP), (3) ComBat, and (4) ComBat with a generalized additive model (ComBat-GAM). Our test case for comparing harmonization methods was cortical thickness data aggregated from 29 sites, which included 1,340 cases with posttraumatic stress disorder (PTSD) (6.2-81.8 years old) and 2,057 trauma-exposed controls without PTSD (6.3-85.2 years old). We found that, compared to the other data harmonization methods, data processed with ComBat-GAM was more sensitive to the detection of significant case-control differences (X-2 (3) = 63.704, p < 0.001) as well as casecontrol differences in age-related cortical thinning (X-2 (3) = 12.082, p = 0.007). Both ComBat and ComBat-GAM outperformed LME methods in detecting sex differences (X-2 (3) = 9.114, p = 0.028) in regional cortical thickness. ComBat-GAM also led to stronger estimates of age-related declines in cortical thickness (corrected p-values < 0.001), stronger estimates of case-related cortical thickness reduction (corrected p-values < 0.001), weaker estimates of age-related declines in cortical thickness in cases than controls (corrected p-values < 0.001), stronger estimates of cortical thickness reduction in females than males (corrected p-values < 0.001), and stronger estimates of cortical thickness reduction in females relative to males in cases than controls (corrected p-values < 0.001). Our results support the use of ComBat-GAM to minimize confounds and increase statistical power when harmonizing data with non-linear effects, and the use of either ComBat or ComBat-GAM for harmonizing data with linear effects. Show less
Background: Depression has been associated with decreased regional grey matter volume, which might partly be explained by an unhealthier lifestyle in depressed individuals which has been ignored by... Show moreBackground: Depression has been associated with decreased regional grey matter volume, which might partly be explained by an unhealthier lifestyle in depressed individuals which has been ignored by most earlier studies. Also, the longitudinal nature of depression, lifestyle and brain structure associations is largely unknown. This study investigates the relationship of depression and lifestyle with brain structure cross-sectionally and longitudinally over up to 9 years.Methods: We used longitudinal structural MRI data of persons with depression and/or anxiety disorders and controls (N-unique participants = 347, N-observations = 609). Cortical thickness of medial orbitofrontal cortex (mOFC), rostral anterior cingulate cortex (rACC) and hippocampal volume were derived using FreeSurfer. Using Generalized Estimating Equations, we investigated associations of depression and lifestyle (Body mass index (BMI), smoking, alcohol consumption, physical activity and sleep duration) with brain structure and change in brain structure over 2 (n = 179) and 9 years (n = 82).Results: Depression status (B =-.053, p = .002) and severity (B =-.002, p = .002) were negatively associated with rACC thickness. mOFC thickness was negatively associated with BMI (B =-.004, p < .001) and positively with moderate alcohol consumption (B = .030, p = .009). All associations were independent of each other. No associations were observed between (change in) depression, disease burden or lifestyle factors with brain change over time.Conclusions: Depressive symptoms and diagnosis were independently associated with thinner rACC, BMI with thinner mOFC, and moderate alcohol consumption with thicker mOFC. No longitudinal associations were observed, suggesting that regional grey matter alterations are a long-term consequence or vulnerability indicator for depression but not dynamically or progressively related to depression course trajectory. Show less
Cremers, H.R.; Demenescu, L.R.; Aleman, A.; Renken, R.; Tol, M.J. van; Wee, N.J.A. van der; ... ; Roelofs, K. 2010
Neuroticism is associated with the experience of negative affect and the development of affective disorders. While evidence exists for a modulatory role of neuroticism on task induced brain... Show moreNeuroticism is associated with the experience of negative affect and the development of affective disorders. While evidence exists for a modulatory role of neuroticism on task induced brain activity, it is unknown how neuroticism affects brain connectivity, especially the crucial coupling between the amygdala and the prefrontal cortex. Here we investigate this relation between functional connectivity and personality in response to negative facial expressions. Sixty healthy control participants, from the Netherlands Study on Depression and Anxiety (NESDA), were scanned during an emotional faces gender decision task. Activity and functional amygdala connectivity (psycho-physiological interaction [PPI]) related to faces of negative emotional valence (angry, fearful and sad) was compared to neutral facial expressions, while neuroticism scores were entered as a regressor. Activity for fearful compared to neutral faces in the dorsomedial prefrontal (dmPFC) cortex was positively correlated with neuroticism scores. PPI analyses revealed that right amygdala-dmPFC connectivity for angry and fearful compared to neutral faces was positively correlated with neuroticism scores. In contrast, left amygdala-anterior cingulate cortex (ACC) connectivity for angry, fearful and sad compared to neutral faces was negatively related to neuroticism levels. DmPFC activity has frequently been associated with self-referential processing in social cognitive tasks. Our results therefore suggest that high neurotic participants display stronger self-referential processing in response to negative emotional faces. Second, in line with previous reports on ACC function, the negative correlation between amygdala-ACC connectivity and neuroticism scores might indicate that those high in neuroticism display diminished control function of the ACC over the amygdala. These connectivity patterns might be associated with vulnerability to developing affective disorders such as depression and anxiety. (C) 2009 Elsevier Inc. All rights reserved. Show less