Computational neuroimaging strategies for single patient predictions.

Neuroimaging increasingly exploits machine learning techniques in an attempt to achieve clinically relevant single-subject predictions. An alternative to machine learning, which tries to establish predictive links between features of the observed data and clinical variables, is the deployment of computational models for inferring on the (patho)physiological and cognitive mechanisms that generate behavioural and neuroimaging responses. This paper discusses the rationale behind a computational approach to neuroimaging-based single-subject inference, focusing on its potential for characterising disease mechanisms in individual subjects and mapping these characterisations to clinical predictions. Following an overview of two main approaches - Bayesian model selection and generative embedding - which can link computational models to individual predictions, we review how these methods accommodate heterogeneity in psychiatric and neurological spectrum disorders, help avoid erroneous interpretations of neuroimaging data, and establish a link between a mechanistic, model-based approach and the statistical perspectives afforded by machine learning.

Levothyroxine effects on depressive symptoms and limbic glucose metabolism in bipolar disorder: a randomized, placebo-controlled positron emission tomography study.

Adding supraphysiologic doses of levothyroxine (L-T4) to standard treatment for bipolar depression shows promise, but the mechanisms underlying clinical improvement are unknown. In a previous pilot study, L-T4 treatment reduced depression scores and activity within the anterior limbic network. Here we extended this work in a randomized, double-blind, placebo-controlled study of patients with bipolar depression. Cerebral glucose metabolism was assessed with positron emission tomography and [F-18]fluorodeoxyglucose before and after 6 weeks of treatment with L-T4 (n=15) or placebo (n=10) in 12 volumes of interest (VOIs): the bilateral thalamus, amygdala, hippocampus, dorsal striatum and ventral striatum, and midline cerebellar vermis and subgenual cingulate cortex. Radioactivity in the VOIs, normalized to whole-brain radioactivity was taken as a surrogate index of glucose metabolism, and markers of thyroid function were assayed. Changes in brain activity and their association with clinical response were assessed using statistical parametric mapping. Adjunctive L-T4 treatment produced a significant decline in depression scores during the 6-week treatment. In patients treated with L-T4, we found a significant decrease in regional activity at P<0.05 after Bonferroni correction in the left thalamus, right amygdala, right hippocampus, left ventral striatum and the right dorsal striatum. Decreases in the left thalamus, left dorsal striatum and the subgenual cingulate were correlated with a reduction in depression scores (P<0.05 after Bonferroni correction). Placebo treatment was associated with a significant decrease in activity only in the right amygdala, and no region had a change in activity that was correlated with change in depression scores. The groups differed significantly in the relationship between the changes in depression scores and in activity in the thalamus bilaterally and the left ventral striatum. The findings provide evidence that administration of supraphysiologic thyroid hormone improves depressive symptoms in patients with bipolar disorder by modulating function in components of the anterior limbic network.

Anticipating agoraphobic situations: the neural correlates of panic disorder with agoraphobia.

BACKGROUND: Panic disorder with agoraphobia is characterized by panic attacks and anxiety in situations where escape might be difficult. However, neuroimaging studies specifically focusing on agoraphobia are rare. Here we used functional magnetic resonance imaging (fMRI) with disorder-specific stimuli to investigate the neural substrates of agoraphobia. METHOD: We compared the neural activations of 72 patients suffering from panic disorder with agoraphobia with 72 matched healthy control subjects in a 3-T fMRI study. To isolate agoraphobia-specific alterations we tested the effects of the anticipation and perception of an agoraphobia-specific stimulus set. During fMRI, 48 agoraphobia-specific and 48 neutral pictures were randomly presented with and without anticipatory stimulus indicating the content of the subsequent pictures (Westphal paradigm). RESULTS: During the anticipation of agoraphobia-specific pictures, stronger activations were found in the bilateral ventral striatum and left insula in patients compared with controls. There were no group differences during the perception phase of agoraphobia-specific pictures. CONCLUSIONS: This study revealed stronger region-specific activations in patients suffering from panic disorder with agoraphobia in anticipation of agoraphobia-specific stimuli. Patients seem to process these stimuli more intensively based on individual salience. Hyperactivation of the ventral striatum and insula when anticipating agoraphobia-specific situations might be a central neurofunctional correlate of agoraphobia. Knowledge about the neural correlates of anticipatory and perceptual processes regarding agoraphobic situations will help to optimize and evaluate treatments, such as exposure therapy, in patients with panic disorder and agoraphobia.

Antibodies in autoimmune thyroiditis affect glucose metabolism of anterior cingulate.

BACKGROUND: Hypothyroidism induced by an autoimmune process is associated with neuropsychiatric symptoms and metabolic abnormalities in the brain. The aim of this study was to examine the relationship between autoimmune thyroiditis and regional brain function in hypothyroid patients. METHODS: Cerebral glucose metabolism, as an index of brain function, was assessed in regional whole-brain analyses using positron emission tomography (PET) and [18F]fluorodeoxyglucose in thirteen hypothyroid patients with autoimmune thyroiditis suffering from neuropsychiatric symptoms. The primary biological measures were radioactivity in pre-selected brain regions, relative to whole-brain radioactivity, as a surrogate index of glucose metabolism, and serum levels of thyroglobulin (TG) and thyroid peroxidase (TPO) antibodies as endocrine markers of autoimmune thyroiditis. RESULTS: Serum levels of anti-TG antibodies in hypothyroid patients were significantly correlated with glucose metabolism in the perigenual anterior cingulate cortex, a brain region previously shown to regulate affect and emotional homeostasis. CONCLUSION: Thyroid autoimmune processes may play an important role in the still poorly defined pathogenic correlates of disturbed function in brain regions critically involved in emotional processing in hypothyroid conditions.

Influence of levothyroxine in augmentation therapy for bipolar depression on central serotonergic function.

INTRODUCTION: Adjunctive treatment with supraphysiological doses of levothyroxine (L-T4) in bipolar depression shows promise, but the neurobiological mechanisms underlying clinical improvement are unknown. It has been postulated from animal studies that exogenous thyroid hormones may exert their modulatory effects in patients with affective disorders via an increase in serotonergic neurotransmission. Therefore, we investigated the loudness dependence of auditory evoked potentials (LDAEP) as a measure of central serotonergic activity and response to L-T4. METHODS: This 6-week, double-blind, randomized, placebo-controlled study assessed the efficacy of L-T4 adjunctive to continuing treatment with mood stabilizer and/or antidepressant medication in 20 patients with bipolar depression. LDAEP was assessed before and after treatment with L-T4. RESULTS: Scores of the Hamilton Depression Rating Scale and Montgomery Asberg Depression Rating Scale decreased significantly during the study. There was no difference in pre- and post-treatment LDAEP between the groups, and no correlation between LDAEP and psychometric measures in the course of the study. DISCUSSION: The hypothesis of a relationship between response of augmentation therapy with levothyroxine in bipolar depression and serotonergic activity could not be confirmed.

Dopamine-modulated aversive emotion processing fails in alcohol-dependent patients.

Negative mood states after alcohol detoxification may enhance the relapse risk. As recently shown in healthy volunteers, dopamine storage capacity (V d) in the left amygdala was positively correlated with functional activation in the left amygdala and anterior cingulate cortex (ACC) during an emotional task; high functional connectivity between the amygdala and the ACC, a region important for emotion regulation, was associated with low trait anxiety. Based on these findings, we now tested whether detoxified alcohol-dependent patients have a disrupted modulation of the anterior cingulate cortex activation in response to aversive stimuli by amygdala dopamine. Furthermore, we asked whether disrupted functional coupling between amygdala and ACC during aversive processing is related to trait anxiety.We used combined 6-[18F]-fluoro-l-DOPA positron emission tomography (PET), functional magnetic resonance imaging (fMRI) and Spielberger's state-trait anxiety questionnaire (STAI) in 11 male detoxified alcohol-dependent patients compared to 13 matched healthy controls.Unlike healthy controls, patients showed no significant correlation between our PET metric for dopamine storage capacity (FDOPA V d), in left amygdala and activation in left ACC. Moreover, the functional connectivity between amygdala and ACC during processing of aversive emotional stimuli was reduced in patients. Voxel-based morphometry did not reveal any discernible group differences in amygdala volume.These results suggest that dopamine-modulated corticolimbic circuit function is important for responding to emotional information such that apparent functional deficits in this neuromodulatory circuitry may contribute to trait anxiety in alcohol-dependent patients.

A new paradigm (Westphal-Paradigm) to study the neural correlates of panic disorder with agoraphobia.

Agoraphobia (with and without panic disorder) is a highly prevalent and disabling anxiety disorder. Its neural complexity can be characterized by specific cues in fMRI studies. Therefore, we developed a fMRI paradigm with agoraphobia-specific stimuli. Pictures of potential agoraphobic situations were generated. Twenty-six patients, suffering from panic disorder and agoraphobia, and 22 healthy controls rated the pictures with respect to arousal, valence, and agoraphobia-related anxiety. The 96 pictures, which discriminated best between groups were chosen, split into two parallel sets and supplemented with matched neutral pictures from the International Affective Picture System. Reliability, criterion, and construct validity of the picture set were determined in a second sample (44 patients, 28 controls). The resulting event-related "Westphal-Paradigm" with cued and uncued pictures was tested in a fMRI pilot study with 16 patients. Internal consistency of the sets was very high; parallelism was given. Positive correlations of picture ratings with Mobility Inventory and Hamilton anxiety scores support construct validity. FMRI data revealed activations in areas associated with the fear circuit including amygdala, insula, and hippocampal areas. Psychometric properties of the Westphal-Paradigm meet necessary quality requirements for further scientific use. The paradigm reliably produces behavioral and fMRI patterns in response to agoraphobia-specific stimuli. To our knowledge, it is the first fMRI paradigm with these properties. This paradigm can be used to further characterize the functional neuroanatomy of panic disorder and agoraphobia and might be useful to contribute data to the differentiation of panic disorder and agoraphobia as related, but conceptually different clinical disorders.

Reward and avoidance learning in the context of aversive environments and possible implications for depressive symptoms.

BACKGROUND: Aversive stimuli in the environment influence human actions. This includes valence-dependent influences on action selection, e.g., increased avoidance but decreased approach behavior. However, it is yet unclear how aversive stimuli interact with complex learning and decision-making in the reward and avoidance domain. Moreover, the underlying computational mechanisms of these decision-making biases are unknown. METHODS: To elucidate these mechanisms, 54 healthy young male subjects performed a two-step sequential decision-making task, which allows to computationally model different aspects of learning, e.g., model-free, habitual, and model-based, goal-directed learning. We used a within-subject design, crossing task valence (reward vs. punishment learning) with emotional context (aversive vs. neutral background stimuli). We analyzed choice data, applied a computational model, and performed simulations. RESULTS: Whereas model-based learning was not affected, aversive stimuli interacted with model-free learning in a way that depended on task valence. Thus, aversive stimuli increased model-free avoidance learning but decreased model-free reward learning. The computational model confirmed this effect: the parameter lambda that indicates the influence of reward prediction errors on decision values was increased in the punishment condition but decreased in the reward condition when aversive stimuli were present. Further, by using the inferred computational parameters to simulate choice data, our effects were captured. Exploratory analyses revealed that the observed biases were associated with subclinical depressive symptoms. CONCLUSION: Our data show that aversive environmental stimuli affect complex learning and decision-making, which depends on task valence. Further, we provide a model of the underlying computations of this affective modulation. Finally, our finding of increased decision-making biases in subjects reporting subclinical depressive symptoms matches recent reports of amplified Pavlovian influences on action selection in depression and suggests a potential vulnerability factor for mood disorders. We discuss our findings in the light of the involvement of the neuromodulators serotonin and dopamine.

Lateral prefrontal model-based signatures are reduced in healthy individuals with high trait impulsivity.

High impulsivity is an important risk factor for addiction with evidence from endophenotype studies. In addiction, behavioral control is shifted toward the habitual end. Habitual control can be described by retrospective updating of reward expectations in 'model-free' temporal-difference algorithms. Goal-directed control relies on the prospective consideration of actions and their outcomes, which can be captured by forward-planning 'model-based' algorithms. So far, no studies have examined behavioral and neural signatures of model-free and model-based control in healthy high-impulsive individuals. Fifty healthy participants were drawn from the upper and lower ends of 452 individuals, completing the Barratt Impulsiveness Scale. All participants performed a sequential decision-making task during functional magnetic resonance imaging (fMRI) and underwent structural MRI. Behavioral and fMRI data were analyzed by means of computational algorithms reflecting model-free and model-based control. Both groups did not differ regarding the balance of model-free and model-based control, but high-impulsive individuals showed a subtle but significant accentuation of model-free control alone. Right lateral prefrontal model-based signatures were reduced in high-impulsive individuals. Effects of smoking, drinking, general cognition or gray matter density did not account for the findings. Irrespectively of impulsivity, gray matter density in the left dorsolateral prefrontal cortex was positively associated with model-based control. The present study supports the idea that high levels of impulsivity are accompanied by behavioral and neural signatures in favor of model-free behavioral control. Behavioral results in healthy high-impulsive individuals were qualitatively different to findings in patients with the same task. The predictive relevance of these results remains an important target for future longitudinal studies.

Medial prefrontal brain activation to anticipated reward and loss in obsessive-compulsive disorder.

Obsessive-compulsive disorder (OCD) is associated with dysfunctional brain activity in several regions which are also involved in the processing of motivational stimuli. Processing of reward and punishment appears to be of special importance to understand clinical symptoms. There is evidence for higher sensitivity to punishment in patients with OCD which raises the question how avoidance of punishment relates to activity within the brain's reward circuitry. We employed the monetary incentive delay task paradigm optimized for modeling the anticipation phase of immediate reward and punishment, in the context of a cross-sectional event-related FMRI study comparing OCD patients and healthy control participants (n = 19 in each group). While overall behavioral performance was similar in both groups, patients showed increased activation upon anticipated losses in a medial and superior frontal cortex region extending into the cingulate cortex, and decreased activation upon anticipated rewards. No evidence was found for altered activation of dorsal or ventral striatal regions. Patients also showed more delayed responses for anticipated rewards than for anticipated losses whereas the reverse was true in healthy participants. The medial prefrontal cortex has been shown to implement a domain-general process comprising negative affect, pain and cognitive control. This process uses information about punishment to control aversively motivated actions by integrating signals arriving from subcortical regions. Our results support the notion that OCD is associated with altered sensitivity to anticipated rewards and losses in a medial prefrontal region whereas there is no significant aberrant activation in ventral or dorsal striatal brain regions during processing of reinforcement anticipation.

Brain glucose metabolism in hypothyroidism: a positron emission tomography study before and after thyroid hormone replacement therapy.

CONTEXT: Hypothyroidism is frequently associated with subtle behavioral and psychiatric symptoms. The consequences of inadequate thyroid hormone availability to brain metabolism are poorly understood. OBJECTIVE: This study assessed the relationships between neuropsychiatric symptoms and changes in relative regional cerebral glucose metabolism in hypothyroid patients undergoing thyroid hormone replacement therapy. DESIGN, SETTING, AND OUTCOME MEASURE: Relative regional cerebral glucose metabolism was compared in 13 previously untreated hypothyroid patients and 10 healthy control participants. Effects of thyroid hormone replacement therapy (levothyroxine, 3 months) were assessed using neuropsychiatric measures and positron emission tomography with [(18)F]fluorodeoxyglucose. RESULTS: Before treatment, hypothyroid patients exhibited lower regional activity than control subjects in the bilateral amygdala, hippocampus, and perigenual anterior cingulate cortex (ACC), left subgenual ACC, and right posterior cingulate cortex. Severity of depressive symptoms covaried negatively with pretreatment activity in the bilateral middle frontal gyrus and right subgenual and dorsal ACC. Thyroid hormone replacement therapy abolished pretreatment group differences in regional activity, robustly increased activity in the ventral ACC, and significantly reduced both clinician-rated and self-rated behavioral and psychiatric symptoms. Increased activity within the ventral ACC was associated with reduced somatic complaints, whereas increased activity within the dorsal ACC was associated with reduced depressive symptoms. CONCLUSIONS: Reduction of the behavioral complaints during thyroid hormone therapy is associated with a restoration of metabolic activity in brain areas that are integral to the regulation of affect and cognition. The findings suggest that thyroid hormone modulates regional glucose metabolism and psychiatric symptoms in the mature brain.

Functional dissociation between medial and lateral prefrontal cortical spatiotemporal activation in negative and positive emotions: a combined fMRI/MEG study.

The orbitofrontal cortex has been cytoarchitectonically and connectionally subdivided into a medial and a lateral part which are assumed to subserve distinct functions in emotional processing. However the exact spatiotemporal mechanisms of negative and positive emotional processing in medial and lateral orbitofrontal cortex remain unclear. We therefore investigated spatiotemporal orbitofrontal and prefrontal cortical activation patterns during emotional stimulation in a combined fMRI/MEG study. We investigated 10 healthy subjects, 5 women and 5 men. Positive and negative pictures from the International Affective Picture system (IAPS) were used for emotional stimulation, whereas neutral and gray pictures were taken as control conditions. fMRI/MEG measurements covered the whole frontal lobe and a time window between -2000 and +200 ms around motor responses (right index finger extension) associated with each picture. Positively and negatively correlated activities were determined in various prefrontal/frontal cortical regions in fMRI. Isocontour maps and single dipoles in MEG were analyzed in 50 ms time windows ranging from -2000 to +200 ms. Dipoles and fMR images were mapped on three-dimensional anatomical MRI so that anatomical localization of single dipoles and regional fMRI activity could be compared. Both negative and positive emotional conditions differed from non-emotional control conditions by strong orbitofrontal and lateral prefrontal activation as well as by the presence of early magnetic fields (-1700 to +1100 ms). Negative emotional processing was characterized by strong medial orbitofrontal activation and earlier (-1700 ms), stronger and more medially oriented orbitofrontal dipoles. In contrast positive emotional processing showed a rather strong activation in lateral prefrontal cortex with later (-1500 ms), weaker and more laterally oriented orbito and prefrontal dipoles. Negative emotional processing can be characterized by strong and early medial orbitofrontal cortical activation, whereas positive emotional processing showed rather later and weaker activation in lateral orbitofrontal/prefrontal cortex. Such a functional dissociation between medial and lateral orbito-frontal/prefrontal cortex during negative and positive emotional processing lends further support to the assumption of a functional subdivision in the orbitofrontal cortex.