Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Deep Neural Network”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 109 records · Page 6Linked to original sources
PubMedDate not supplied

Interrelated modification of excitatory and inhibitory connections in the olivocerebellar neural network.

A model of plasticity is proposed for the olivocerebellar neural network in which the efficiency of the synaptic inputs to different neurons changes simultaneously and interdependently. This effect is based on the following functional characteristics of the network: simultaneous arrival of an afferent signal via mossy fibers to input granule cells and output neurons in the deep cerebellar nuclei; synchronous arrival of the signal from the inferior olive, via climbing fibers and their collaterals, at cells in the input and output layers, and to Purkinje cells, and the existence of local excitatory, inhibitory, and disinhibitory feedback circuits. Increases (decreases) in post-tetanic Ca2+ concentrations relative to the level evoked by the preceding stimulation in these cells are accompanied by decreases (increases) in the activity of cGMP-dependent protein kinase G, with increases (decreases) in the activity of protein phosphatase I. As a result, dephosphorylation (phosphorylation) of ionotropic receptors is accompanied by simultaneous depression (potentiation) of the excitatory input to a given neuron and potentiation (depression) of the inhibitory input to the same neuron. The depolarizing signal from the inferior olive affects synapse modification in different layers of the network in such a way that its presence (absence) depresses (potentiates) the signal sent from the output cells of the cerebellum to other structures.

Animals↗
PubMed2026

Graph neural network-based risk stratification of prostate cancer using gene expression and SHAP interpretability.

Accurate risk stratification is essential for guiding treatment decisions and preventing over treatment of prostate cancer, which remains one of the most prevalent cancers among adult men. While the Gleason score, obtained from prostate biopsies, is routinely used to assess tumor aggressiveness, the biopsy procedure carries risks such as pain, infection, and, in some cases, serious complications such as sepsis. In this study, we proposed an artificial intelligence-based framework that integrates mRNA expression profiles with functional interaction networks to classify prostate cancer patients into low-, medium-, and high-risk groups defined by Gleason scores. The pipeline comprised five steps: (1) data collection from The Cancer Genome Atlas (TCGA), (2) preprocessing of gene expression data, (3) two-stage feature selection to identify informative biomarkers, (4) risk classification using a dual-branch graph neural network (GNN) that combines gene-gene interaction graphs with sample-level expression features, and (5) model interpretation using SHAP to quantify feature contributions. Differentially expressed genes were identified in the High (ASPN, GMNN, PEBP4, C2, KNCK17), Medium (C2, IGSF1, ASPN, CDKN3, AMH), and Low (TNMD, VWA5B2, ST6GALNAC5, CYP3A5, PHGR1) risk groups, underscoring the molecular heterogeneity of disease progression. On an independent held-out test set, the model achieved AUCs of 0.86, 0.88, and 0.95 for the low-, medium-, and high-risk groups, respectively, with an overall accuracy of 80%. These results suggest that combining GNN-based modeling with explainable AI can capture both global and local molecular patterns relevant to tumor aggressiveness. However, as the model was developed and evaluated solely on the TCGA cohort, the findings should be regarded as exploratory, and external validation will be required to establish generalizability. Within these limitations, the proposed framework highlights the potential of molecular profiling and graph-based deep learning to support more precise, potentially less invasive, risk assessment and individualized treatment planning in prostate cancer.

Prostatic Neoplasms↗
PubMed1985

[Surface of the vascular plexuses of human cerebral ventricles].

The racemose part of vascular plexuses of the lateral ventricles has been studied in fetuses and persons at various age in the scanning and transmissive electron microscopes. In the magistral arteries of the vascular plexuses the neural trunks have been studied. They form the periadventitial plexus with a more complex network of connections in mature persons. The sulci and the elongated folds between them make the relief of the plexus. A rough surface of epitheliocytes has deep craters and irregular protrusions, microvilli, cilia and spherical bodies. The epitheliocytes are arranged in a single layer and connected with each other by means of protoplasmic peduncles. Over the epithelial layer, as single groups, Kolmer cells are situated.

Adolescent↗
PubMedDate not supplied

Altered brain activation during cognitive control in patients with moderate to severe traumatic brain injury.

BACKGROUND: Persistent deficits in cognitive control have been documented following traumatic brain injury (TBI) but are inconsistently related to the presence and location of focal lesions. OBJECTIVE: Functional magnetic resonance imaging (fMRI) was used to examine brain activation during a cognitive control task in patients with moderate to severe TBI or orthopedic injury (OI). METHODS: Fourteen TBI patients and 10 OI patients underwent fMRI at 3 months postinjury using a stimulus-response compatibility task in which response accuracy and reaction time were measured. Performance between the groups was equated by individually adjusting the amount of training. Groups did not differ in age, gender, or education. RESULTS: Brain activation during stimulus-response incompatibility was greater in TBI patients than in OI patients within the cingulate, medial frontal, middle frontal, and superior frontal gyri. However, the positive regression of activation with response accuracy during stimulus-response incompatibility indicated a stronger relationship for OI patients than the TBI group within the anterior cingulate gyrus, medial frontal, and parietal regions, as well as deep brain structures (eg, brainstem). The number of focal lesions within either the whole brain or within prefrontal areas was not related to brain activation, but there was a relationship between activation and TBI severity. CONCLUSIONS: These findings suggest that neural networks mediating cognitive control are altered after moderate to severe TBI, possibly as a result of diffuse axonal injury, and that the typical relationship of brain activation to performance is disrupted.

Adolescent↗
PubMed1999

Core assessment program for surgical interventional therapies in Parkinson's disease (CAPSIT-PD).

In 1992 the Core Assessment Program for Intracerebral Transplantations (CAPIT) was published providing the minimal requirements for a common patient evaluation protocol. Despite the intent, the program was thought to be too laborious to carry out in large scale trials, and it also lacked evaluations of cognitive functions and quality of life. Moreover, the CAPIT was designed for neural transplantation only and has not been revised since. Since then, pallidotomy and deep brain stimulation have emerged as additional treatment modalities but there exists no common tool for evaluation of, and between, the techniques. In 1996, within the framework of NECTAR (Network for European CNS Transplantation and Restoration), a dedicated program entitled "Neurosurgical Interventions in Parkinson's Disease" (NIPD) was funded by the European Union Biomed 2 program to develop a new Core Assessment Program for Surgical Interventional Therapies in PD (CAPSIT-PD) and to establish an European registry for patients with PD subjected to functional neurosurgery. This article presents the recommendations of this new program.

Clinical Protocols↗
PubMed1995

Impaired development of interstitial cells and intestinal electrical rhythmicity in steel mutants.

Electrical rhythmicity in the gastrointestinal tract may originate in interstitial cells of Cajal (IC). Development of IC in the small intestine is linked to signaling via the tyrosine kinase receptor, c-kit. IC express c-kit protein, and disruption of c-kit signaling causes breakdown in IC networks and loss of slow waves. We tested whether mutations in steel factor, the ligand for c-kit, affect the development of IC networks. IC were found in the region of the myenteric plexus (IC-MY) in mice with steel mutations (i.e., Sl/Sld) at 5-10 days postpartum, but these cells formed an abnormal network. IC-MY were not observed in adult Sl/Sld animals. IC in the deep muscular plexus (IC-DMP) appeared normal in Sl/Sld animals. Electrical slow waves, normally present in the small intestine, were absent in Sl/Sld animals (10-30 days postpartum). Neural inputs were intact in Sl/Sld animals. Steel factor appears important for the development of certain classes of IC, and IC-MY appear to be involved in the generation of electrical rhythmicity in the small intestine.

Animals↗
PubMed2004

Control of phase synchronization of neuronal activity in the rat hippocampus.

Analysis of the synchronization mechanisms of neural activity is crucial to the understanding of the generation, propagation and control of epileptiform activity. Recently, phase synchronization (PS) analysis was applied to quantify the partial synchrony that exists in complex chaotic or noisy systems. In a previous study, we have shown that neural activity between two remotely located sites can be synchronized through a complete cut of the tissue by endogenous non-synaptic signals. Therefore, it should be possible to apply signals to control PS. In this study, we test the hypothesis that stimulation amplitudes below excitation level (sub-threshold) can be used to control phase synchronization of two neural signals and we investigate the underlying mechanisms. PS of neuronal activity is first analysed in two coupled Rossler neuron models. Both synchronization and desynchronization could be generated with sub-threshold sinusoidal stimulation. Phase synchronization was then studied in in vitro brain slices. Neuronal activity between two sites was modulated by the application of small sinusoidal electric fields. PS between two remote sites could be achieved by the application of two identical waveforms while phase desynchronization of two close sites was generated by the application of a stimulus at a single site. These results show that sub-threshold stimuli are able to phase synchronize or desynchronize two networks and suggest that small signals could play an important role in normal neural activity and epilepsy.

Action Potentials↗
PubMed2001

Attention to movement modulates activity in sensori-motor areas, including primary motor cortex.

Attention to sensory stimulation modulates behavioural responses and cortical activity. Attention to movement can also modulate motor responses. For example, directing attention away from cued movements can increase reaction times. This study used fMRI to determine where in the motor cortex attention to movement modulates activity. Attention to movement was reduced by asking subjects to perform a concurrent distractor task (counting backwards). Sensori-motor areas showing a negative interaction between counting and movement (i.e. reduced activation in the dual task condition relative to the sum of the single task conditions) included the supplementary motor area (SMA), cingulate cortex, insula and post-central gyrus. A separate volumes-of-interest analysis revealed significant reductions in mean percent signal change in the dual task compared to the single task in a portion of the pre-central gyrus, deep in the central sulcus (thought to correspond to area 4p) and SMA. We conclude that the brain network for motor control is modulated by attention at multiple sites, including the primary motor cortex. These results are also discussed with reference to theories concerning the neural correlates of dual task performance and mental calculation and have implications for the interpretation of functional imaging studies of normal and impaired motor performance.

Adult↗
PubMed2003

Brain activation patterns during video sexual stimulation following the administration of apomorphine: results of a placebo-controlled study.

OBJECTIVES: To evaluate the in vivo effect of apomorphine sublingual versus placebo on cortical and subcortical brain activation during video sexual stimulation. METHODS: Ten patients with psychogenic erectile dysfunction and six potent controls underwent functional magnetic resonance of the brain during video sexual stimulation after the administration of either apomorphine sublingual 4mg or placebo following a randomized, double blind design. Functional magnetic resonance sessions were performed with a 7-day interval. RESULTS: In potent controls, viewing erotic versus neutral films induced bilateral activations in a network of occipito-parietal and temporal inferior regions, in dorsolateral and premotor frontal cortex, in anterior temporal limbic areas and the thalamus, which were comparable to the patient activations during erotic stimulation in the placebo condition. However, a striking difference was found in patients, who demonstrated a significant and extended activation in the cingulate gyrus, frontal mesial and frontal basal cortex, bilaterally, in comparison with potent controls. These activated neural systems were modulated by apomorphine administration which produced a picture that was similar to the one seen in potent controls. In patients with spychogenic erectile dysfunction apomorphine sublingual caused an increase in the extension of the activated networks, plus additional activation foci in subcortical and deep structures, namely in the nucleus accumbens, hypothalamus and mesencephalon: this activation was greater than that seen with placebo. Interestingly, a down-regulation in the frontal basal and temporal limbic cortex was present as shown by a decrease of functional magnetic resonance imaging signal reflecting a deactivation of these regions. CONCLUSIONS: Apomorphine significantly enhances the activation of cortical and subcortical brain function during video sexual stimulation. Patients with psychogenic erectile dysfunction may have an underlying functional abnormality of the brain acting as a previously unrecognised aetiological factor.

Administration, Sublingual↗
PubMed2005

Applications of two-photon microscopy in the neurosciences.

Thanks to deep tissue penetration of infrared light (IR), two-photon laser scanning microscopy (2PLSM) has become an important tool in modern neuroscience. Examples range from the high-resolution visualization of dendritic spines in single neurons to the monitoring of intracellular Ca++ transients in small neural networks. In the present review we give an overview of the current achievements and discuss emerging applications of the technique.

Brain Mapping↗
PubMed2006

Comparison of manual sleep staging with automated neural network-based analysis in clinical practice.

We have compared sleep staging by an automated neural network (ANN) system, BioSleep (Oxford BioSignals) and a human scorer using the Rechtschaffen and Kales scoring system. Sleep study recordings from 114 patients with suspected obstructed sleep apnoea syndrome (OSA) were analysed by ANN and by a blinded human scorer. We also examined human scorer reliability by calculating the agreement between the index scorer and a second independent blinded scorer for 28 of the 114 studies. For each study, we built contingency tables on an epoch-by-epoch (30 s epochs) comparison basis. From these, we derived kappa (kappa) coefficients for different combinations of sleep stages. The overall agreement of automatic and manual scoring for the 114 studies for the classification {wake / light-sleep / deep-sleep / REM} was poor (median kappa = 0.305) and only a little better (kappa = 0.449) for the crude {wake / sleep} distinction. For the subgroup of 28 randomly selected studies, the overall agreement of automatic and manual scoring was again relatively low (kappa = 0.331 for {wake light-sleep / deep-sleep REM} and kappa = 0.505 for {wake / sleep}), whereas inter-scorer reliability was higher (kappa = -0.641 for {wake / light-sleep / deep-sleep / REM} and kappa = 0.737 for {wake / sleep}). We conclude that such an ANN-based analysis system is not sufficiently accurate for sleep study analyses using the R&K classification system.

Adult↗
PubMed1994

Integration of quantitative and qualitative reasoning: an expert system for cardiosurgical patients.

In this work the possibility of building an expert system to reason on the status of post-operative cardiac patients in intensive care units is analysed. The long-term knowledge consists of causal network which describes the main relationships between hemodynamic and metabolic quantities involved in the evolution after cardiac surgery. The inference engine uses an original hybrid formalism, which integrates numerical simulation and qualitative methods. If available, the numerical values of quantities and their exact mathematical relationships are employed; otherwise, the inference engine reasons by using a discrete qualitative representation of quantities. Simulations performed using real data indicate that integration of quantitative and qualitative methods reduces the number of diagnostic scenarios compatible with patient data, and constitutes a valid tool for reasoning about physiological disorders in terms of deep causal knowledge.

Atrial Function, Right↗
PubMed2003

On the intersubject generalization ability in extracting kinematic information from afferent nervous signals.

In the recent past, many efforts have been carried out in order to evaluate the feasibility of implementing closed-loop controlled neuroprostheses based on the processing of sensory electroneurographic (ENG) signals. The success of these techniques mostly relies on the development of processing algorithms capable of extracting the necessary kinematic information from these signals. Soft-computing algorithms can be very useful when dealing with the complexity of the neuromuscular system because of their generalization ability and model-free structure. In this paper, these techniques were used to extract angular position information from the ENG signals recorded from muscle afferents in animal model using cuff electrodes. Specifically, a genetic algorithm-based dynamic nonsingleton fuzzy logic system (named GA-DNSFLS) was developed and tested on different types of angular trajectories (characterized by small or large angular excursions). In particular, two different Takagi-Sugeno-Kang (TSK)-like structures were used in the consequent part of the neuro-fuzzy model in order to verify which one could improve the generalization abilities (intrasubject and intersubject). The results showed that the GA-DNSFLS was able to reconstruct the trajectories giving interesting results in terms of correlation between the actual and the predicted trajectories for small excursion movements during intrasubject and intersubject tests. Particularly, one of the TSK models showed better results in terms of intersubject generalization. The simulations conducted with the large excursion movements led in some cases to interesting results but further experiments are necessary in order to analyze this point more in deep.

Action Potentials↗
PubMed1997

How neural interactions form neural responses in the salamander retina.

A wide range of experimental data characterizing properties of individual salamander retinal cells and synaptic interactions are integrated to form a quantitative computational model of visual function in the salamander retina. The model is used to show how specific interactions between neurons and between networks of neurons can lead-to the integrated response behavior of individual cells deep in the retina. The model is also used to illustrate how the representation of moving and stationary stimuli is encoded in a series of layer-by-layer transformations leading to the final retinal output at the ganglion cell layer.

Animals↗
PubMed2003

Deep brain stimulation for Parkinson's disease.

Deep brain stimulation at high frequency was first used in 1997 to replace thalamotomy in treating the characteristic tremor of Parkinson's disease, and has subsequently been applied to the pallidum and the subthalamic nucleus. The subthalamic nucleus is a key node in the functional control of motor activity in the basal ganglia. Its inhibition suppresses symptoms in animal models of Parkinson's disease, and high frequency chronic stimulation does the same in human patients. Acute and long-term results after deep brain stimulation show a dramatic and stable improvement of a patient's clinical condition, which mimics the effects of levodopa treatment. The mechanism of action may involve a functional disruption of the abnormal neural messages associated with the disease. Long-term changes, neural plasticity and neural protection might be induced in the network. Similar effects of stimulation and lesioning have led to the extension of this technique for other targets and diseases.

Electric Stimulation Therapy↗
PubMed2006

[Part I: Face recognition].

Faces represent a crucial vector of interhuman communication. The message transmitted by the face has multiple features. Recognition of each feature can be impaired independently or in combination with others. In order to understand the behavioral consequences of such impairments, which can be a major social handicap, we first must specify the neural networks involved in face recognition. We propose in this first part to present the systems involved in face recognition, in particular the question of identity and prosopagnosia. Different neural networks are indeed implicated in the recognition of invariant facial features such as identity, gender, ethnicity, and recognition of variant features like facial expression and eye gaze. This paper is illustrated by some of our scalp and intracranial electrophysiological studies performed in humans allowing us to describe some aspects of face recognition dynamics combining an excellent spatial and temporal resolution. Intracranial recordings were performed in drug refractory epileptical patients implanted with depth electrodes. These studies demonstrate that numerous deep brain and cortical structures participate early and sometimes in a sustained manner in face recognition.

Face↗
PubMed2005

Craniofacial inputs to upper cervical dorsal horn: implications for somatosensory information processing.

The aim of this study was to characterize the properties of somatosensory neurons in the first 2 cervical spinal dorsal horns (C1 and C2 DHs) and compare them with those previously described for the rostral subnucleus caudalis (rVc). A total of 74 nociceptive neurons classified as wide-dynamic-range (WDR) or nociceptive-specific (NS), as well as 72 low-threshold mechanoreceptive (LTM) neurons, was studied in urethane/chloralose-anesthetized rats. The majority of LTM neurons were located in laminae III/IV and had a small mechanoreceptive field (RF) that included the posterior face and cervical tissues. In contrast, the nociceptive neurons were located in laminae I/II or V/VI, and the RF of each C1 and C2 DH nociceptive neuron included a part of the face and in 47% of them the RF included a region supplied by upper cervical afferents. There was a gradual caudal shift in the neuronal RF from nasal/intraoral tissues towards the neck as recording sites progressed from rVc to C1 and C2 DHs. In contrast to LTM neurons, many C1 and C2 DH nociceptive neurons received mechanosensitive convergent afferent inputs from cervical and craniofacial deep tissues (e.g., tongue muscles or temporomandibular joint), and over 50% could be activated by hypoglossal (XII) nerve electrical stimulation. We propose that C1 and C2 DHs represent part of the caudal extension of the Vc, and that Vc and C1 and C2 DHs may act together as one functional unit to process nociceptive information from craniofacial and cervical tissues, including that from deep craniofacial tissues.

Afferent Pathways↗
PubMed2026

Accelerated long-read variant calling with Clair3 for whole-genome sequencing.

SUMMARY: The rapid growth of genomic data and increasing adoption of long-read sequencing technologies have rendered variant calling one of the most computationally demanding tasks in genomic analysis. Although deep learning-based methods currently outperform conventional approaches in distinguishing true variants from complex sequencing noise, they impose prohibitive computational and time requirements. To address this limitation, we present a computational framework based on Clair3 that integrates parallelized feature generation, enhanced variant phasing, in-memory read haplotagging, and GPU-accelerated neural network inference to accelerate variant calling. By dynamically optimizing the use of both GPU and CPU resources, our method achieves substantial runtime improvements without compromising accuracy. We evaluated our framework across a range of sequencing depths, diverse samples, and multiple hardware configurations. Our results demonstrate that the optimized pipeline completes variant calling for a 30× whole-genome sequence in 12-20 minutes using standard computational resources (32 CPU threads and one NVIDIA GPU), and in 12-15 minutes on an Apple Mac Studio (32 threads), which is ∼10-20-fold speedup compared with its initial release. In addition to exceptional efficiency, our method maintains state-of-the-art accuracy, achieving SNP F1-scores of 99.32% and 99.70% on 30× ONT and PacBio GIAB HG003 datasets, respectively. This work introduces a rapid, accurate, and scalable variant calling framework that effectively supports large-cohort genomic studies and time-sensitive clinical applications. AVAILABILITY AND IMPLEMENTATION: The accelerated implementation of Clair3 is open source and available at: https://github.com/HKU-BAL/Clair3/tree/gpu.

Whole Genome Sequencing↗