Evaluating Deep Neural Network Architectures with Transfer Learning for Pneumonitis Diagnosis.
Ontology highlight
ABSTRACT:
SUBMITTER: Krishnamurthy S
PROVIDER: S-EPMC8452401 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:ObjectiveSocial media is an important pharmacovigilance data source for adverse drug reaction (ADR) identification. Human review of social media data is infeasible due to data quantity, thus natural language processing techniques are necessary. Social media includes informal vocabulary and irregular grammar, which challenge natural language processing methods. Our objective is to develop a scalable, deep-learning approach that exceeds state-of-the-art ADR detection performance in social media.Materials and methodsWe developed a recurrent neural network (RNN) model that labels words in an input sequence with ADR membership tags. The only input features are word-embedding vectors, which can be formed through task-independent pretraining or during ADR detection training.ResultsOur best-performing RNN model used pretrained word embeddings created from a large, non-domain-specific Twitter dataset. It achieved an approximate match F-measure of 0.755 for ADR identification on the dataset, compared to 0.631 for a baseline lexicon system and 0.65 for the state-of-the-art conditional random field model. Feature analysis indicated that semantic information in pretrained word embeddings boosted sensitivity and, combined with contextual awareness captured in the RNN, precision.DiscussionOur model required no task-specific feature engineering, suggesting generalizability to additional sequence-labeling tasks. Learning curve analysis showed that our model reached optimal performance with fewer training examples than the other models.ConclusionADR detection performance in social media is significantly improved by using a contextually aware model and word embeddings formed from large, unlabeled datasets. The approach reduces manual data-labeling requirements and is scalable to large social media datasets.
| S-EPMC7651964 | biostudies-literature
Project description:PURPOSE:The nonalcoholic fatty liver disease is the most common liver abnormality. Up to date, liver biopsy is the reference standard for direct liver steatosis quantification in hepatic tissue samples. In this paper we propose a neural network-based approach for nonalcoholic fatty liver disease assessment in ultrasound. METHODS:We used the Inception-ResNet-v2 deep convolutional neural network pre-trained on the ImageNet dataset to extract high-level features in liver B-mode ultrasound image sequences. The steatosis level of each liver was graded by wedge biopsy. The proposed approach was compared with the hepatorenal index technique and the gray-level co-occurrence matrix algorithm. After the feature extraction, we applied the support vector machine algorithm to classify images containing fatty liver. Based on liver biopsy, the fatty liver was defined to have more than 5% of hepatocytes with steatosis. Next, we used the features and the Lasso regression method to assess the steatosis level. RESULTS:The area under the receiver operating characteristics curve obtained using the proposed approach was equal to 0.977, being higher than the one obtained with the hepatorenal index method, 0.959, and much higher than in the case of the gray-level co-occurrence matrix algorithm, 0.893. For regression the Spearman correlation coefficients between the steatosis level and the proposed approach, the hepatorenal index and the gray-level co-occurrence matrix algorithm were equal to 0.78, 0.80 and 0.39, respectively. CONCLUSIONS:The proposed approach may help the sonographers automatically diagnose the amount of fat in the liver. The presented approach is efficient and in comparison with other methods does not require the sonographers to select the region of interest.
| S-EPMC6223753 | biostudies-literature
Project description:BackgroundPredicting hepatocellular carcinoma (HCC) prognosis is important for treatment selection, and it is increasingly interesting to predict prognosis through gene expression data. Currently, the prognosis remains of low accuracy due to the high dimension but small sample size of liver cancer omics data. In previous studies, a transfer learning strategy has been developed by pre-training models on similar cancer types and then fine-tuning the pre-trained models on the target dataset. However, transfer learning has limited performance since other cancer types are similar at different levels, and it is not trivial to balance the relations with different cancer types.MethodsHere, we propose an adaptive transfer-learning-based deep Cox neural network (ATRCN), where cancers are represented by 12 phenotype and 10 genotype features, and suitable cancers were adaptively selected for model pre-training. In this way, the pre-trained model can learn valuable prior knowledge from other cancer types while reducing the biases.ResultsATRCN chose pancreatic and stomach adenocarcinomas as the pre-training cancers, and the experiments indicated that our method improved the C-index of 3.8% by comparing with traditional transfer learning methods. The independent tests on three additional HCC datasets proved the robustness of our model. Based on the divided risk subgroups, we identified 10 HCC prognostic markers, including one new prognostic marker, TTC36. Further wet experiments indicated that TTC36 is associated with the progression of liver cancer cells.ConclusionThese results proved that our proposed deep-learning-based method for HCC prognosis prediction is robust, accurate, and biologically meaningful.
| S-EPMC8504135 | biostudies-literature
Project description:Combining machine learning with neuroimaging data has a great potential for early diagnosis of mild cognitive impairment (MCI) and Alzheimer's disease (AD). However, it remains unclear how well the classifiers built on one population can predict MCI/AD diagnosis of other populations. This study aimed to employ a spectral graph convolutional neural network (graph-CNN), that incorporated cortical thickness and geometry, to identify MCI and AD based on 3089 T1-weighted MRI data of the ADNI-2 cohort, and to evaluate its feasibility to predict AD in the ADNI-1 cohort (n = 3602) and an Asian cohort (n = 347). For the ADNI-2 cohort, the graph-CNN showed classification accuracy of controls (CN) vs. AD at 85.8% and early MCI (EMCI) vs. AD at 79.2%, followed by CN vs. late MCI (LMCI) (69.3%), LMCI vs. AD (65.2%), EMCI vs. LMCI (60.9%), and CN vs. EMCI (51.8%). We demonstrated the robustness of the graph-CNN among the existing deep learning approaches, such as Euclidean-domain-based multilayer network and 1D CNN on cortical thickness, and 2D and 3D CNNs on T1-weighted MR images of the ADNI-2 cohort. The graph-CNN also achieved the prediction on the conversion of EMCI to AD at 75% and that of LMCI to AD at 92%. The find-tuned graph-CNN further provided a promising CN vs. AD classification accuracy of 89.4% on the ADNI-1 cohort and >90% on the Asian cohort. Our study demonstrated the feasibility to transfer AD/MCI classifiers learned from one population to the other. Notably, incorporating cortical geometry in CNN has the potential to improve classification performance.
| S-EPMC6627731 | biostudies-literature
Project description:Early diagnosis remains a significant challenge for many neurological disorders, especially for rare disorders where studying large cohorts is not possible. A novel solution that investigators have undertaken is combining advanced machine learning algorithms with resting-state functional Magnetic Resonance Imaging to unveil hidden pathological brain connectome patterns to uncover diagnostic and prognostic biomarkers. Recently, state-of-the-art deep learning techniques are outperforming traditional machine learning methods and are hailed as a milestone for artificial intelligence. However, whole brain classification that combines brain connectome with deep learning has been hindered by insufficient training samples. Inspired by the transfer learning strategy employed in computer vision, we exploited previously collected resting-state functional MRI data for healthy subjects from existing databases and transferred this knowledge for new disease classification tasks. We developed a deep transfer learning neural network (DTL-NN) framework for enhancing the classification of whole brain functional connectivity patterns. Briefly, we trained a stacked sparse autoencoder (SSAE) prototype to learn healthy functional connectivity patterns in an offline learning environment. Then, the SSAE prototype was transferred to a DTL-NN model for a new classification task. To test the validity of our framework, we collected resting-state functional MRI data from the Autism Brain Imaging Data Exchange (ABIDE) repository. Using autism spectrum disorder (ASD) classification as a target task, we compared the performance of our DTL-NN approach with a traditional deep neural network and support vector machine models across four ABIDE data sites that enrolled at least 60 subjects. As compared to traditional models, our DTL-NN approach achieved an improved performance in accuracy, sensitivity, specificity and area under receiver operating characteristic curve. These findings suggest that DTL-NN approaches could enhance disease classification for neurological conditions, where accumulating large neuroimaging datasets has been challenging.
| S-EPMC6066582 | biostudies-literature
Project description:The ability of deep learning architectures to identify glaucomatous optic neuropathy (GON) in fundus photographs was evaluated. A large database of fundus photographs (n = 14,822) from a racially and ethnically diverse group of individuals (over 33% of African descent) was evaluated by expert reviewers and classified as GON or healthy. Several deep learning architectures and the impact of transfer learning were evaluated. The best performing model achieved an overall area under receiver operating characteristic (AUC) of 0.91 in distinguishing GON eyes from healthy eyes. It also achieved an AUC of 0.97 for identifying GON eyes with moderate-to-severe functional loss and 0.89 for GON eyes with mild functional loss. A sensitivity of 88% at a set 95% specificity was achieved in detecting moderate-to-severe GON. In all cases, transfer improved performance and reduced training time. Model visualizations indicate that these deep learning models relied on, in part, anatomical features in the inferior and superior regions of the optic disc, areas commonly used by clinicians to diagnose GON. The results suggest that deep learning-based assessment of fundus images could be useful in clinical decision support systems and in the automation of large-scale glaucoma detection and screening programs.
| S-EPMC6232132 | biostudies-literature
Project description:Computer-aided diagnosis (CAD) systems hold potential to improve the diagnostic accuracy of thyroid ultrasound (US). We aimed to develop a deep learning-based US CAD system (dCAD) for the diagnosis of thyroid nodules and compare its performance with those of a support vector machine (SVM)-based US CAD system (sCAD) and radiologists. dCAD was developed by using US images of 4919 thyroid nodules from three institutions. Its diagnostic performance was prospectively evaluated between June 2016 and February 2017 in 286 nodules, and was compared with those of sCAD and radiologists, using logistic regression with the generalized estimating equation. Subgroup analyses were performed according to experience level and separately for small thyroid nodules 1-2?cm. There was no difference in overall sensitivity, specificity, positive predictive value (PPV), negative predictive value and accuracy (all p?>?0.05) between radiologists and dCAD. Radiologists and dCAD showed higher specificity, PPV, and accuracy than sCAD (all p?<?0.001). In small nodules, experienced radiologists showed higher specificity, PPV and accuracy than sCAD (all p?<?0.05). In conclusion, dCAD showed overall comparable diagnostic performance with radiologists and assessed thyroid nodules more effectively than sCAD, without loss of sensitivity.
| S-EPMC6882804 | biostudies-literature
Project description:Isolated dystonia is a neurological disorder of heterogeneous pathophysiology, which causes involuntary muscle contractions leading to abnormal movements and postures. Its diagnosis is remarkably challenging due to the absence of a biomarker or gold standard diagnostic test. This leads to a low agreement between clinicians, with up to 50% of cases being misdiagnosed and diagnostic delays extending up to 10.1 y. We developed a deep learning algorithmic platform, DystoniaNet, to automatically identify and validate a microstructural neural network biomarker for dystonia diagnosis from raw structural brain MRIs of 612 subjects, including 392 patients with three different forms of isolated focal dystonia and 220 healthy controls. DystoniaNet identified clusters in corpus callosum, anterior and posterior thalamic radiations, inferior fronto-occipital fasciculus, and inferior temporal and superior orbital gyri as the biomarker components. These regions are known to contribute to abnormal interhemispheric information transfer, heteromodal sensorimotor processing, and executive control of motor commands in dystonia pathophysiology. The DystoniaNet-based biomarker showed an overall accuracy of 98.8% in diagnosing dystonia, with a referral of 3.5% of cases due to diagnostic uncertainty. The diagnostic decision by DystoniaNet was computed in 0.36 s per subject. DystoniaNet significantly outperformed shallow machine-learning algorithms in benchmark comparisons, showing nearly a 20% increase in its diagnostic performance. Importantly, the microstructural neural network biomarker and its DystoniaNet platform showed substantial improvement over the current 34% agreement on dystonia diagnosis between clinicians. The translational potential of this biomarker is in its highly accurate, interpretable, and generalizable performance for enhanced clinical decision-making.
| S-EPMC7586425 | biostudies-literature
Project description:In computer-aided analysis of cardiac MRI data, segmentations of the left ventricle (LV) and myocardium are performed to quantify LV ejection fraction and LV mass, and they are performed after the identification of a short axis slice coverage, where automatic classification of the slice range of interest is preferable. Standard cardiac image post-processing guidelines indicate the importance of the correct identification of a short axis slice range for accurate quantification. We investigated the feasibility of applying transfer learning of deep convolutional neural networks (CNNs) as a means to automatically classify the short axis slice range, as transfer learning is well suited to medical image data where labeled data is scarce and expensive to obtain. The short axis slice images were classified into out-of-apical, apical-to-basal, and out-of-basal, on the basis of short axis slice location in the LV. We developed a custom user interface to conveniently label image slices into one of the three categories for the generation of training data and evaluated the performance of transfer learning in nine popular deep CNNs. Evaluation with unseen test data indicated that among the CNNs the fine-tuned VGG16 produced the highest values in all evaluation categories considered and appeared to be the most appropriate choice for the cardiac slice range classification.
| S-EPMC7815707 | biostudies-literature
Project description:OBJECTIVES:To develop a proof-of-concept "interpretable" deep learning prototype that justifies aspects of its predictions from a pre-trained hepatic lesion classifier. METHODS:A convolutional neural network (CNN) was engineered and trained to classify six hepatic tumor entities using 494 lesions on multi-phasic MRI, described in Part 1. A subset of each lesion class was labeled with up to four key imaging features per lesion. A post hoc algorithm inferred the presence of these features in a test set of 60 lesions by analyzing activation patterns of the pre-trained CNN model. Feature maps were generated that highlight regions in the original image that correspond to particular features. Additionally, relevance scores were assigned to each identified feature, denoting the relative contribution of a feature to the predicted lesion classification. RESULTS:The interpretable deep learning system achieved 76.5% positive predictive value and 82.9% sensitivity in identifying the correct radiological features present in each test lesion. The model misclassified 12% of lesions. Incorrect features were found more often in misclassified lesions than correctly identified lesions (60.4% vs. 85.6%). Feature maps were consistent with original image voxels contributing to each imaging feature. Feature relevance scores tended to reflect the most prominent imaging criteria for each class. CONCLUSIONS:This interpretable deep learning system demonstrates proof of principle for illuminating portions of a pre-trained deep neural network's decision-making, by analyzing inner layers and automatically describing features contributing to predictions. KEY POINTS:• An interpretable deep learning system prototype can explain aspects of its decision-making by identifying relevant imaging features and showing where these features are found on an image, facilitating clinical translation. • By providing feedback on the importance of various radiological features in performing differential diagnosis, interpretable deep learning systems have the potential to interface with standardized reporting systems such as LI-RADS, validating ancillary features and improving clinical practicality. • An interpretable deep learning system could potentially add quantitative data to radiologic reports and serve radiologists with evidence-based decision support.
| S-EPMC7243989 | biostudies-literature