Machine learning is becoming an increasingly popular approach for investigating spatially distributed and subtle neuroanatomical alterations in brain‐based disorders. However, some machine learning ...models have been criticized for requiring a large number of cases in each experimental group, and for resembling a “black box” that provides little or no insight into the nature of the data. In this article, we propose an alternative conceptual and practical approach for investigating brain‐based disorders which aim to overcome these limitations. We used an artificial neural network known as “deep autoencoder” to create a normative model using structural magnetic resonance imaging data from 1,113 healthy people. We then used this model to estimate total and regional neuroanatomical deviation in individual patients with schizophrenia and autism spectrum disorder using two independent data sets (n = 263). We report that the model was able to generate different values of total neuroanatomical deviation for each disease under investigation relative to their control group (p < .005). Furthermore, the model revealed distinct patterns of neuroanatomical deviations for the two diseases, consistent with the existing neuroimaging literature. We conclude that the deep autoencoder provides a flexible and promising framework for assessing total and regional neuroanatomical deviations in neuropsychiatric populations.
Brain morphology varies across the ageing trajectory and the prediction of a person's age using brain features can aid the detection of abnormalities in the ageing process. Existing studies on such ...“brain age prediction” vary widely in terms of their methods and type of data, so at present the most accurate and generalisable methodological approach is unclear. Therefore, we used the UK Biobank data set (N = 10,824, age range 47–73) to compare the performance of the machine learning models support vector regression, relevance vector regression and Gaussian process regression on whole‐brain region‐based or voxel‐based structural magnetic resonance imaging data with or without dimensionality reduction through principal component analysis. Performance was assessed in the validation set through cross‐validation as well as an independent test set. The models achieved mean absolute errors between 3.7 and 4.7 years, with those trained on voxel‐level data with principal component analysis performing best. Overall, we observed little difference in performance between models trained on the same data type, indicating that the type of input data had greater impact on performance than model choice. All code is provided online in the hope that this will aid future research.
We compared the machine learning models support vector regression, relevance vector regression and Gaussian process regression for brain age prediction using different types of morphometric input and sample sizes of more than 10,000 subjects. The mean absolute error across the different models ranged from 3.7 to 4.7 years. The type of data input (region‐ or voxel‐level) had a greater impact on performance than the choice of model.
Abstract
The transcription factor interferon regulatory factor-5 (IRF5) plays an important role in innate immune responses via the TLR-MyD88 (Toll-like receptor - myeloid differentiation primary ...response 88) pathway. IRF5 is also involved in the pathogenesis of the autoimmune disease systemic lupus erythematosus (SLE). Recent studies have identified new regulators, both positive and negative, which act on IRF5 activation events in the TLR-MyD88 pathway such as post-translational modifications, dimerization and nuclear translocation. A model of the causal relationship between IRF5 activation and SLE pathogenesis proposes that a loss of the negative regulation of IRF5 causes its hyperactivation, resulting in hyperproduction of type I interferons and other cytokines, and ultimately in the development of SLE. Importantly, to our knowledge, all murine models of SLE studied thus far have shown that IRF5 is required for the pathogenesis of SLE-like diseases. During the development of SLE-like diseases, IRF5 plays key roles in various cell types, including dendritic cells and B cells. It is noteworthy that the onset of SLE-like diseases can be inhibited by reducing the activity or amount of IRF5 by half. Therefore, IRF5 is an important therapeutic target of SLE, and selective suppression of its activity and expression may potentially lead to the development of new therapies.
A recent study showed that people evaluate products more positively when they are physically associated with art images than similar non-art images. Neuroimaging studies of visual art have ...investigated artistic style and esthetic preference but not brain responses attributable specifically to the artistic status of images. Here we tested the hypothesis that the artistic status of images engages reward circuitry, using event-related functional magnetic resonance imaging (fMRI) during viewing of art and non-art images matched for content. Subjects made animacy judgments in response to each image. Relative to non-art images, art images activated, on both subject- and item-wise analyses, reward-related regions: the ventral striatum, hypothalamus and orbitofrontal cortex. Neither response times nor ratings of familiarity or esthetic preference for art images correlated significantly with activity that was selective for art images, suggesting that these variables were not responsible for the art-selective activations. Investigation of effective connectivity, using time-varying, wavelet-based, correlation-purged Granger causality analyses, further showed that the ventral striatum was driven by visual cortical regions when viewing art images but not non-art images, and was not driven by regions that correlated with esthetic preference for either art or non-art images. These findings are consistent with our hypothesis, leading us to propose that the appeal of visual art involves activation of reward circuitry based on artistic status alone and independently of its hedonic value.
► Reward circuitry activated by artistic status of image independent of content. ► Activation independent of esthetic preference, familiarity, and response time. ► Ventral striatum activity driven by visual cortex for art images. ► Ventral striatum not activated by non-art images matched for content. ► Ventral striatum not driven by regions correlated with esthetic preference ratings.
Affective decoding is the inference of human emotional states using brain signal measurements. This approach is crucial to develop new therapeutic approaches for psychiatric rehabilitation, such as ...affective neurofeedback protocols. To reduce the training duration and optimize the clinical outputs, an ideal clinical neurofeedback could be trained using data from an independent group of volunteers before being used by new patients. Here, we investigated if this subject-independent design of affective decoding can be achieved using functional near-infrared spectroscopy (fNIRS) signals from frontal and occipital areas. For this purpose, a linear discriminant analysis classifier was first trained in a dataset (49 participants, 24.65±3.23 years) and then tested in a completely independent one (20 participants, 24.00±3.92 years). Significant balanced accuracies between classes were found for positive vs. negative (64.50 ± 12.03%, p<0.01) and negative vs. neutral (68.25 ± 12.97%, p<0.01) affective states discrimination during a reactive block consisting in viewing affective-loaded images. For an active block, in which volunteers were instructed to recollect personal affective experiences, significant accuracy was found for positive vs. neutral affect classification (71.25 ± 18.02%, p<0.01). In this last case, only three fNIRS channels were enough to discriminate between neutral and positive affective states. Although more research is needed, for example focusing on better combinations of features and classifiers, our results highlight fNIRS as a possible technique for subject-independent affective decoding, reaching significant classification accuracies of emotional states using only a few but biologically relevant features.
Assessing the neural correlates of motor and cognitive processes under naturalistic experimentation is challenging due to the movement constraints of traditional brain imaging technologies. The ...recent advent of portable technologies that are less sensitive to motion artifacts such as Functional Near Infrared Spectroscopy (fNIRS) have been made possible the study of brain function in freely-moving participants. In this paper, we describe a series of proof-of-concept experiments examining the potential of fNIRS in assessing the neural correlates of cognitive and motor processes in unconstrained environments. We show illustrative applications for practicing a sport (i.e., table tennis), playing a musical instrument (i.e., piano and violin) alone or in duo and performing daily activities for many hours (i.e., continuous monitoring). Our results expand upon previous research on the feasibility and robustness of fNIRS to monitor brain hemodynamic changes in different real life settings. We believe that these preliminary results showing the flexibility and robustness of fNIRS measurements may contribute by inspiring future work in the field of applied neuroscience.
The signal and resolution during in vivo imaging of the mouse brain is limited by sample-induced optical aberrations. We find that, although the optical aberrations can vary across the sample and ...increase in magnitude with depth, they remain stable for hours. As a result, two-photon adaptive optics can recover diffraction-limited performance to depths of 450 μm and improve imaging quality over fields of view of hundreds of microns. Adaptive optical correction yielded fivefold signal enhancement for small neuronal structures and a threefold increase in axial resolution. The corrections allowed us to detect smaller neuronal structures at greater contrast and also improve the signal-to-noise ratio during functional Ca2+ imaging in single neurons.
Sleep is characterized by unique patterns of cortical activity alternating between the stages of slow-wave sleep (SWS) and rapid-eye movement (REM) sleep. How these patterns relate to the balanced ...activity of excitatory pyramidal cells and inhibitory interneurons in cortical circuits is unknown. We investigated cortical network activity during wakefulness, SWS, and REM sleep globally and locally using in vivo calcium imaging in mice. Wide-field imaging revealed a reduction in pyramidal cell activity during SWS compared with wakefulness and, unexpectedly, a further profound reduction in activity during REM sleep. Two-photon imaging on local circuits showed that this suppression of activity during REM sleep was accompanied by activation of parvalbumin (PV)+ interneurons, but not of somatostatin (SOM)+ interneurons. PV+ interneurons most active during wakefulness were also most active during REM sleep. Our results reveal a sleep-stage-specific regulation of the cortical excitation/inhibition balance, with PV+ interneurons conveying maximum inhibition during REM sleep, which might help shape memories in these networks.
•Cortical activity is suppressed globally during sleep, being lowest during REM sleep•During REM sleep, a subset of PV+ interneurons increase their activity•Neurons that are active during wake tend to show higher activity during REM sleep
Niethard et al. show that REM sleep is associated with a global suppression of cortical neural activity, which is accompanied by a specific activation of parvalbumin-positive inhibitory interneurons.
Meditation is a mental training, which involves attention and the ability to maintain focus on a particular object. In this study we have applied a specific attentional task to simply measure the ...performance of the participants with different levels of meditation experience, rather than evaluating meditation practice per se or task performance during meditation. Our objective was to evaluate the performance of regular meditators and non-meditators during an fMRI adapted Stroop Word-Colour Task (SWCT), which requires attention and impulse control, using a block design paradigm. We selected 20 right-handed regular meditators and 19 non-meditators matched for age, years of education and gender. Participants had to choose the colour (red, blue or green) of single words presented visually in three conditions: congruent, neutral and incongruent. Non-meditators showed greater activity than meditators in the right medial frontal, middle temporal, precentral and postcentral gyri and the lentiform nucleus during the incongruent conditions. No regions were more activated in meditators relative to non-meditators in the same comparison. Non-meditators showed an increased pattern of brain activation relative to regular meditators under the same behavioural performance level. This suggests that meditation training improves efficiency, possibly via improved sustained attention and impulse control.
► An fMRI adapted Stroop Word-Color Task (an attention task) was applied in two groups. ► Non-meditators showed greater brain activity than meditators in the comparisons. ► No brain regions were more activated in meditators relative to non-meditators. ► Behavioural performance was equivalent in both groups. ► Meditation training can increase brain efficiency in attention and impulse control.
Cortical computation is distributed across multiple areas of the cortex by networks of reciprocal connectivity. However, how such connectivity contributes to the communication between the connected ...areas is not clear. In this study, we examine the communication between sensory and motor cortices. We develop an eye movement task in mice and combine it with optogenetic suppression and two-photon calcium imaging techniques. We identify a small region in the secondary motor cortex (MO
) that controls eye movements and reciprocally connects with a rostrolateral part of the higher visual areas (V
). These two regions encode both motor signals and visual information; however, the information flow between the regions depends on the direction of the connectivity: motor information is conveyed preferentially from the MO
to the V
, and sensory information is transferred primarily in the opposite direction. We propose that reciprocal connectivity streamlines information flow, enhancing the computational capacity of a distributed network.
PDF
