R R Rawlings

National Institutes of Health, Bethesda, MD, United States

Are you R R Rawlings?

Claim your profile

Publications (101)633.47 Total impact

  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: A linear time-invariant model based on statistical time series analysis in the Fourier domain for single subjects is further developed and applied to functional MRI (fMRI) blood-oxygen level-dependent (BOLD) multivariate data. This methodology was originally developed to analyze multiple stimulus input evoked response BOLD data. However, to analyze clinical data generated using a repeated measures experimental design, the model has been extended to handle multivariate time series data and demonstrated on control and alcoholic subjects taken from data previously analyzed in the temporal domain. Analysis of BOLD data is typically carried out in the time domain where the data has a high temporal correlation. These analyses generally employ parametric models of the hemodynamic response function (HRF) where prewhitening of the data is attempted using autoregressive (AR) models for the noise. However, this data can be analyzed in the Fourier domain. Here, assumptions made on the noise structure are less restrictive, and hypothesis tests can be constructed based on voxel-specific nonparametric estimates of the hemodynamic transfer function (HRF in the Fourier domain). This is especially important for experimental designs involving multiple states (either stimulus or drug induced) that may alter the form of the response function.
    Computational and Mathematical Methods in Medicine 01/2013; 2013:645043. · 0.79 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Oxidative stress has been proposed as one of the mechanisms of alcohol-induced brain shrinkage and alcohol-induced hepatotoxicity. The aim of this study was to assess the correlations between liver function and brain volume (BV) measurements in patients with alcohol dependence. We recruited 124 patients with alcohol dependence and 111 healthy control subjects from National Institute of Health, National Institute on Alcohol Abuse and Alcoholism inpatient alcohol treatment program. Gamma-glutamyl transferase (GGT), aspartate aminotransferase (AST), alanine aminotransferase (ALT), as well as hematocrit (Hct) and albumin were assayed shortly after admission. Magnetic resonance imaging examination was conducted in both groups (after 3-week abstinence in the patient group). We used stepwise linear regression analyses to determine the variables most strongly correlated with brain shrinkage. Patients with alcohol dependence had lower BV, and greater brain shrinkage as measured by gray matter ratio (GMR), white matter ratio (WMR), brain ratio (BR), and higher cerebrospinal fluid ratio ratio (CSFR) compared with their healthy counterparts. Age and sex were significantly correlated with some BV measurements in both patient and control groups. Body mass index (BMI) was significantly correlated with CSFR, BR, GMR, and WMR; Hct with CSFR and BR; serum GGT level with BV, CSFR, BR, GMR, and WMF in the patient group. No biological variables were correlated with BV indices in the control group. In gender-stratified analysis, age was significantly correlated with brain shrinkage in male patients but not in female patients. Serum GGT level in male and female patients, Hct in male patients, and AST levels in female patients were significantly correlated with brain shrinkage. Our results showed that the higher levels of liver function indices, especially GGT, correlated with BV shrinkage as measured using CSFR, BR, GMR, and WMR in patients with alcohol dependence but not in controls. Serum GGT level outweighed aging effect on brain shrinkage in female patients.
    Alcoholism Clinical and Experimental Research 10/2011; 36(4):625-32. · 3.42 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Studies have shown that various brain structure abnormalities are associated with chronic alcohol abuse and impulsive aggression. However, few imaging studies have focused on violent individuals with a diagnosis of alcohol dependence. The present study used volumetric magnetic resonance imaging (MRI) to compare the volumes of different structural components of prefrontal cortex and six subcortical structures in perpetrators of intimate partner violence with alcohol dependence (IPV-ADs), non-violent alcohol-dependent patients (non-violent ADs) and healthy controls (HCs). Caucasian men (n = 54), ages 24-55, who had participated in National Institutes of Alcohol Abuse and Alcoholism treatment programs, were grouped together as IPV-ADs (n = 27), non-violent ADs (n = 14) and HCs (n = 13). The MRI scan was performed at least 3 weeks from the participant's last alcohol use. T1-weighted images were used to measure the volumes of intracranial space, gray and white matter, orbitofrontal cortex, medial prefrontal cortex, lateral prefrontal cortex, and six subcortical structures. Results revealed that IPV-ADs, compared with non-violent ADs and HCs, had a significant volume reduction in the right amygdala. No significant volumetric difference was found in other structures. This finding suggests that structural deficits in the right amygdala may underlie impulsive types of aggression often seen in alcohol-dependent patients with a history of IPV. It adds to a growing literature suggesting that there are fundamental differences between alcohol-dependent patients with and without IPV.
    Addiction Biology 10/2011; · 5.91 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The consumption of omega-3 (n-3) and omega-6 (n-6) essential fatty acids in Western diets is thought to have changed markedly during the 20th century. We sought to quantify changes in the apparent consumption of essential fatty acids in the United States from 1909 to 1999. We calculated the estimated per capita consumption of food commodities and availability of essential fatty acids from 373 food commodities by using economic disappearance data for each year from 1909 to 1999. Nutrient compositions for 1909 were modeled by using current foods (1909-C) and foods produced by traditional early 20th century practices (1909-T). The estimated per capita consumption of soybean oil increased >1000-fold from 1909 to 1999. The availability of linoleic acid (LA) increased from 2.79% to 7.21% of energy (P < 0.000001), whereas the availability of α-linolenic acid (ALA) increased from 0.39% to 0.72% of energy by using 1909-C modeling. By using 1909-T modeling, LA was 2.23% of energy, and ALA was 0.35% of energy. The ratio of LA to ALA increased from 6.4 in 1909 to 10.0 in 1999. The 1909-T but not the 1909-C data showed substantial declines in dietary availability (percentage of energy) of n-6 arachidonic acid, eicosapentaenoic acid (EPA), and docosahexaenoic acid (DHA). Predicted net effects of these dietary changes included declines in tissue n--3 highly unsaturated fatty acid status (36.81%, 1909-T; 31.28%, 1909-C; 22.95%, 1999) and declines in the estimated omega-3 index (8.28, 1909-T; 6.51, 1909-C; 3.84, 1999). The apparent increased consumption of LA, which was primarily from soybean oil, has likely decreased tissue concentrations of EPA and DHA during the 20th century.
    American Journal of Clinical Nutrition 03/2011; 93(5):950-62. · 6.50 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Behaviorally based therapies for the treatment of perpetrators who initiate intimate partner violence (IPV) have generally shown minimal therapeutic efficacy. To explore a new treatment approach for IPV, we examined the effects of a selective serotonin reuptake inhibitor on the irritability subscale score of the Modified Overt Aggression Scale. This score served as a surrogate marker for the anger and physical aggression that characterize perpetrators of IPV. A 12-week, double-blind, randomized, placebo-controlled intervention study employing fluoxetine, alcohol treatment, and cognitive-behavioral therapy was performed. Sixty (46 men) non-court-mandated, DSM-IV-diagnosed alcoholic perpetrators of IPV with a history of at least 2 episodes of IPV in the year prior to participation in the study were evaluated. The primary outcome measure was the score on the irritability subscale of the Modified Overt Aggression Scale. Secondary measures included anxiety, depression, and ratings by the perpetrator's spouse/significant other. The study was conducted from January 2002 through December 2007. A repeated-measures analysis of variance using the irritability subscale scores obtained from perpetrators who completed the 12-week study (n = 24) showed a significant drug effect (F(1,21) = 12.09, P = .002). Last observation carried forward (F(1,32) = 4.24, P = .048) as well as intent-to-treat analysis (F(1,54) = 5.0, P = .034) also showed a significant drug effect. Spouses'/significant others' physical and nonphysical Partner Abuse Scale ratings showed a significant reduction of abuse over time (F(1,11) = 10.2, P = .009 and F(1,11) = 24.2, P = .0005, respectively). This is the first controlled study to show that a pharmacologic intervention employing a selective serotonin reuptake inhibitor, in conjunction with alcohol treatment and cognitive-behavioral therapy, can reduce measures of anger and physical aggression in alcoholic perpetrators of IPV.
    The Journal of Clinical Psychiatry 01/2011; 72(1):60-5. · 5.81 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Alcoholism-associated liver disease is a potentially preventable hepatic injury. By use of 25 routinely ordered clinical laboratory tests, alcoholic cirrhosis was distinguishable from noncirrhotic alcoholic liver disease in 53 male inpatients with biopsy-verified liver disease (31 diagnosed with alcoholic cirrhosis and 22 diagnosed with fatty liver and/or alcoholic hepatitis). Linear discriminant analysis correctly classified 77% of subjects with cirrhotic liver disease and 68% of subjects with noncirrhotic liver disease (overall accuracy of 74%). Quadratic discriminant analysis correctly classified 94% of subjects with cirrhotic liver disease and 68% of subjects with noncirrhotic liver disease (81% overall accuracy). These pattern recognition techniques may allow earlier clinical intervention in alcoholic patients with liver disease.
    American Journal on Addictions 02/2010; 1(2):133 - 138. · 1.74 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: RationaleThere is an extensive literature showing that the CB1 cannabinoid receptor antagonist rimonabant (SR141716) decreases alcohol consumption in animals, but little is known about its effects in human alcohol drinkers. MethodsIn this study, 49 nontreatment-seeking heavy alcohol drinkers participated in a 3-week study. After a 1-week baseline, participants received either 20mg/day of rimonabant or placebo for 2weeks under double-blind conditions. During these 3weeks, participants reported their daily alcohol consumption by telephone. Subsequently, they participated in an alcohol self-administration paradigm in which they received a priming dose of alcohol followed by the option of consuming either eight alcohol drinks or receiving $3.00 for each nonconsumed drink. Endocrine measures and self-rating scales were also obtained. ResultsRimonabant did not change alcohol consumption during the 2weeks of daily call-ins. Similarly, the drug did not change either alcohol self-administration or endocrine measures during the laboratory session. ConclusionWe conclude that the daily administration of 20mg of rimonabant for 2weeks has no effect on alcohol consumption in nontreatment-seeking heavy alcohol drinkers.
    Psychopharmacology 01/2010; 208(1):37-44. · 4.06 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The general linear model (GLM) has been extensively applied to fMRI data in the time domain. However, traditionally time series data can be analyzed in the Fourier domain where the assumptions made as to the noise in the signal can be less restrictive and statistical tests are mathematically more rigorous. A complex form of the GLM in the Fourier domain has been applied to the analysis of fMRI (BOLD) data. This methodology has a number of advantages over temporal methods: 1. Noise in the fMRI data is modeled more generally and closer to that actually seen in the data. 2. Any input function is allowed regardless of the timing. 3. Non-parametric estimation of the transfer functions at each voxel are possible. 4. Rigorous statistical inference of single subjects is possible. This is demonstrated in the analysis of an experimental design with random exponentially distributed stimulus inputs (a two way ANOVA design with input stimuli images of alcohol, non-alcohol beverage and positive or negative images) sampled at 400 milliseconds. This methodology applied to a pair of subjects showed precise and interesting results (e.g. alcoholic beverage images attenuate the response of negative images in an alcoholic as compared to a control subject).
    Proc SPIE 02/2009;
  • DE Rio, RR Rawlings, LA Woltz, DW Hommer
    NeuroImage 01/2009; 47. · 6.25 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Alcohol dependence is a major public health challenge in need of new treatments. As alcoholism evolves, stress systems in the brain play an increasing role in motivating continued alcohol use and relapse. We investigated the role of the neurokinin 1 receptor (NK1R), a mediator of behavioral stress responses, in alcohol dependence and treatment. In preclinical studies, mice genetically deficient in NK1R showed a marked decrease in voluntary alcohol consumption and had an increased sensitivity to the sedative effects of alcohol. In a randomized controlled experimental study, we treated recently detoxified alcoholic inpatients with an NK1R antagonist (LY686017; n = 25) or placebo (n = 25). LY686017 suppressed spontaneous alcohol cravings, improved overall well-being, blunted cravings induced by a challenge procedure, and attenuated concomitant cortisol responses. Brain functional magnetic resonance imaging responses to affective stimuli likewise suggested beneficial LY686017 effects. Thus, as assessed by these surrogate markers of efficacy, NK1R antagonism warrants further investigation as a treatment in alcoholism.
    Science 04/2008; 319(5869):1536-9. · 31.20 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Alcoholism is characterized by deficits in emotional functioning as well as by deficits in cognitive functioning. However, most brain imaging research on alcoholism has focused on cognition rather than emotion. We used an event-related functional magnetic imaging approach to examine alcoholics' brain blood oxygenation level dependent (BOLD) response to evaluation of emotional stimuli and to compare their response to that of nonalcoholic controls. The task used was a simplified variant of a facial emotion-decoding task in which subjects determined the intensity level of a target emotion displayed as a facial expression. Facial expressions of happy, sad, anger, disgust, and fear were used as stimuli. Alcoholics and controls did not differ in accurately identifying the intensity level on the simple emotional decoding task but there were significant differences in their BOLD response during evaluation of facial emotion. In general, alcoholics showed less brain activation than nonalcoholic controls. The greatest differences in activation were during decoding of facial expressions of fear and disgust during which alcoholics had significantly less activation than controls in the affective division of the anterior cingulate cortex (ACC). Alcoholics also had significantly less activation than controls in the affective division of the ACC, while viewing sad faces. Only to facial expressions of anger did the alcoholics show significant activation in the affective ACC and in this case, their BOLD response did not significantly differ from that of the controls. Alcoholics show a deficit in the function of the affective division of the ACC during evaluation of negative facial emotions that can serve as cues for flight or avoidance. This deficit may underlie some of the behavioral dysfunction in alcoholism.
    Alcoholism Clinical and Experimental Research 10/2007; 31(9):1490-504. · 3.42 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: In this paper we consider a two-group discriminant analysis problem where each group is a mixture of two subgroups. Based upon data from a clinical study of alcohol involvement and diseases, simulation experiments were performed for three different configurations of means and covariance matrices. Expected actual non-error rates are estimated for the linear, quadratic, and kernel discriminant functions for sample sizes 30, 50, 75, 100, 150 and 200. A conclusion of the article is that the kernel discriminant function performs as well as or better than quadratic discriminant function. However, the linear discriminant function was clearly inferior to either the quadratic or kernel discriminant functions.
    Biometrical Journal 01/2007; 26(8):923 - 930. · 1.15 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The characterization of measurement error is critical in assessing the significance of diffusion tensor imaging (DTI) findings in longitudinal and cohort studies of psychiatric disorders. We studied 20 healthy volunteers, each one scanned twice (average interval between scans of 51 +/- 46.8 days) with a single shot echo planar DTI technique. Intersession variability for fractional anisotropy (FA) and Trace (D) was represented as absolute variation (standard deviation within subjects: SDw), percent coefficient of variation (CV) and intra-class correlation coefficient (ICC). The values from the two sessions were compared for statistical significance with repeated measures analysis of variance or a non-parametric equivalent of a paired t-test. The results showed good reproducibility for both FA and Trace (CVs below 10% and ICCs at or above 0.70 in most regions of interest) and evidence of systematic global changes in Trace between scans. The regional distribution of reproducibility described here has implications for the interpretation of regional findings and for rigorous pre-processing. The regional distribution of reproducibility measures was different for SDw, CV and ICC. Each one of these measures reveals complementary information that needs to be taken into consideration when performing statistical operations on groups of DT images.
    Psychiatry Research 07/2006; 147(1):69-78. · 2.68 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: A linear time invariant model is applied to functional fMRI blood flow data. Based on traditional time series analysis, this model assumes that the fMRI stochastic output sequence can be determined by a constant plus a linear filter (hemodynamic response function) of several fixed deterministic inputs and an error term assumed stationary with zero mean. The input function consists of multiple exponential distributed (time delay between images) visual stimuli consisting of negative and erotic images. No a priori assumptions are made about the hemodynamic response function that, in essence, is calculated at each spatial position from the data. The sampling rate for the experiment is 400 ms in order to allow for filtering out higher frequencies associated with the cardiac rate. Since the statistical analysis is carried out in the Fourier domain, temporal correlation problems associated with inference in the time domain are avoided. This formal model easily lends itself to further development based on previously developed statistical techniques.
    Computer Methods and Programs in Biomedicine 05/2006; 82(1):10-9. · 1.56 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Commonly used neuropsychological tests were administered to 91 detoxified alcoholics at the beginning of treatment. Statistically significant relationships were observed between test scores and post treatment consumption determined 8 months after completing treatment for 72 patients. The results varied depending upon the particular measure of posttreatment consumption evaluated and the type of statistical analysis used. The most consistent relationships were often counter to the notion that increased neuropsychological performance is correlated with a more favorable treatment outcome. Neuropsychological evaluation is of limited clinical utility in predicting posttreatment alcohol consumption.
    Alcoholism Clinical and Experimental Research 04/2006; 12(1):88 - 93. · 3.42 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Docosahexaenoic acid (DHA, 22:6n-3), an essential omega 3 fatty acid, may protect against disorders of emotional regulation as well as cardiovascular disease. Animal studies demonstrate that dietary folate can increase tissue concentrations of DHA, although the literature, to date, includes no human studies examining the possibility that folate status may affect plasma DHA concentrations. The objective of this study is to determine if the blood concentrations of folate and DHA are correlated in humans. Retrospective study. An American research hospital. A total of 15 normal and 22 hostile and aggressive subjects, with a mean age of 38 years. Concentrations of plasma polyunsaturated essential fatty acids and red blood cell folate (RBC folate) were obtained prior to 1996, before American flour was enriched with folate. RBC folate was significantly correlated with plasma DHA, r=0.57, P=0.005 in the aggressive group. Age, smoking and alcohol consumption did not alter the results. No other essential fatty acids were significantly associated with RBC folate in either group. The positive relationship between plasma DHA and RBC folate concentrations suggests that these two nutrients should be examined together in order to make the most accurate inferences about their relative contributions to disease pathogenesis. Our findings present one explanation why some conditions associated with hostility and low DHA status, such as cardiovascular disease and emotional disorders, are also associated with low folate status. National Institutes of Health, National Institute on Alcohol Abuse and Alcoholism.
    European Journal of Clinical Nutrition 04/2006; 60(3):352-7. · 2.76 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Research indicates that perpetrators of domestic violence have abnormalities in central serotonin and testosterone metabolism, an increased sensitivity to anxiogenic stimuli, and an impaired neuro-connection between their cortex and the amygdala. Clinical evaluations show that perpetrators of domestic violence also have a distinguishing set of behaviors and diagnoses related to anxiety, depression, intermittent explosive disorder, and borderline personality disorder. In this paper we propose a model to understand how the biological abnormalities can potentially explain the behaviors and diagnoses exhibited by the perpetrators. Changes in the perpetrator's neurotransmitters lead to a heightened sensitivity to environmental stimuli, anxiety, and conditioned fear. Lack of cortical input to the amygdala impairs the perpetrator's ability to extinguish anxiety and/or conditioned fear and gives rise to either innate behaviors (e.g., fight, flight, and shut down) or learned fear avoidant behaviors designed to avoid anxiety (e.g., alcohol consumption, self-injurious acts, and obsessive behaviors). Linking conditioned fear and fear avoidance to the behaviors and psychiatric diagnoses will serve to change the way the medical community perceives and treats perpetrators of domestic violence.
    Medical Hypotheses 02/2006; 67(2):345-53. · 1.18 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Childhood-onset schizophrenia (COS), a severe form of the disorder, is of interest for etiologic studies. Smooth pursuit eye-tracking dysfunction (ETD) is a biological marker for schizophrenia. To compare familial eye-tracking abnormalities for COS and adult-onset schizophrenia (AOS). Eye-tracking performance for 70 COS parents, 64 AOS parents and 20 COS siblings was compared to their respective age-matched control groups. COS and AOS parents had higher rate of dichotomously rated eye-tracking dysfunction than their respective controls (16% vs. 1% and 22% vs. 4%, respectively). COS parents and siblings also differed from controls on several continuous measures. However, scores for COS, AOS and control groups overlapped extensively. Genetic factors underlying eye-tracking dysfunction appear more salient for COS. However, eye-tracking measures have to be used with caution for endophenotypic definition due to low predictive power. The study was done at the National Institutes of Health.
    Schizophrenia Research 04/2005; 73(2-3):243-52. · 4.59 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: A subject of increasing importance in magnetic resonance imaging (MRI) is the analysis of intersubject structural differences, particularly when comparing groups of subjects with different conditions or diagnoses. On the other hand, determining structural homogeneity across subjects using voxel-based morphological (VBM) methods has become even more important to investigators who test for group brain activation using functional magnetic resonance images (fMRI) or positron emission tomography (PET). In the absence of methods that evaluate structural differences, one does not know how much reliability to assign to the functional differences. Here, we describe a voxel-based method for quantitatively assessing the homogeneity of tissues from structural magnetic resonance images of groups. Specifically, this method determines the homogeneity of gray matter for a group of subjects. Homogeneity probability maps (HPMs) of a given tissue type (e.g., gray matter) are generated by using a confidence interval based on binomial distribution. These maps indicate for each voxel the probability that the tissue type is gray for the population being studied. Therefore, HPMs can accompany functional analyses to indicate the confidence one can assign to functional difference at any given voxel. In this paper, examples of HPMs generated for a group of control subjects are shown and discussed. The application of this method to functional analysis is demonstrated.
    NeuroImage 04/2004; 21(3):965-72. · 6.25 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: In an earlier study, we reported that some perpetrators of domestic violence evidenced exaggerated fear-related responses to the panicogenic agent sodium lactate. In the current study, we employed positron emission tomography (PET) to investigate our hypothesis that there are differences in the neural structures and/or pathways that mediate and control the expression of fear-induced aggression in perpetrators of domestic violence. Regional cerebral glucose metabolism was measured in eight male perpetrators of domestic violence who fulfilled DSM-III-R criteria for alcohol dependence (DV-ALC), 11 male participants who fulfilled DSM-III-R criteria for alcohol dependence and had no history of interpersonal aggression (ALC) and 10 healthy male participants who did not fulfill criteria for any DSM-III-R axis I diagnosis and had no history of interpersonal aggression (HCS). DV-ALC had a significantly lower mean glucose uptake in the right hypothalamus compared to ALC and HCS. Correlations were performed between measures of glucose utilization in the brain structures involved in fear-induced aggression. The comparison of DV-ALC to HCS and to ALC differed in six and seven comparisons, respectively, involving various cortical and subcortical structures. HCS and ALC differed between the left thalamus and the left posterior orbitofrontal cortex. These PET findings show that some perpetrators of domestic violence differ from control participants in showing lower metabolism in the right hypothalamus and decreased correlations between cortical and subcortical brain structures. A possible psychological covariate of these changes in regional activity might be fear-induced aggression, but this hypothesis should be examined in larger study groups that undergo provocation during imaging.
    Psychiatry Research 02/2004; 130(1):11-25. · 2.68 Impact Factor

Publication Stats

3k Citations
633.47 Total Impact Points

Institutions

  • 1992–2013
    • National Institutes of Health
      • • Laboratory of Clinical and Translational Studies
      • • Laboratory of Membrane Biochemistry and Biophysics
      Bethesda, MD, United States
    • Vanderbilt University
      Nashville, Michigan, United States
  • 1984–2011
    • National Institute on Alcohol Abuse and Alcoholism
      Maryland, United States
  • 1990–2001
    • National Institute of Mental Health (NIMH)
      • Clinical Brain Disorders Branch
      Bethesda, MD, United States
  • 1998
    • National Eye Institute
      Maryland, United States
  • 1994–1998
    • University of Helsinki
      • Department of Psychiatry
      Helsinki, Province of Southern Finland, Finland
    • Washington DC VA Medical Center
      Washington, Washington, D.C., United States
  • 1997
    • Helsinki University Central Hospital
      Helsinki, Southern Finland Province, Finland
  • 1995
    • Washington University in St. Louis
      • Department of Psychiatry
      San Luis, Missouri, United States