Hosung Kim

Multidimensional classification of hippocampal shape features discriminates Alzheimer's disease and mild cognitive impairment from normal aging

We describe a new method to automatically discriminate between patients with Alzheimers disease (AD) or mild cognitive impairment (MCI) and elderly controls, based on multidimensional classification of hippocampal shape features. This approach uses spherical harmonics (SPHARM) coefficients to model the shape of the hippocampi, which are segmented from magnetic resonance images (MRI) using a fully automatic method that we previously developed. SPHARM coefficients are used as features in a classification procedure based on support vector machines (SVM). The most relevant features for classification are selected using a bagging strategy. We evaluate the accuracy of our method in a group of 23 patients with AD (10 males, 13 females, age+/-standard-deviation (SD)=73+/-6 years, mini-mental score (MMS)=24.4+/-2.8), 23 patients with amnestic MCI (10 males, 13 females, age+/-SD=74+/-8 years, MMS=27.3+/-1.4) and 25 elderly healthy controls (13 males, 12 females, age+/-SD=64+/-8 years), using leave-one-out cross-validation. For AD vs controls, we obtain a correct classification rate of 94%, a sensitivity of 96%, and a specificity of 92%. For MCI vs controls, we obtain a classification rate of 83%, a sensitivity of 83%, and a specificity of 84%. This accuracy is superior to that of hippocampal volumetry and is comparable to recently published SVM-based whole-brain classification methods, which relied on a different strategy. This new method may become a useful tool to assist in the diagnosis of Alzheimers disease.

Longitudinal and cross-sectional analysis of atrophy in pharmacoresistant temporal lobe epilepsy

Background: Whether recurrent epileptic seizures induce brain damage is debated. Disease progression in epilepsy has been evaluated only in a few community-based studies involving patients with seizures well controlled by medication. These studies concluded that epilepsy does not inevitably lead to global cerebral damage. Objective: To track the progression of neocortical atrophy in pharmacoresistant temporal lobe epilepsy (TLE) using longitudinal and cross-sectional designs. Methods: Using a fully automated measure of cortical thickness on MRI, we studied a homogeneous sample of patients with pharmacoresistant TLE. In the longitudinal analysis (n = 18), fixed-effect models were used to quantify cortical atrophy over a mean interscan interval of 2.5 years (range = 7 to 90 months). In the cross-sectional analysis (n = 121), we correlated epilepsy duration and thickness. To dissociate normal aging from pathologic progression, we compared aging effects in TLE to healthy controls. Results: The longitudinal analysis mapped progression in ipsilateral temporopolar and central and contralateral orbitofrontal, insular, and angular regions. In patients with more than 14 years of disease, atrophy progressed more rapidly in frontocentral and parietal regions that in those with shorter duration. The cross-sectional study showed progressive atrophy in the mesial and superolateral frontal, and parietal cortices. Conclusions: Our combined cross-sectional and longitudinal analysis in patients with pharmacoresistant temporal lobe epilepsy demonstrated progressive neocortical atrophy over a mean interval of 2.5 years that is distinct from normal aging, likely representing seizure-induced damage. The cumulative character of atrophy underlies the importance of early surgical treatment in this group of patients.

Mapping thalamocortical network pathology in temporal lobe epilepsy

Objective: Although experimental work has provided evidence that the thalamus is a crucial relay structure in temporal lobe epilepsy (TLE), the relation of the thalamus to neocortical pathology remains unclear. To assess thalamocortical network pathology in TLE, we mapped pointwise patterns of thalamic atrophy and statistically related them to neocortical thinning. Methods: We studied cross-sectionally 36 patients with drug-resistant TLE and 19 age- and sex-matched healthy control subjects using high-resolution MRI. To localize thalamic pathology, we converted manual labels into surface meshes using the spherical harmonic description and calculated local deformations relative to a template. In addition, we measured cortical thickness by means of the constrained Laplacian anatomic segmentation using proximity algorithm. Results: Compared with control subjects, patients with TLE showed ipsilateral thalamic atrophy that was located along the medial surface, encompassing anterior, medial, and posterior divisions. Unbiased analysis correlating the degree of medial thalamic atrophy with cortical thickness measurements mapped bilateral frontocentral, lateral temporal, and mesiotemporal cortices. These areas overlapped with those of cortical thinning found when patients were compared with control subjects. Thalamic atrophy intensified with a longer duration of epilepsy and was more severe in patients with a history of febrile convulsions. Conclusion: The degree and distribution of thalamic pathology relates to the topography and extent of neocortical atrophy, lending support to the concept that the thalamus is an important hub in the pathologic network of TLE.

Automated detection of cortical dysplasia type II in MRI-negative epilepsy

Objective: To detect automatically focal cortical dysplasia (FCD) type II in patients with extratemporal epilepsy initially diagnosed as MRI-negative on routine inspection of 1.5 and 3.0T scans. Methods: We implemented an automated classifier relying on surface-based features of FCD morphology and intensity, taking advantage of their covariance. The method was tested on 19 patients (15 with histologically confirmed FCD) scanned at 3.0T, and cross-validated using a leave-one-out strategy. We assessed specificity in 24 healthy controls and 11 disease controls with temporal lobe epilepsy. Cross-dataset classification performance was evaluated in 20 healthy controls and 14 patients with histologically verified FCD examined at 1.5T. Results: Sensitivity was 74%, with 100% specificity (i.e., no lesions detected in healthy or disease controls). In 50% of cases, a single cluster colocalized with the FCD lesion, while in the remaining cases a median of 1 extralesional cluster was found. Applying the classifier (trained on 3.0T data) to the 1.5T dataset yielded comparable performance (sensitivity 71%, specificity 95%). Conclusion: In patients initially diagnosed as MRI-negative, our fully automated multivariate approach offered a substantial gain in sensitivity over standard radiologic assessment. The proposed method showed generalizability across cohorts, scanners, and field strengths. Machine learning may assist presurgical decision-making by facilitating hypothesis formulation about the epileptogenic zone. Classification of evidence: This study provides Class II evidence that automated machine learning of MRI patterns accurately identifies FCD among patients with extratemporal epilepsy initially diagnosed as MRI-negative.

Spatial patterns of water diffusion along white matter tracts in temporal lobe epilepsy.

Objectives: Diffusion tensor imaging (DTI) tractography has shown tract-specific pathology in temporal lobe epilepsy (TLE). This technique normally yields a single value per diffusion parameter per tract, potentially reducing the sensitivity for the detection of focal changes. Our goal was to spatially characterize diffusion abnormalities of fasciculi carrying temporal lobe connections. Methods: We studied 30 patients with drug-resistant TLE and 21 healthy control subjects. Twenty-four patients underwent DTI toward the end of video-EEG telemetry, with an average of 50 ± 54 hours between the last seizure and DTI examination. After manual dissection of the uncinate and inferior longitudinal and arcuate bundle, they were spatially matched based on their distance to the temporal lobe, providing between-subject correspondence of tract segments. We evaluated point-wise differences in diffusion parameters along each tract at group and subject levels. Results: Our approach localized increased mean diffusivity restricted to or more prominent within the ipsilateral temporal lobe. These abnormalities tapered off as tracts exited the temporal lobe. We observed that the shorter the interval between the last seizure and DTI, the higher the mean diffusivity (MD) of the ipsilateral tracts. Linear discriminant analysis of tract segments correctly lateralized 87% of patients. Conclusions: The centrifugal pattern of white matter diffusion abnormalities probably reflects astrogliosis and microstructure derangement related to seizure activity in the vicinity of the focus. The negative correlation between the interval from last seizure and MD suggests a role for postictal vasogenic edema. The ability to assess tracts segmentally may contribute to a better understanding of the extent of white matter pathology in epilepsy and assist in the presurgical evaluation of patients with TLE, particularly those with unremarkable conventional imaging results.

Hippocampal substructural vulnerability to sleep disturbance and cognitive impairment in patients with chronic primary insomnia: magnetic resonance imaging morphometry.

STUDY OBJECTIVES Despite compelling evidence from animal studies indicating hippocampal subfield-specific vulnerability to poor sleep quality and related cognitive impairment, there have been no human magnetic resonance imaging (MRI) studies investigating the relationship between hippocampal subfield volume and sleep disturbance. Our aim was to investigate the pattern of volume changes across hippocampal subfields in patients with primary insomnia relative to controls. DESIGN Pointwise morphometry allowed for volume measurements of hippocampal regions on T1-weighted MRI. SETTING University hospital. PATIENTS Twenty-seven unmedicated patients (age: 51.2 ± 9.6 y) and 30 good sleepers as controls (50.4 ± 7.1 y). INTERVENTIONS N/A. MEASUREMENTS We compared hippocampal subfield volumes between patients and controls and correlated volume with clinical and neuropsychological features in patients. RESULTS Patients exhibited bilateral atrophy across all hippocampal subfields (P < 0.05 corrected). Cornu ammonis (CA) 1 subfield atrophy was associated with worse sleep quality (higher Pittsburgh Sleep Quality Index and higher arousal index of polysomnography) (r < -0.45, P < 0.005). The volume of the combined region, including the dentate gyrus (DG) and CA3-4, negatively correlated with verbal memory, verbal information processing, and verbal fluency in patients (|r| > 0.45, P < 0.05). Hemispheric volume asymmetry of this region (left smaller than right) was associated with impaired verbal domain functions (r = 0.50, P < 0.005). CONCLUSION Hippocampal subfield atrophy in chronic insomnia suggests reduced neurogenesis in the dentate gyrus (DG) and neuronal loss in the cornu ammonis (CA) subfields in conditions of sleep fragmentation and related chronic stress condition. Atrophy in the CA3-4-DG region was associated with impaired cognitive functions in patients. These observations may provide insight into pathophysiological mechanisms that make patients with chronic sleep disturbance vulnerable to cognitive impairment. CITATION Joo EY, Kim H, Suh S, Hong SB. Hippocampal substructural vulnerability to sleep disturbance and cognitive impairment in patients with chronic primary insomnia: magnetic resonance imaging morphometry.

Extensive migration of young neurons into the infant human frontal lobe

Building the human brain As the brain develops, neurons migrate from zones of proliferation to their final locations, where they begin to build circuits. Paredes et al. have discovered that shortly after birth, a group of neurons that proliferates near the ventricles migrates in chains alongside circulatory vessels into the frontal lobes (see the Perspective by McKenzie and Fishell). Young neurons that migrate postnatally into the anterior cingulate cortex then develop features of inhibitory interneurons. The number of migratory cells decreases over the first 7 months of life, and by 2 years of age, migratory cells are not evident. Any damage during migration, such as hypoxia, may affect the childs subsequent physical and behavioral development. Science, this issue p. 81; see also p. 38 Neurons are still finding their places as inhibitory circuits are established in the developing postnatal brain. [Also see Perspective by McKenzie] INTRODUCTION Inhibitory interneurons balance the excitation and inhibition of neural networks and therefore are key to normal brain function. In the developing brain, young interneurons migrate from their sites of birth into distant locations, where they functionally integrate. Although this neuronal migration is largely complete before birth, some young inhibitory interneurons continue to travel and add to circuits in restricted regions of the juvenile and adult mammalian brain. For example, postnatally migrating inhibitory neurons travel from the walls of the lateral ventricle, along the rostral migratory stream (RMS) into the olfactory bulb. In humans, an additional ventral route branching off the RMS, the medial migratory stream (MMS), takes young neurons into the medial prefrontal cortex. It has been suggested that recruitment of neurons during postnatal life could help shape neural circuits according to experience. Specifically, inhibitory interneuron maturation during postnatal development is associated with critical periods of brain plasticity. We asked whether neuronal recruitment continues into early childhood in the frontal lobe, a region of the human brain that has greatly increased in size and complexity during evolution. RATIONALE Migrating young neurons persist for several months after birth in an extensive region of the subventricular zone (SVZ) around the anterior lateral ventricles in the human brain. Are all these young neurons migrating into the RMS and MMS, or do they have other destinations? Using high-resolution magnetic resonance imaging (MRI), histology, and time-lapse confocal microscopy, we observed the migration of many young inhibitory interneurons around the dorsal anterior walls of the lateral ventricle and into multiple cortical regions of the human frontal cortex. We determined the location and orientation of these young neurons, demonstrated their active translocation, and inferred their fates in the postnatal anterior forebrain. RESULTS A large collection of cells expressing doublecortin (DCX), a marker of young migrating neurons, traveled and integrated within the infant frontal lobe. This migratory stream, which was most prominent during the first 2 months after birth and persisted until at least 5 months, formed a caplike structure surrounding the anterior body of the lateral ventricle. We refer to this population of young neurons as the Arc. This structure could also be visualized by brain MRI. Young neurons in the Arc appeared to move long distances in distinct regions around the ventricular wall and the developing white matter. The orientation of elongated DCX+ cells suggested that migratory neurons closer to the ventricular wall dispersed tangentially. In contrast, migratory neurons within the developing white matter tended to be orientated toward the overlying cortex. These cells expressed markers of interneurons, and their entry into the anterior cingulate cortex (a major target of the Arc used for quantification) was correlated with the emergence of specific subtypes of γ-aminobutyric acid (GABA)–expressing interneurons (neuropeptide Y, somatostatin, calretinin, and calbindin). Expression of transcription factors associated with specific sites of origin suggested that these neurons arise from ventral telencephalon progenitor domains. CONCLUSION Widespread neuronal migration into the human frontal lobe continues for several months after birth. Young neurons express markers of cortical inhibitory interneurons and originate outside the cortex, likely in the ventral forebrain. The postnatal recruitment of large populations of inhibitory neurons may contribute to maturation and plasticity in the human frontal cortex. Defects in the migration of these neurons could result in circuit dysfunction associated with neurodevelopmental disorders. Widespread neuronal migration into the human frontal lobe continues during postnatal life. (A) Sagittal schematic of the newborn forebrain shows prominent collections of young migratory neurons (illustrated in green) adjacent to the lateral ventricle (LV) and in the overlying white matter. Directional axes: D, dorsal; A, anterior. (B and C) DCX+ cells coexpress GABA and GAD67, markers of inhibitory interneurons (marked by arrows). The first few months after birth, when a child begins to interact with the environment, are critical to human brain development. The human frontal lobe is important for social behavior and executive function; it has increased in size and complexity relative to other species, but the processes that have contributed to this expansion are unknown. Our studies of postmortem infant human brains revealed a collection of neurons that migrate and integrate widely into the frontal lobe during infancy. Chains of young neurons move tangentially close to the walls of the lateral ventricles and along blood vessels. These cells then individually disperse long distances to reach cortical tissue, where they differentiate and contribute to inhibitory circuits. Late-arriving interneurons could contribute to developmental plasticity, and the disruption of their postnatal migration or differentiation may underlie neurodevelopmental disorders.

Patterns of subregional mesiotemporal disease progression in temporal lobe epilepsy

Objective: Evidence for disease progression in the mesiotemporal lobe is mainly derived from global volumetry of the hippocampus. In this study, we tracked progressive structural changes in the hippocampus, amygdala, and entorhinal cortex in drug-resistant temporal lobe epilepsy at a subregional level. Furthermore, we evaluated the relation between disease progression and surgical outcome. Methods: We combined cross-sectional modeling of disease duration in a large cohort of patients (n = 134) and longitudinal analysis in a subset that delayed surgery (n = 31). To track subregional pathology, we applied surface-shape analysis techniques on manual mesiotemporal labels. Results: Longitudinal and cross-sectional designs showed consistent patterns of progressive atrophy in hippocampal CA1, anterolateral entorhinal, and the amygdalar laterobasal group bilaterally. These regions also exhibited more marked age-related volume loss in patients compared with controls. We found a faster progression of hippocampal atrophy in patients with a seizure frequency ≥6 per month. High rates of contralateral entorhinal cortex atrophy predicted postsurgical seizure relapse. Conclusion: We observed progressive atrophy in hippocampal, amygdalar, and entorhinal subregions that frequently display neuronal loss on histology. The bilateral character of cumulative atrophy highlights the importance of early surgery. In patients who nevertheless delay this procedure, serial scanning may provide markers of surgical outcome.

Increased temporolimbic cortical folding complexity in temporal lobe epilepsy

Objective: Converging evidence suggests that abnormalities of brain development may play a role in the pathogenesis of temporal lobe epilepsy (TLE). As sulco-gyral patterns are thought to be a footprint of cortical development, we set out to quantitatively map folding complexity across the neocortex in TLE. Additionally, we tested whether there was a relationship between cortical complexity and features of hippocampal maldevelopment, commonly referred to as malrotation. Methods: To quantify folding complexity, we obtained whole-brain surface-based measures of absolute mean cortical curvature from MRI scans acquired in 43 drug-resistant patients with TLE with unilateral hippocampal atrophy, and 40 age- and sex-matched healthy controls. In patients, we correlated changes in cortical curvature with 3-dimensional measures of hippocampal positioning. Results: We found increased folding complexity in the temporolimbic cortices encompassing parahippocampal, temporopolar, insular, and fronto-opercular regions. Increased complexity was observed ipsilateral to the seizure focus in patients with left TLE (LTLE), whereas these changes were bilateral in patients with right TLE (RTLE). In both TLE groups, increased temporolimbic complexity was associated with increased hippocampal malrotation. We found tendencies for increased complexity in bilateral posterior temporal cortices in LTLE and contralateral parahippocampal cortices in RTLE to be predictive of unfavorable seizure outcome after surgery. Conclusion: The anatomic distribution of increased cortical complexity overlapping with limbic seizure networks in TLE and its association with hippocampal maldevelopment further imply that neurodevelopmental factors may play a role in the epileptogenic process of TLE.

Automatic hippocampal segmentation in temporal lobe epilepsy: Impact of developmental abnormalities

In drug-resistant temporal lobe epilepsy (TLE), detecting hippocampal atrophy on MRI is important as it allows defining the surgical target. The performance of automatic segmentation in TLE has so far been considered unsatisfactory. In addition to atrophy, about 40% of patients present with developmental abnormalities (referred to as malrotation) characterized by atypical morphologies of the hippocampus and collateral sulcus. Our purpose was to evaluate the impact of malrotation and atrophy on the performance of three state-of-the-art automated algorithms. We segmented the hippocampus in 66 patients and 35 sex- and age-matched healthy subjects using a region-growing algorithm constrained by anatomical priors (SACHA), a freely available atlas-based software (FreeSurfer) and a multi-atlas approach (ANIMAL-multi). To quantify malrotation, we generated 3D models from manual hippocampal labels and automatically extracted collateral sulci. The accuracy of automated techniques was evaluated relative to manual labeling using the Dice similarity index and surface-based shape mapping, for which we computed vertex-wise displacement vectors between automated and manual segmentations. We then correlated segmentation accuracy with malrotation features and atrophy. ANIMAL-multi demonstrated similar accuracy in patients and healthy controls (p > 0.1), whereas SACHA and FreeSurfer were less accurate in patients (p < 0.05). Surface-based analysis of contour accuracy revealed that SACHA over-estimated the lateral border of malrotated hippocampi (r = 0.61; p < 0.0001), but performed well in the presence of atrophy (|r |< 0.34; p > 0.2). Conversely, FreeSurfer and ANIMAL-multi were affected by both malrotation (FreeSurfer: r = 0.57; p = 0.02, ANIMAL-multi: r = 0.50; p = 0.05) and atrophy (FreeSurfer: r = 0.78, p < 0.0001, ANIMAL-multi: r = 0.61; p < 0.0001). Compared to manual volumetry, automated procedures underestimated the magnitude of atrophy (Cohens d: manual: 1.68; ANIMAL-multi: 1.11; SACHA: 1.10; FreeSurfer: 0.90, p < 0.0001). In addition, they tended to lateralize the seizure focus less accurately in the presence of malrotation (manual: 64%; ANIMAL-multi: 55%, p = 0.4; SACHA: 50%, p = 0.1; FreeSurfer: 41%, p = 0.05). Hippocampal developmental anomalies and atrophy had a negative impact on the segmentation performance of three state-of-the-art automated methods. These shape variants should be taken into account when designing segmentation algorithms.