Corrigendum: A 5-min Cognitive Task With Deep Learning Accurately Detects Early Alzheimer's Disease.

[This corrects the article DOI: 10.3389/fnagi.2020.603179.].

Depression is associated with hippocampal volume loss in adults with HIV.

Depressive symptoms are more prevalent in persons with HIV (PWH) than HIV-uninfected individuals. In HIV-uninfected individuals, depression has been associated with atrophy in the hippocampus and other brain regions. In the present study, we investigated the impact of depression on brain structure in PWH. One hundred PWH participated in a cross-sectional study (56.6 ± 6.4 yrs, range 41-70 yrs, 24 females, 63 African Americans). The Beck's Depression Inventory-II (BDI-II) was used to assess depressive symptoms. Structural MRI images were collected. Both the voxel-based morphometry (VBM) technique and a region of interest (ROI) based approach were used to examine the relationship between hippocampal gray matter volume (GMv) and depressive symptoms. The impact of HIV CD4 nadir and antidepressants was also investigated. Both VBM and ROI approaches revealed that higher BDI-II scores (implicating more severe depressive symptoms) were associated with loss of hippocampal GMv, especially in the right hippocampus and the right entorhinal cortex. Low CD4 nadir predicted additional hippocampal volume loss independent of depressive symptoms. Taking antidepressants did not have a detectable effect on hippocampal volume. In summary, having more depressive symptoms is associated with smaller hippocampal volume in PWH, and a history of severe immunosuppression (i.e., low CD4 nadir) correlates with additional hippocampal volume reduction. However, the impact of depression on hippocampal volume may be independent of HIV-disease severity such as low CD4 nadir.

A 5-min Cognitive Task With Deep Learning Accurately Detects Early Alzheimer's Disease.

Introduction: The goal of this study was to investigate and compare the classification performance of machine learning with behavioral data from standard neuropsychological tests, a cognitive task, or both. Methods: A neuropsychological battery and a simple 5-min cognitive task were administered to eight individuals with mild cognitive impairment (MCI), eight individuals with mild Alzheimer's disease (AD), and 41 demographically match controls (CN). A fully connected multilayer perceptron (MLP) network and four supervised traditional machine learning algorithms were used. Results: Traditional machine learning algorithms achieved similar classification performances with neuropsychological or cognitive data. MLP outperformed traditional algorithms with the cognitive data (either alone or together with neuropsychological data), but not neuropsychological data. In particularly, MLP with a combination of summarized scores from neuropsychological tests and the cognitive task achieved ~90% sensitivity and ~90% specificity. Applying the models to an independent dataset, in which the participants were demographically different from the ones in the main dataset, a high specificity was maintained (100%), but the sensitivity was dropped to 66.67%. Discussion: Deep learning with data from specific cognitive task(s) holds promise for assisting in the early diagnosis of Alzheimer's disease, but future work with a large and diverse sample is necessary to validate and to improve this approach.

Low CD4 nadir linked to widespread cortical thinning in adults living with HIV.

BACKGROUND: The history of immune suppression, especially CD4 nadir, has been shown to be a strong predictor of HIV-associated neurocognitive disorders (HAND). However, the potential mechanism of this association is not well understood. METHODS: High resolution structural MRI images and neuropsychological data were obtained from fifty-nine HIV+ adults (mean age, 56.5 ± 5.8) to investigate the correlation between CD4 nadir and cortical thickness. RESULTS: Low CD4 nadir was associated with widespread cortical thinning, especially in the frontal and temporal regions, and global mean cortical thickness correlated with CD4 nadir. In addition, worse global neurocognitive function was associated with bilateral frontal cortical thinning, and the association largely persisted (especially in the left frontal cortex) in the subset of participants who did not meet HAND criteria. CONCLUSIONS: These results suggest that low CD4 nadir may be associated with widespread neural injury in the brain, especially in the frontal and temporal regions. The diffuse neural injury might contribute to the prevalence and the phenotypes of HAND, as well as the difficulty treating HAND due to a broad network of brain regions affected. Low CD4 nadir related neural injury to the frontal cortex might contribute to subtle neurocognitive impairment/decline, even in the absence of HAND diagnosis.

Sequence learning in the human brain: A functional neuroanatomical meta-analysis of serial reaction time studies.

Sequence learning underlies numerous motor, cognitive, and social skills. Previous models and empirical investigations of sequence learning in humans and non-human animals have implicated cortico-basal ganglia-cerebellar circuitry as well as other structures. To systematically examine the functional neuroanatomy of sequence learning in humans, we conducted a series of neuroanatomical meta-analyses. We focused on the serial reaction time (SRT) task. This task, which is the most widely used paradigm for probing sequence learning in humans, allows for the rigorous control of visual, motor, and other factors. Controlling for these factors (in sequence-random block contrasts), sequence learning yielded consistent activation only in the basal ganglia, across the striatum (anterior/mid caudate nucleus and putamen) and the globus pallidus. In contrast, when visual, motor, and other factors were not controlled for (in a global analysis with all sequence-baseline contrasts, not just sequence-random contrasts), premotor cortical and cerebellar activation were additionally observed. The study provides solid evidence that, at least as tested with the visuo-motor SRT task, sequence learning in humans relies on the basal ganglia, whereas cerebellar and premotor regions appear to contribute to aspects of the task not related to sequence learning itself. The findings have both basic research and translational implications.

Language learning in the adult brain: A neuroanatomical meta-analysis of lexical and grammatical learning.

Language learning as an adult, though often difficult, is quite common. Nevertheless, the neural substrates of this process remain unclear, even though identifying them should clarify how language is learned and could lead to improved success at this endeavor. We addressed this gap by conducting multiple neuroanatomical meta-analyses to synthesize the functional neuroimaging literature of language learning. We focused on learning lexical and grammatical knowledge, two building blocks of language. Lexical and grammatical learning yielded overlapping activation in frontal (e.g., BA 44/45) and posterior parietal regions. Only lexical learning showed ventral occipito-temporal (ventral stream) activation, while only grammatical learning showed basal ganglia (anterior caudate/putamen) activation. To further elucidate the neurocognition of grammar learning, we also tested specific predictions of the declarative/procedural model of language. Consistent with the model, grammar learning predicted to rely especially on declarative memory (e.g., with explicit training) showed hippocampal involvement, while grammar learning predicted to rely particularly on procedural memory (e.g., with implicit training) showed anterior caudate/putamen involvement. Finally, given the prevalence of research on artificial grammars, we performed separate analyses of artificial grammar and non-artificial grammar (e.g., miniature language) paradigms. These yielded overlapping activation, especially in BA 44, underscoring the validity of artificial grammars as models for grammar learning in natural languages. In sum, the study elucidates the empirical and theoretical landscape of language learning and has applied implications.

Task-based changes in proton MR spectroscopy signal during configural working memory in human medial temporal lobe.

PURPOSE: To detect local cholinergic changes in human medial temporal lobe during configural working memory performance. MATERIALS AND METHODS: Proton magnetic resonance spectroscopy (1 H-MRS) measurements were acquired at 3T from a 2 × 2 × 3 cm voxel in right medial temporal lobe from 36 subjects during performance of a configural visual working memory task (cWMT). In order to compensate for expected task-based blood oxygenation level-dependent (BOLD) T2 * effects, resonance signal changes of unbound choline-containing metabolites (Cho) were referenced to an internal standard of creatine + phosphocreatine metabolites (Cre) and compared between four task blocks: rest, memorization, active memory maintenance, and recognition. An unannounced memory retention test was conducted in 21 subjects. Quality assurance analyses examined task-based Cho and Cre individually as well as referenced to resonance signal from N-acetylaspartate (NAA). RESULTS: Increases from a resting baseline in the Cho/Cre ratio were observed during 60-second blocks of active memory maintenance across the group (P = 0.0042). Behavioral accuracy during task performance correlated with memory retention (r = 0.48, P = 0.027). Quality assurance measures showed task-based changes in Cre resonance signal both individually (P = 0.00099) and when utilized as a noncholinergic internal reference (NAA/Cre, P = 0.00079). CONCLUSION: Increases in human medial temporal lobe 1 H-MRS Cho/Cre ratio occur during the maintenance of configural working memory information. However, interpretation of these results as driven by cholinergic activity cannot be assumed, as NAA, a noncholinergic metabolite, shows similar results when utilizing Cre as a reference. Caution is advised when considering Cre as an internal standard for task-based 1 H-MRS measurements. LEVEL OF EVIDENCE: 2 Technical Efficacy: Stage 1 J. Magn. Reson. Imaging 2018;47:682-691.

Diffusion tensor imaging in arginase deficiency reveals damage to corticospinal tracts.

Individuals with a proximal urea cycle disorder, such as carbamoyl phosphate synthetase deficiency 1 or ornithine transcarbamylase deficiency, may present with encephalopathy resulting from hyperammonemia. The clinical presentation of arginase deficiency is considerably different, characterized by progressive spasticity involving the lower extremities and usually dementia. Diagnosis may be delayed, and patients are often thought to have cerebral palsy. The true etiology of brain injury in arginase deficiency is unknown, but is not thought to be due to hyperammonemia and brain swelling, the mechanism of injury recognized in ornithine transcarbamylase deficiency. Elevated arginine could augment nitric oxide synthesis, leading to oxidative damage. The hypothesis for the present study was that specific brain vulnerability in arginase deficiency would involve microstructural alterations in corticospinal tracts and that this finding, as measured by diffusion tensor imaging, would differ from age-matched control subjects and those with ornithine transcarbamylase deficiency. Diffusion tensor imaging data were compared for a 17-year-old male patient with arginase deficiency, age-matched normal control subjects, and age-matched individuals with ornithine transcarbamylase deficiency. Significant differences were found in suspected areas of interest, specifically in the corticospinal tracts. This finding confirms the hypothesis that the mechanism of injury in arginase deficiency, although still unknown, is unlikely to be similar to that causing ornithine transcarbamylase deficiency.