An ensemble of neural networks provides expert-level prenatal detection of complex congenital heart disease.
Ontology highlight
ABSTRACT:
SUBMITTER: Arnaout R
PROVIDER: S-EPMC8380434 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:Objectives:Scoring laboratory polysomnography (PSG) data remains a manual task of visually annotating 3 primary categories: sleep stages, sleep disordered breathing, and limb movements. Attempts to automate this process have been hampered by the complexity of PSG signals and physiological heterogeneity between patients. Deep neural networks, which have recently achieved expert-level performance for other complex medical tasks, are ideally suited to PSG scoring, given sufficient training data. Methods:We used a combination of deep recurrent and convolutional neural networks (RCNN) for supervised learning of clinical labels designating sleep stages, sleep apnea events, and limb movements. The data for testing and training were derived from 10 000 clinical PSGs and 5804 research PSGs. Results:When trained on the clinical dataset, the RCNN reproduces PSG diagnostic scoring for sleep staging, sleep apnea, and limb movements with accuracies of 87.6%, 88.2% and 84.7% on held-out test data, a level of performance comparable to human experts. The RCNN model performs equally well when tested on the independent research PSG database. Only small reductions in accuracy were noted when training on limited channels to mimic at-home monitoring devices: frontal leads only for sleep staging, and thoracic belt signals only for the apnea-hypopnea index. Conclusions:By creating accurate deep learning models for sleep scoring, our work opens the path toward broader and more timely access to sleep diagnostics. Accurate scoring automation can improve the utility and efficiency of in-lab and at-home approaches to sleep diagnostics, potentially extending the reach of sleep expertise beyond specialty clinics.
| S-EPMC6289549 | biostudies-literature
Project description:ImportanceConvolutional neural networks (CNNs) achieve expert-level accuracy in the diagnosis of pigmented melanocytic lesions. However, the most common types of skin cancer are nonpigmented and nonmelanocytic, and are more difficult to diagnose.ObjectiveTo compare the accuracy of a CNN-based classifier with that of physicians with different levels of experience.Design, setting, and participantsA CNN-based classification model was trained on 7895 dermoscopic and 5829 close-up images of lesions excised at a primary skin cancer clinic between January 1, 2008, and July 13, 2017, for a combined evaluation of both imaging methods. The combined CNN (cCNN) was tested on a set of 2072 unknown cases and compared with results from 95 human raters who were medical personnel, including 62 board-certified dermatologists, with different experience in dermoscopy.Main outcomes and measuresThe proportions of correct specific diagnoses and the accuracy to differentiate between benign and malignant lesions measured as an area under the receiver operating characteristic curve served as main outcome measures.ResultsAmong 95 human raters (51.6% female; mean age, 43.4 years; 95% CI, 41.0-45.7 years), the participants were divided into 3 groups (according to years of experience with dermoscopy): beginner raters (<3 years), intermediate raters (3-10 years), or expert raters (>10 years). The area under the receiver operating characteristic curve of the trained cCNN was higher than human ratings (0.742; 95% CI, 0.729-0.755 vs 0.695; 95% CI, 0.676-0.713; P < .001). The specificity was fixed at the mean level of human raters (51.3%), and therefore the sensitivity of the cCNN (80.5%; 95% CI, 79.0%-82.1%) was higher than that of human raters (77.6%; 95% CI, 74.7%-80.5%). The cCNN achieved a higher percentage of correct specific diagnoses compared with human raters (37.6%; 95% CI, 36.6%-38.4% vs 33.5%; 95% CI, 31.5%-35.6%; P = .001) but not compared with experts (37.3%; 95% CI, 35.7%-38.8% vs 40.0%; 95% CI, 37.0%-43.0%; P = .18).Conclusions and relevanceNeural networks are able to classify dermoscopic and close-up images of nonpigmented lesions as accurately as human experts in an experimental setting.
| S-EPMC6439580 | biostudies-literature
Project description:Altered brain development is a common feature of the neurological sequelae of complex congenital heart disease (CHD). These alterations include abnormalities in brain size and growth that begin prenatally and persist postnatally. However, the longitudinal trajectory of changes in brain volume from the prenatal to postnatal environment have not been investigated. We aimed to evaluate the trajectory of brain growth in a cohort of patients with complex CHD (n?=?16) and healthy controls (n?=?15) to test the hypothesis that patients with complex CHD would have smaller total brain volume (TBV) prenatally, which would become increasingly prominent by three months of age. Participants underwent fetal magnetic resonance imaging (MRI) at a mean of 32?weeks gestation, a preoperative/neonatal MRI shortly after birth, a postoperative MRI (CHD only), and a 3-month MRI to evaluate the trajectory of brain growth. Three-dimensional volumetric analysis was applied to the MRI data to measure TBV, as well as tissue-specific volumes of the cortical gray matter (CGM), white matter (WM), subcortical (deep nuclear) gray matter (SCGM), cerebellum, and cerebrospinal fluid (CSF). A random coefficients model was used to investigate longitudinal changes in TBV and demonstrated an altered trajectory of brain growth in the CHD population. The estimated slope for TBV from fetal to 3-month MRI was 11.5?cm3 per week for CHD infants compared to 16.7?cm3 per week for controls (p?=?0.0002). Brain growth followed a similar trajectory for the CGM (p?<?0.0001), SCGM (p?=?0.002), and cerebellum (p?=?0.005). There was no difference in growth of the WM (p?=?0.30) or CSF (p?=?0.085). Brain injury was associated with reduced TBV at 3-month MRI (p?=?0.02). After removing infants with brain injury from the model, an altered trajectory of brain growth persisted in CHD infants (p?=?0.006). These findings extend the existing literature by demonstrating longitudinal impairments in brain development in the CHD population and emphasize the global nature of disrupted brain growth from the prenatal environment through early infancy.
| S-EPMC6178192 | biostudies-literature
Project description:The cardiac neural crest arises in the hindbrain, then migrates to the heart and contributes to critical structures, including the outflow tract septum. Chick cardiac crest ablation results in failure of this septation, phenocopying the human heart defect persistent truncus arteriosus (PTA), which trunk neural crest fails to rescue. Here, we probe the molecular mechanisms underlying the cardiac crest's unique potential. Transcriptional profiling identified cardiac-crest-specific transcription factors, with single-cell RNA sequencing revealing surprising heterogeneity, including an ectomesenchymal subpopulation within the early migrating population. Loss-of-function analyses uncovered a transcriptional subcircuit, comprised of Tgif1, Ets1, and Sox8, critical for cardiac neural crest and heart development. Importantly, ectopic expression of this subcircuit was sufficient to imbue trunk crest with the ability to rescue PTA after cardiac crest ablation. Together, our results reveal a transcriptional program sufficient to confer cardiac potential onto trunk neural crest cells, thus implicating new genes in cardiovascular birth defects.
| S-EPMC7255058 | biostudies-literature
Project description:ImportancePrenatal diagnosis of complex congenital heart defects reduces mortality and morbidity in affected infants. However, fetal echocardiography can be limited by poor acoustic windows, and there is a need for improved diagnostic methods.ObjectiveTo assess the clinical utility of fetal cardiovascular magnetic resonance imaging in cases in which fetal echocardiography could not visualize all relevant anatomy.Design, setting, and participantsThis cohort study was conducted between January 20, 2017, and June 29, 2020, at Skåne University Hospital (Lund, Sweden), a tertiary center for pediatric cardiology and thoracic surgery. Participants were fetuses referred for fetal cardiovascular magnetic resonance examination by a pediatric cardiologist after an inconclusive echocardiograph.ExposuresFetal cardiovascular magnetic resonance examination requested by the patient's pediatric cardiologist.Main outcomes and measuresAny change in patient management because of diagnostic information gained from fetal cardiovascular magnetic resonance imaging.ResultsA total of 31 fetuses underwent cardiovascular magnetic resonance examination at a median gestational age of 36 weeks (range, 31-39 weeks). Overall, fetal cardiovascular magnetic resonance imaging had clinical utility, affecting patient management and/or parental counseling in 26 cases (84%). For aortic arch anatomy including signs of coarctation (20 fetuses), fetal cardiovascular magnetic resonance imaging added diagnostic information in 16 cases (80%). For assessment of univentricular vs biventricular outcome in borderline left ventricle, unbalanced atrioventricular septal defect, and pulmonary atresia with intact ventricular septum (15 fetuses), fetal cardiovascular magnetic resonance imaging visualized intracardiac anatomy and ventricular function, allowing assessment of outcome in 13 cases (87%). In 4 fetuses with hypoplastic left heart syndrome, fetal cardiovascular magnetic resonance imaging helped delivery planning in 3 cases (75%). Finally, fetal cardiovascular magnetic resonance imaging provided valuable information for parental counseling in 21 cases (68%).Conclusions and relevanceIn this cohort study, fetal cardiovascular magnetic resonance imaging added clinically useful information to what was available from echocardiography. These findings suggest that fetal CMR has the potential to affect clinical decision-making in challenging cases of congenital heart defects with inconclusive data from echocardiography. Fetal cardiovascular magnetic resonance imaging showed an association with clinical decision-making, including mode of delivery and early postnatal care, as well as with parental counseling.
| S-EPMC8008290 | biostudies-literature
Project description:BackgroundDeep learning-based head and neck lymph node level (HN_LNL) autodelineation is of high relevance to radiotherapy research and clinical treatment planning but still underinvestigated in academic literature. In particular, there is no publicly available open-source solution for large-scale autosegmentation of HN_LNL in the research setting.MethodsAn expert-delineated cohort of 35 planning CTs was used for training of an nnU-net 3D-fullres/2D-ensemble model for autosegmentation of 20 different HN_LNL. A second cohort acquired at the same institution later in time served as the test set (n = 20). In a completely blinded evaluation, 3 clinical experts rated the quality of deep learning autosegmentations in a head-to-head comparison with expert-created contours. For a subgroup of 10 cases, intraobserver variability was compared to the average deep learning autosegmentation accuracy on the original and recontoured set of expert segmentations. A postprocessing step to adjust craniocaudal boundaries of level autosegmentations to the CT slice plane was introduced and the effect of autocontour consistency with CT slice plane orientation on geometric accuracy and expert rating was investigated.ResultsBlinded expert ratings for deep learning segmentations and expert-created contours were not significantly different. Deep learning segmentations with slice plane adjustment were rated numerically higher (mean, 81.0 vs. 79.6, p = 0.185) and deep learning segmentations without slice plane adjustment were rated numerically lower (77.2 vs. 79.6, p = 0.167) than manually drawn contours. In a head-to-head comparison, deep learning segmentations with CT slice plane adjustment were rated significantly better than deep learning contours without slice plane adjustment (81.0 vs. 77.2, p = 0.004). Geometric accuracy of deep learning segmentations was not different from intraobserver variability (mean Dice per level, 0.76 vs. 0.77, p = 0.307). Clinical significance of contour consistency with CT slice plane orientation was not represented by geometric accuracy metrics (volumetric Dice, 0.78 vs. 0.78, p = 0.703).ConclusionsWe show that a nnU-net 3D-fullres/2D-ensemble model can be used for highly accurate autodelineation of HN_LNL using only a limited training dataset that is ideally suited for large-scale standardized autodelineation of HN_LNL in the research setting. Geometric accuracy metrics are only an imperfect surrogate for blinded expert rating.
| S-EPMC9978473 | biostudies-literature
Project description:MotivationProtein model quality estimation, in many ways, informs protein structure prediction. Despite their tight coupling, existing model quality estimation methods do not leverage inter-residue distance information or the latest technological breakthrough in deep learning that has recently revolutionized protein structure prediction.ResultsWe present a new distance-based single-model quality estimation method called QDeep by harnessing the power of stacked deep residual neural networks (ResNets). Our method first employs stacked deep ResNets to perform residue-level ensemble error classifications at multiple predefined error thresholds, and then combines the predictions from the individual error classifiers for estimating the quality of a protein structural model. Experimental results show that our method consistently outperforms existing state-of-the-art methods including ProQ2, ProQ3, ProQ3D, ProQ4, 3DCNN, MESHI, and VoroMQA in multiple independent test datasets across a wide-range of accuracy measures; and that predicted distance information significantly contributes to the improved performance of QDeep.Availability and implementationhttps://github.com/Bhattacharya-Lab/QDeep.Supplementary informationSupplementary data are available at Bioinformatics online.
| S-EPMC7355297 | biostudies-literature
Project description:White matter hyperintensities (WMHs) represent the most common neuroimaging marker of cerebral small vessel disease (CSVD). The volume and location of WMHs are important clinical measures. We present a pipeline using deep fully convolutional network and ensemble models, combining U-Net, SE-Net, and multi-scale features, to automatically segment WMHs and estimate their volumes and locations. We evaluated our method in two datasets: a clinical routine dataset comprising 60 patients (selected from Chinese National Stroke Registry, CNSR) and a research dataset composed of 60 patients (selected from MICCAI WMH Challenge, MWC). The performance of our pipeline was compared with four freely available methods: LGA, LPA, UBO detector, and U-Net, in terms of a variety of metrics. Additionally, to access the model generalization ability, another research dataset comprising 40 patients (from Older Australian Twins Study and Sydney Memory and Aging Study, OSM), was selected and tested. The pipeline achieved the best performance in both research dataset and the clinical routine dataset with DSC being significantly higher than other methods (p < .001), reaching .833 and .783, respectively. The results of model generalization ability showed that the model trained on the research dataset (DSC = 0.736) performed higher than that trained on the clinical dataset (DSC = 0.622). Our method outperformed widely used pipelines in WMHs segmentation. This system could generate both image and text outputs for whole brain, lobar and anatomical automatic labeling WMHs. Additionally, software and models of our method are made publicly available at https://www.nitrc.org/projects/what_v1.
| S-EPMC8764480 | biostudies-literature
Project description:ObjectiveTwitter posts are now recognized as an important source of patient-generated data, providing unique insights into population health. A fundamental step toward incorporating Twitter data in pharmacoepidemiologic research is to automatically recognize medication mentions in tweets. Given that lexical searches for medication names suffer from low recall due to misspellings or ambiguity with common words, we propose a more advanced method to recognize them.Materials and methodsWe present Kusuri, an Ensemble Learning classifier able to identify tweets mentioning drug products and dietary supplements. Kusuri (, "medication" in Japanese) is composed of 2 modules: first, 4 different classifiers (lexicon based, spelling variant based, pattern based, and a weakly trained neural network) are applied in parallel to discover tweets potentially containing medication names; second, an ensemble of deep neural networks encoding morphological, semantic, and long-range dependencies of important words in the tweets makes the final decision.ResultsOn a class-balanced (50-50) corpus of 15 005 tweets, Kusuri demonstrated performances close to human annotators with an F1 score of 93.7%, the best score achieved thus far on this corpus. On a corpus made of all tweets posted by 112 Twitter users (98 959 tweets, with only 0.26% mentioning medications), Kusuri obtained an F1 score of 78.8%. To the best of our knowledge, Kusuri is the first system to achieve this score on such an extremely imbalanced dataset.ConclusionsThe system identifies tweets mentioning drug names with performance high enough to ensure its usefulness, and is ready to be integrated in pharmacovigilance, toxicovigilance, or more generally, public health pipelines that depend on medication name mentions.
| S-EPMC6857507 | biostudies-literature
Project description:BackgroundThere are still inconsistent conclusions about the association of prenatal alcohol drinking with congenital heart defects (CHDs). We conducted this meta-analysis to investigate the association between prenatal alcohol exposure and the risk of overall CHDs and the CHDs subtypes.MethodsCase-control and cohort studies published before March 2015 were searched through PubMed and Embase. Two authors independently extracted data and scored the study quality according to the Newcastle-0ttawa Scale. The pooled ORs and 95%CI were estimated using the random-effects model and heterogeneity was assessed by the Q test and I2 statistic.ResultsA total of 20 studies were finally included. The results provided no evidence of the association between prenatal alcohol exposure and the risk of overall CHDs (OR = 1.06, 95%CI = 0.93-1.22), ventricular septal defects (VSDs) (OR = 1.04, 95%CI = 0.86-1.25), or atrial septal defects (ASDs) (OR = 1.40, 95%CI = 0.88-2.23). However, prenatal alcohol drinking was marginally significantly associated with conotruncal defects (CTDs) (OR = 1.24, 95%CI = 0.97-1.59) and statistically significantly associated with d-Transposition of the Great Arteries (dTGA) (OR = 1.64, 95%CI = 1.17-2.30). Moreover, both prenatal heavy drinking and binge drinking have a strong association with overall CHDs (heavy drinking: OR = 3.76, 95%CI = 1.00-14.10; binge drinking: OR = 2.49, 95%CI = 1.04-5.97), and prenatal moderate drinking has a modest association with CTDs (OR = 1.35, 95%CI = 1.05-1.75) and dTGA (OR = 1.86, 95%CI = 1.09-3.20).ConclusionsIn conclusion, the results suggested that prenatal alcohol exposure was not associated with overall CHDs or some subtypes, whereas marginally significant association was found for CTDs and statistically significant association was found for dTGA. Further prospective studies with large population and better designs are needed to explore the association of prenatal alcohol exposure with CHDs including the subtypes in specific groups.
| S-EPMC4482023 | biostudies-literature