We propose a simple but generally applicable approach to improving the accuracy of automatic image segmentation algorithms relative to manual segmentations. The approach is based on the hypothesis that a large fraction of the errors produced by automatic segmentation are systematic, i.e., occur consistently from subject to subject, and serves as a wrapper method around a given host segmentation method. The wrapper method attempts to learn the intensity, spatial and contextual patterns associated with systematic segmentation errors produced by the host method on training data for which manual segmentations are available. The method then attempts to correct such errors in segmentations produced by the host method on new images. One practical use of the proposed wrapper method is to adapt existing segmentation tools, without explicit modification, to imaging data and segmentation protocols that are different from those on which the tools were trained and tuned. An open-source implementation of the proposed wrapper method is provided, and can be applied to a wide range of image segmentation problems. The wrapper method is evaluated with four host brain MRI segmentation methods: hippocampus segmentation using FreeSurfer (Fischl et al., 2002); hippocampus segmentation using multi-atlas label fusion (Artaechevarria et al., 2009); brain extraction using BET (Smith, 2002); and brain tissue segmentation using FAST (Zhang et al., 2001). The wrapper method generates 72%, 14%, 29% and 21% fewer erroneously segmented voxels than the respective host segmentation methods. In the hippocampus segmentation experiment with multi-atlas label fusion as the host method, the average Dice overlap between reference segmentations and segmentations produced by the wrapper method is 0.908 for normal controls and 0.893 for patients with mild cognitive impairment. Average Dice overlaps of 0.964, 0.905 and 0.951 are obtained for brain extraction, white matter segmentation and gray matter segmentation, respectively.

Major Depressive Disorder is a serious medical illness which is responsible for considerable morbidity and disability. Despite decades of research, the neural basis for depression is still incompletely understood. In this review, evidence from neuroimaging, neuropsychiatric and brain stimulations studies are explored to answer the question regarding the localization of depression in the brain. Neuroimaging studies indicate that although many regions of the brain have been repeatedly implicated in the pathophysiology of depression, not many consistent findings have been found until present. In recent times, the focus of neuroimaging has shifted from regional brain abnormalities to circuit level connectivity abnormalities. However, connectivity models are inherently more complicated, and the validity of these models remains to be tested. Neuropsychiatric studies of illnesses such as Parkinson’s Disease and stroke provide promising clues regarding areas involved in depression, but again consistent findings are rare. Similarly, stimulation of a variety of brain regions and circuits has been reported as being effective in depression. Therefore, the current knowledge indicates that the pathophysiology of depression may be distributed across many brain regions and circuits. In future studies, this distributed nature of depression needs to be further investigated, primary and secondary areas affected need to be identified, and new paradigms to explain complex mental functions need to be explored.

Measurement of brain change due to neurodegenerative disease and treatment is one of the fundamental tasks of neuroimaging. Deformation-based morphometry (DBM) has been long recognized as an effective and sensitive tool for estimating the change in the volume of brain regions over time. This paper demonstrates that a straightforward application of DBM to estimate the change in the volume of the hippocampus can result in substantial bias, i.e., an overestimation of the rate of change in hippocampal volume. In ADNI data, this bias is manifested as a non-zero intercept of the regression line fitted to the 6 and 12 month rates of hippocampal atrophy. The bias is further confirmed by applying DBM to repeat scans of subjects acquired on the same day. This bias appears to be the result of asymmetry in the interpolation of baseline and followup images during longitudinal image registration. Correcting this asymmetry leads to bias-free atrophy estimation.

Research on resting functional brain networks in bipolar disorder (BP) has been unable to differentiate between disturbances related to mania or depression, which is necessary to understand the mechanisms leading to each state. Past research has also been unable to elucidate the impact of BP-related network disturbances on the organizational properties of the brain (eg, communication efficiency). Thus, the present work sought to isolate network disturbances related to BP, fractionate these into components associated with manic and depressive symptoms, and characterize the impact of disturbances on network function. Graph theory was used to analyze resting functional magnetic resonance imaging data from 60 medication-free patients meeting the criteria for BP and either a current hypomanic (n=30) or depressed (n=30) episode and 30 closely age/sex-matched healthy controls. Correction for multiple comparisons was carried out. Compared with controls, BP patients evidenced hyperconnectivity in a network involving right amygdala. Fractionation revealed that (hypo)manic symptoms were associated with hyperconnectivity in an overlapping network and disruptions in the brain's 'small-world' network organization. Depressive symptoms predicted hyperconnectivity in a network involving orbitofrontal cortex along with a less resilient global network organization. Findings provide deeper insight into the differential pathophysiological processes associated with hypomania and depression, along with the particular impact these differential processes have on network function.

Automatic segmentation using multi-atlas label fusion has been widely applied in medical image analysis. To simplify the label fusion problem, most methods implicitly make a strong assumption that the segmentation errors produced by different atlases are uncorrelated. We show that violating this assumption significantly reduces the efficiency of multi-atlas segmentation. To address this problem, we propose a regression-based approach for label fusion. Our experiments on segmenting the hippocampus in magnetic resonance images (MRI) show significant improvement over previous label fusion techniques.

Multi-atlas based segmentation has been applied widely in medical image analysis. For label fusion, previous studies show that image similarity-based local weighting techniques produce the most accurate results. However, these methods ignore the correlations between results produced by different atlases. Furthermore, they rely on preselected weighting models and ad hoc methods to choose model parameters. We propose a novel label fusion method to address these limitations. Our formulation directly aims at reducing the expectation of the combined error and can be efficiently solved in a closed form. In our hippocampus segmentation experiment, our method significantly outperforms similarity-based local weighting. Using 20 atlases, we produce results with 0.898 ± 0.019 Dice overlap to manual labelings for controls.

Bipolar disorder (BP) is characterized by periods of depression (BPD) and (hypo)mania (BPM), but the underlying state-related brain circuit abnormalities are not fully understood. Striatal functional activation and connectivity abnormalities have been noted in BP, but consistent findings have not been reported. To further elucidate striatal abnormalities in different BP states, this study investigated differences in resting-state functional connectivity of six striatal subregions in BPD, BPM, and healthy control (HC) subjects. Ninety medication-free subjects (30 BPD, 30 BPM, and 30 HC), closely matched for age and gender, were scanned using 3T functional magnetic resonance imaging (fMRI) acquired at resting state. Correlations of low-frequency blood oxygen level dependent signal fluctuations for six previously described striatal subregions were used to obtain connectivity maps of each subregion. Using a factorial design, main effects for differences between groups were obtained and post hoc pairwise group comparisons performed. BPD showed increased connectivity of the dorsal caudal putamen with somatosensory areas such as the insula and temporal gyrus. BPM group showed unique increased connectivity between left dorsal caudate and midbrain regions, as well as increased connectivity between ventral striatum inferior and thalamus. In addition, both BPD and BPM exhibited widespread functional connectivity abnormalities between striatal subregions and frontal cortices, limbic regions, and midbrain structures. In summary, BPD exhibited connectivity abnormalities of associative and somatosensory subregions of the putamen, while BPM exhibited connectivity abnormalities of associative and limbic caudate. Most other striatal subregion connectivity abnormalities were common to both groups and may be trait related.

We performed a cross-sectional analysis of the prevalence of psychiatric diagnoses among transgender patients in clinical care using an all-payer electronic health record database. Of 10,270 transgender patients identified, 58% (n = 5940) had at least one psychiatric diagnosis compared with 13.6% (n = 7,311,780) in the control patient population (p < 0.0005). Transgender patients had a statistically significant increase in prevalence for all psychiatric diagnoses queried, with major depressive disorder and generalized anxiety disorder being the most common diagnoses (31% and 12%, respectively). Utilizing an all-payer database, although not without limitations, enables assessment of mental health and substance use diagnoses in this otherwise small population.