Search

Deep neural networks have demonstrated promising performance in screening mammography with recent studies reporting performance at or above the level of trained radiologists on internal datasets. However, it remains unclear whether the performance of these trained models is robust and replicates across external datasets. In this study, we evaluate four state-of-the-art publicly available models using four publicly available mammography datasets (CBIS-DDSM, INbreast, CMMD, OMI-DB). Where test data was available, published results were replicated. The best-performing model, which achieved an area under the ROC curve (AUC) of 0.88 on internal data from NYU, achieved here an AUC of 0.9 on the external CMMD dataset (N = 826 exams). On the larger OMI-DB dataset (N = 11,440 exams), it achieved an AUC of 0.84 but did not match the performance of individual radiologists (at a specificity of 0.92, the sensitivity was 0.97 for the radiologist and 0.53 for the network for a 1-year follow-up). The network showed higher performance for in situ cancers, as opposed to invasive cancers. Among invasive cancers, it was relatively weaker at identifying asymmetries and was relatively stronger at identifying masses. The three other trained models that we evaluated all performed poorly on external datasets. Independent validation of trained models is an essential step to ensure safe and reliable use. Future progress in AI for mammography may depend on a concerted effort to make larger datasets publicly available that span multiple clinical sites. [ABSTRACT FROM AUTHOR]

In this work, we present a dataset that combines functional magnetic imaging (fMRI) and electroencephalography (EEG) to use as a resource for understanding human brain function in these two imaging modalities. The dataset can also be used for optimizing preprocessing methods for simultaneously collected imaging data. The dataset includes simultaneously collected recordings from 22 individuals (ages: 23–51) across various visual and naturalistic stimuli. In addition, physiological, eye tracking, electrocardiography, and cognitive and behavioral data were collected along with this neuroimaging data. Visual tasks include a flickering checkerboard collected outside and inside the MRI scanner (EEG-only) and simultaneous EEG-fMRI recordings. Simultaneous recordings include rest, the visual paradigm Inscapes, and several short video movies representing naturalistic stimuli. Raw and preprocessed data are openly available to download. We present this dataset as part of an effort to provide open-access data to increase the opportunity for discoveries and understanding of the human brain and evaluate the correlation between electrical brain activity and blood oxygen level-dependent (BOLD) signals. [ABSTRACT FROM AUTHOR]

Our continuous visual experience in daily life is dominated by change. Previous research has focused on visual change due to stimulus motion, eye movements or unfolding events, but not their combined impact across the brain, or their interactions with semantic novelty. We investigate the neural responses to these sources of novelty during film viewing. We analyzed intracranial recordings in humans across 6328 electrodes from 23 individuals. Responses associated with saccades and film cuts were dominant across the entire brain. Film cuts at semantic event boundaries were particularly effective in the temporal and medial temporal lobe. Saccades to visual targets with high visual novelty were also associated with strong neural responses. Specific locations in higher-order association areas showed selectivity to either high or low-novelty saccades. We conclude that neural activity associated with film cuts and eye movements is widespread across the brain and is modulated by semantic novelty. Movies are complex, continuous stimuli that are characterized by visual and semantic novelty. Here, by leveraging intracranial recordings from 23 humans, the authors find that responses to novelty across film cuts and saccades are widespread in the brain. [ABSTRACT FROM AUTHOR]

In large clinical centers a small subset of patients present with hydrocephalus that requires surgical treatment. We aimed to develop a screening tool to detect such cases from the head MRI with performance comparable to neuroradiologists. We leveraged 496 clinical MRI exams collected retrospectively at a single clinical site from patients referred for any reason. This diagnostic dataset was enriched to have 259 hydrocephalus cases. A 3D convolutional neural network was trained on 16 manually segmented exams (ten hydrocephalus) and subsequently used to automatically segment the remaining 480 exams and extract volumetric anatomical features. A linear classifier of these features was trained on 240 exams to detect cases of hydrocephalus that required treatment with surgical intervention. Performance was compared to four neuroradiologists on the remaining 240 exams. Performance was also evaluated on a separate screening dataset of 451 exams collected from a routine clinical population to predict the consensus reading from four neuroradiologists using images alone. The pipeline was also tested on an external dataset of 31 exams from a 2nd clinical site. The most discriminant features were the Magnetic Resonance Hydrocephalic Index (MRHI), ventricle volume, and the ratio between ventricle and brain volume. At matching sensitivity, the specificity of the machine and the neuroradiologists did not show significant differences for detection of hydrocephalus on either dataset (proportions test, p > 0.05). ROC performance compared favorably with the state-of-the-art (AUC 0.90–0.96), and replicated in the external validation. Hydrocephalus cases requiring treatment can be detected automatically from MRI in a heterogeneous patient population based on quantitative characterization of brain anatomy with performance comparable to that of neuroradiologists. [ABSTRACT FROM AUTHOR]

Severe brain injuries can lead to long-lasting disorders of consciousness (DoC) such as vegetative state/unresponsive wakefulness syndrome (VS/UWS) or minimally conscious state (MCS). While behavioral assessment remains the gold standard to determine conscious state, EEG has proven to be a promising complementary tool to monitor the effect of new therapeutics. Encouraging results have been obtained with invasive electrical stimulation of the brain, and recent studies identified transcranial direct current stimulation (tDCS) as an effective approach in randomized controlled trials. This non-invasive and inexpensive tool may turn out to be the preferred treatment option. However, its mechanisms of action and physiological effects on brain activity remain unclear and debated. Here, we stimulated 60 DoC patients with the anode placed over left-dorsolateral prefrontal cortex in a prospective open-label study. Clinical behavioral assessment improved in twelve patients (20%) and none deteriorated. This behavioral response after tDCS coincided with an enhancement of putative EEG markers of consciousness: in comparison with non-responders, responders showed increases of power and long-range cortico-cortical functional connectivity in the theta-alpha band, and a larger and more sustained P300 suggesting improved conscious access to auditory novelty. The EEG changes correlated with electric fields strengths in prefrontal cortices, and no correlation was found on the scalp. Taken together, this prospective intervention in a large cohort of DoC patients strengthens the validity of the proposed EEG signatures of consciousness, and is suggestive of a direct causal effect of tDCS on consciousness. [ABSTRACT FROM AUTHOR]

Transcranial electrical stimulation has widespread clinical and research applications, yet its effect on ongoing neural activity in humans is not well established. Previous reports argue that transcranial alternating current stimulation (tACS) can entrain and enhance neural rhythms related to memory, but the evidence from non-invasive recordings has remained inconclusive. Here, we measure endogenous spindle and theta activity intracranially in humans during low-frequency tACS and find no stable entrainment of spindle power during non-REM sleep, nor of theta power during resting wakefulness. As positive controls, we find robust entrainment of spindle activity to endogenous slow-wave activity in 66% of electrodes as well as entrainment to rhythmic noise-burst acoustic stimulation in 14% of electrodes. We conclude that low-frequency tACS at common stimulation intensities neither acutely modulates spindle activity during sleep nor theta activity during waking rest, likely because of the attenuated electrical fields reaching the cortical surface. [ABSTRACT FROM AUTHOR]

Music tends to be highly repetitive, both in terms of musical structure and in terms of listening behavior, yet little is known about how engagement changes with repeated exposure. Here we postulate that engagement with music affects the inter-subject correlation of brain responses during listening. We predict that repeated exposure to music will affect engagement and thus inter-subject correlation. Across repeated exposures to instrumental music, inter-subject correlation decreased for music written in a familiar style. Participants with formal musical training showed more inter-subject correlation, and sustained it across exposures to music in an unfamiliar style. This distinguishes music from other domains, where repetition has consistently been shown to decrease inter-subject correlation. Overall, the study suggests that listener engagement tends to decrease across repeated exposures of familiar music, but that unfamiliar musical styles can sustain an audience's interest, in particular in individuals with some musical training. Future work needs to validate the link proposed here between music engagement and inter-subject correlation of brain responses during listening. [ABSTRACT FROM AUTHOR]

Noninvasive brain stimulation techniques are used in experimental and clinical fields for their potential effects on brain network dynamics and behavior. Transcranial electrical stimulation (TES), including transcranial direct current stimulation (tDCS) and transcranial alternating current stimulation (tACS), has gained popularity because of its convenience and potential as a chronic therapy. However, a mechanistic understanding of TES has lagged behind its widespread adoption. Here, we review data and modelling on the immediate neurophysiological effects of TES in vitro as well as in vivo in both humans and other animals. While it remains unclear how typical TES protocols affect neural activity, we propose that validated models of current flow should inform study design and artifacts should be carefully excluded during signal recording and analysis. Potential indirect effects of TES (e.g., peripheral stimulation) should be investigated in more detail and further explored in experimental designs. We also consider how novel technologies may stimulate the next generation of TES experiments and devices, thus enhancing validity, specificity, and reproducibility. Transcranial electrical stimulation techniques, such as tDCS and tACS, are popular tools for neuroscience and clinical therapy, but how low-intensity current might modulate brain activity remains unclear. In this review, the authors review the evidence on mechanisms of transcranial electrical stimulation. [ABSTRACT FROM AUTHOR]

The original version of this Article contained an error in the last sentence of the first paragraph of the Introduction, which incorrectly read ‘Correlation of brain activity is typically measured using functional magnetic resonance imaging (fMRI), and the correlation structure is often referred to as “fu’. The correct version states ‘referred to as “functional connectivity”2-6’ in place of ‘referred to as “fu’. This has been corrected in both the PDF and HTML versions of the Article. [ABSTRACT FROM AUTHOR]

Global integration of information in the brain results from complex interactions of segregated brain networks. Identifying the most influential neuronal populations that efficiently bind these networks is a fundamental problem of systems neuroscience. Here, we apply optimal percolation theory and pharmacogenetic interventions in vivo to predict and subsequently target nodes that are essential for global integration of a memory network in rodents. The theory predicts that integration in the memory network is mediated by a set of low-degree nodes located in the nucleus accumbens. This result is confirmed with pharmacogenetic inactivation of the nucleus accumbens, which eliminates the formation of the memory network, while inactivations of other brain areas leave the network intact. Thus, optimal percolation theory predicts essential nodes in brain networks. This could be used to identify targets of interventions to modulate brain function. Complex networks can be used to model brain networks. Here the authors identify the essential nodes in a model of a brain network and then validate these predictions by means of in vivo pharmacogenetic interventions. They find that the nucleus accumbens is a central region for brain integration. [ABSTRACT FROM AUTHOR]

We performed simultaneous recordings of electroencephalography (EEG) from multiple students in a classroom, and measured the inter-subject correlation (ISC) of activity evoked by a common video stimulus. The neural reliability, as quantified by ISC, has been linked to engagement and attentional modulation in earlier studies that used high-grade equipment in laboratory settings. Here we reproduce many of the results from these studies using portable low-cost equipment, focusing on the robustness of using ISC for subjects experiencing naturalistic stimuli. The present data shows that stimulus-evoked neural responses, known to be modulated by attention, can be tracked for groups of students with synchronized EEG acquisition. This is a step towards real-time inference of engagement in the classroom. [ABSTRACT FROM AUTHOR]

A correction to the article "Origins of powerlaw degree distribution in the heterogeneity of human activity in social networks" is presented.

Deep neural networks have demonstrated promising performance in screening mammography with recent studies reporting performance at or above the level of trained radiologists on internal datasets. However, it remains unclear whether the performance of these trained models is robust and replicates across external datasets. In this study, we evaluate four state-of-the-art publicly available models using four publicly available mammography datasets (CBIS-DDSM, INbreast, CMMD, OMI-DB). Where test data was available, published results were replicated. The best-performing model, which achieved an area under the ROC curve (AUC) of 0.88 on internal data from NYU, achieved here an AUC of 0.9 on the external CMMD dataset (N = 826 exams). On the larger OMI-DB dataset (N = 11,440 exams), it achieved an AUC of 0.84 but did not match the performance of individual radiologists (at a specificity of 0.92, the sensitivity was 0.97 for the radiologist and 0.53 for the network for a 1-year follow-up). The network showed higher performance for in situ cancers, as opposed to invasive cancers. Among invasive cancers, it was relatively weaker at identifying asymmetries and was relatively stronger at identifying masses. The three other trained models that we evaluated all performed poorly on external datasets. Independent validation of trained models is an essential step to ensure safe and reliable use. Future progress in AI for mammography may depend on a concerted effort to make larger datasets publicly available that span multiple clinical sites., (© 2024. The Author(s).)