Site-specific automated contouring model generalisibiliy enhancement

Abstract Funding Acknowledgements Type of funding sources: None. Background Segmentation of cardiac structures in echocardiography is a pre-requisite for accurately assessing cardiac morphology and function. Manual or semi-automated segmentation are both routinely used in clinical practice, although these can be time-consuming, and can introduce high inter- and intra- operator variability resulting in decreased reproducibility. Effective contouring with no manual input has proven to be challenging due to variations in image quality, image noise, motion during the acquisition and the lack of a well-defined geometry. Methods This work proposes a coordinate regression method for automated left ventricle (LV) segmentation, presented in Figure 1 (a). The proposed method is based on a modified U-net architecture that outputs the likelihood of coordinates of landmark points. The obtained likelihood heatmaps are converted to 2D coordinates using a differentiable spatial to numerical transform. The model was trained and validated on UK multisite data (1383 subjects) comprising apical 2 and 4 chamber views for both contrast and non-contrast echocardiographic images. The Cardiac Acquisitions for Multi-structure Ultrasound Segmentation (CAMUS) echocardiographic image segmentation database was used to assess the performance of the proposed method acting as data from a new clinical site. The CAMUS dataset comprises apical 2 and 4 chamber views acquired from 500 patients with manually annotated cardiac structures for end-diastole and end-systole frames. The original CAMUS dataset was split into training (450 patients) and testing (50 patients), with manual contours being available only for the training dataset. Therefore, we used the CAMUS training dataset to both test and improve our model, by using a random sample of 100 studies as an independent testing dataset and the remaining 350 studies were used for retraining the initial model to improve performance for this dataset. Results The results obtained on the testing images are presented in Figure 1 (b). When the model was trained using no CAMUS data for the LV segmentation, a mean Dice coefficient of 0.890 and a median of 0.911 was obtained. Including 350 studies with the original 1383 UK dataset and retraining the same model improved the average Dice coefficient to 0.930 and the median to 0.939. The CAMUS dataset authors reported the best average Dice coefficient of 0.924 on the 50 CAMUS testing images, therefore the proposed points regression method introduces a promising alternative to mask-based segmentation models. Conclusions In conclusion, the auto-contouring framework has proven to be effective in terms of its performance and ability to generalise to new data. Furthermore, this work highlights the importance of both evaluating model performance on data from new clinical sites and also enhancing model performance. Abstract Figure.

Download Full-text

Effectiveness of transfer learning for enhancing tumor classification with a convolutional neural network on frozen sections

Scientific Reports ◽

10.1038/s41598-020-78129-0 ◽

2020 ◽

Vol 10 (1) ◽

Author(s):

Young-Gon Kim ◽

Sungchul Kim ◽

Cristina Eunbee Cho ◽

In Hye Song ◽

Hee Jin Lee ◽

...

Keyword(s):

Neural Network ◽

Deep Learning ◽

Convolutional Neural Network ◽

Transfer Learning ◽

Frozen Section ◽

Medical Center ◽

External Validation ◽

Model Performance ◽

Classification Model ◽

Training Dataset

AbstractFast and accurate confirmation of metastasis on the frozen tissue section of intraoperative sentinel lymph node biopsy is an essential tool for critical surgical decisions. However, accurate diagnosis by pathologists is difficult within the time limitations. Training a robust and accurate deep learning model is also difficult owing to the limited number of frozen datasets with high quality labels. To overcome these issues, we validated the effectiveness of transfer learning from CAMELYON16 to improve performance of the convolutional neural network (CNN)-based classification model on our frozen dataset (N = 297) from Asan Medical Center (AMC). Among the 297 whole slide images (WSIs), 157 and 40 WSIs were used to train deep learning models with different dataset ratios at 2, 4, 8, 20, 40, and 100%. The remaining, i.e., 100 WSIs, were used to validate model performance in terms of patch- and slide-level classification. An additional 228 WSIs from Seoul National University Bundang Hospital (SNUBH) were used as an external validation. Three initial weights, i.e., scratch-based (random initialization), ImageNet-based, and CAMELYON16-based models were used to validate their effectiveness in external validation. In the patch-level classification results on the AMC dataset, CAMELYON16-based models trained with a small dataset (up to 40%, i.e., 62 WSIs) showed a significantly higher area under the curve (AUC) of 0.929 than those of the scratch- and ImageNet-based models at 0.897 and 0.919, respectively, while CAMELYON16-based and ImageNet-based models trained with 100% of the training dataset showed comparable AUCs at 0.944 and 0.943, respectively. For the external validation, CAMELYON16-based models showed higher AUCs than those of the scratch- and ImageNet-based models. Model performance for slide feasibility of the transfer learning to enhance model performance was validated in the case of frozen section datasets with limited numbers.

Download Full-text

Automatic detection and segmentation of adenomatous colorectal polyps during colonoscopy using Mask R-CNN

Open Life Sciences ◽

10.1515/biol-2020-0055 ◽

2020 ◽

Vol 15 (1) ◽

pp. 588-596 ◽

Cited By ~ 1

Author(s):

Jie Meng ◽

Linyan Xue ◽

Ying Chang ◽

Jianguang Zhang ◽

Shilong Chang ◽

...

Keyword(s):

Deep Neural Networks ◽

Alimentary Tract ◽

Multiple Scale ◽

Colorectal Polyps ◽

Training Dataset ◽

Cad System ◽

Multicenter Trials ◽

Testing Dataset ◽

Adenoma Detection ◽

Aided Diagnosis

AbstractColorectal cancer (CRC) is one of the main alimentary tract system malignancies affecting people worldwide. Adenomatous polyps are precursors of CRC, and therefore, preventing the development of these lesions may also prevent subsequent malignancy. However, the adenoma detection rate (ADR), a measure of the ability of a colonoscopist to identify and remove precancerous colorectal polyps, varies significantly among endoscopists. Here, we attempt to use a convolutional neural network (CNN) to generate a unique computer-aided diagnosis (CAD) system by exploring in detail the multiple-scale performance of deep neural networks. We applied this system to 3,375 hand-labeled images from the screening colonoscopies of 1,197 patients; of whom, 3,045 were assigned to the training dataset and 330 to the testing dataset. The images were diagnosed simply as either an adenomatous or non-adenomatous polyp. When applied to the testing dataset, our CNN-CAD system achieved a mean average precision of 89.5%. We conclude that the proposed framework could increase the ADR and decrease the incidence of interval CRCs, although further validation through large multicenter trials is required.

Download Full-text

Fully-automated root image analysis (faRIA)

Scientific Reports ◽

10.1038/s41598-021-95480-y ◽

2021 ◽

Vol 11 (1) ◽

Author(s):

Narendra Narisetti ◽

Michael Henke ◽

Christiane Seiler ◽

Astrid Junker ◽

Jörn Ostermann ◽

...

Keyword(s):

Near Infrared ◽

Climatic Factors ◽

Plant Root ◽

Ground Truth ◽

Parameter Tuning ◽

Dice Coefficient ◽

Powerful Analytical Tool ◽

Manual Interaction ◽

And Function ◽

Noisy Background

AbstractHigh-throughput root phenotyping in the soil became an indispensable quantitative tool for the assessment of effects of climatic factors and molecular perturbation on plant root morphology, development and function. To efficiently analyse a large amount of structurally complex soil-root images advanced methods for automated image segmentation are required. Due to often unavoidable overlap between the intensity of fore- and background regions simple thresholding methods are, generally, not suitable for the segmentation of root regions. Higher-level cognitive models such as convolutional neural networks (CNN) provide capabilities for segmenting roots from heterogeneous and noisy background structures, however, they require a representative set of manually segmented (ground truth) images. Here, we present a GUI-based tool for fully automated quantitative analysis of root images using a pre-trained CNN model, which relies on an extension of the U-Net architecture. The developed CNN framework was designed to efficiently segment root structures of different size, shape and optical contrast using low budget hardware systems. The CNN model was trained on a set of 6465 masks derived from 182 manually segmented near-infrared (NIR) maize root images. Our experimental results show that the proposed approach achieves a Dice coefficient of 0.87 and outperforms existing tools (e.g., SegRoot) with Dice coefficient of 0.67 by application not only to NIR but also to other imaging modalities and plant species such as barley and arabidopsis soil-root images from LED-rhizotron and UV imaging systems, respectively. In summary, the developed software framework enables users to efficiently analyse soil-root images in an automated manner (i.e. without manual interaction with data and/or parameter tuning) providing quantitative plant scientists with a powerful analytical tool.

Download Full-text

Predicting endoscopic remission in Crohn’s disease by the modified multiplier SES-CD (MM-SES-CD)

Gut ◽

10.1136/gutjnl-2020-323799 ◽

2021 ◽

pp. gutjnl-2020-323799

Author(s):

Neeraj Narula ◽

Emily C L Wong ◽

Jean-Frederic Colombel ◽

William J Sandborn ◽

John Kenneth Marshall ◽

...

Keyword(s):

Logistic Regression ◽

Crohn’S Disease ◽

Crohn's Disease ◽

Assessment Tool ◽

Training Dataset ◽

Mucosal Inflammation ◽

Youden Index ◽

Active Therapy ◽

Testing Dataset ◽

Endoscopic Remission

Background and aimsThe Simple Endoscopic Score for Crohn’s disease (SES-CD) is the primary tool for measurement of mucosal inflammation in clinical trials but lacks prognostic potential. We set to develop and validate a modified multiplier of the SES-CD (MM-SES-CD), which takes into consideration each individual parameter’s prognostic value for achieving endoscopic remission (ER) while on active therapy.MethodsIn this posthoc analysis of three CD clinical trial programmes (n=350 patients, baseline SES-CD ≥ 3 with confirmed ulceration), data were pooled and randomly split into a 70% training and 30% testing cohort. The MM-SES-CD was designed using weights for individual parameters as determined by logistic regression modelling, with 1-year ER (SES-CD < 3) being the dependent variable. A cut point score for low and high probability of ER was determined by using the maximum Youden Index and validated in the testing cohort.ResultsBaseline ulcer size, extent of ulceration and presence of non-passable strictures had the strongest association with 1-year ER as compared with affected surface area, with differential weighting of individual parameters across disease segments being observed during logistic regression. The MM-SES-CD was generated using this weighted regression model and demonstrated strong discrimination for ER in the training dataset (area under the receiver operator curve (AUC) 0.83, 95% CI 0.78 to 0.94) and in the testing dataset (AUC 0.82, 95% CI 0.77 to 0.92). In comparison to the MM-SES-CD scoring model, the original SES-CD score lacks accuracy (AUC 0.60, 95% CI 0.55 to 0.65) for predicting the achievement of ER.ConclusionsWe developed and internally validated the MM-SES-CD as an endoscopic severity assessment tool to predict one-year ER in patients with CD on active therapy.

Download Full-text

A Novel Model Based on Deep Convolutional Neural Network Improves Diagnostic Accuracy of Intramucosal Gastric Cancer (With Video)

Frontiers in Oncology ◽

10.3389/fonc.2021.622827 ◽

2021 ◽

Vol 11 ◽

Author(s):

Dehua Tang ◽

Jie Zhou ◽

Lei Wang ◽

Muhan Ni ◽

Min Chen ◽

...

Keyword(s):

Gastric Cancer ◽

Diagnostic Accuracy ◽

Diagnostic Performance ◽

Training Dataset ◽

Deep Convolutional Neural Networks ◽

Testing Dataset ◽

Kappa Value ◽

The Mean ◽

Gastric Cancer Patients ◽

Novel Model

Background and AimsPrediction of intramucosal gastric cancer (GC) is a big challenge. It is not clear whether artificial intelligence could assist endoscopists in the diagnosis.MethodsA deep convolutional neural networks (DCNN) model was developed via retrospectively collected 3407 endoscopic images from 666 gastric cancer patients from two Endoscopy Centers (training dataset). The DCNN model’s performance was tested with 228 images from 62 independent patients (testing dataset). The endoscopists evaluated the image and video testing dataset with or without the DCNN model’s assistance, respectively. Endoscopists’ diagnostic performance was compared with or without the DCNN model’s assistance and investigated the effects of assistance using correlations and linear regression analyses.ResultsThe DCNN model discriminated intramucosal GC from advanced GC with an AUC of 0.942 (95% CI, 0.915–0.970), a sensitivity of 90.5% (95% CI, 84.1%–95.4%), and a specificity of 85.3% (95% CI, 77.1%–90.9%) in the testing dataset. The diagnostic performance of novice endoscopists was comparable to those of expert endoscopists with the DCNN model’s assistance (accuracy: 84.6% vs. 85.5%, sensitivity: 85.7% vs. 87.4%, specificity: 83.3% vs. 83.0%). The mean pairwise kappa value of endoscopists was increased significantly with the DCNN model’s assistance (0.430–0.629 vs. 0.660–0.861). The diagnostic duration reduced considerably with the assistance of the DCNN model from 4.35s to 3.01s. The correlation between the perseverance of effort and diagnostic accuracy of endoscopists was diminished using the DCNN model (r: 0.470 vs. 0.076).ConclusionsAn AI-assisted system was established and found useful for novice endoscopists to achieve comparable diagnostic performance with experts.

Download Full-text

On the Effectiveness of Bayesian AutoML methods for Physics Emulators

10.20944/preprints202010.0595.v2 ◽

2020 ◽

Author(s):

Peetak Mitra ◽

Niccolò Dal Santo ◽

Majid Haghshenas ◽

Shounak Mitra ◽

Conor Daly ◽

...

Keyword(s):

Neural Networks ◽

Network Architecture ◽

Model Performance ◽

Parameter Tuning ◽

Bayesian Optimization ◽

Eddy Simulation ◽

Large Eddy ◽

Frame Work ◽

Network Weight ◽

And Function

The adoption of Machine Learning (ML) for building emulators for complex physical processes has seen an exponential rise in the recent years. While neural networks are good function approximators, optimizing the hyper-parameters of the network to reach a global minimum is not trivial, and often needs human knowl- edge and expertise. In this light, automatic ML or autoML methods have gained large interest as they automate the process of network hyper-parameter tuning. In addition, Neural Architecture Search (NAS) has shown promising outcomes for improving model performance. While autoML methods have grown in popularity for image, text and other applications, their effectiveness for high-dimensional, complex scientific datasets remains to be investigated. In this work, a data driven emulator for turbulence closure terms in the context of Large Eddy Simulation (LES) models is trained using Artificial Neural Networks and an autoML frame- work based on Bayesian Optimization, incorporating priors to jointly optimize the hyper-parameters as well as conduct a full neural network architecture search to converge to a global minima, is proposed. Additionally, we compare the effect of using different network weight initializations and optimizers such as ADAM, SGDM and RMSProp, to explore the best performing setting. Weight and function space similarities during the optimization trajectory are investigated, and critical differences in the learning process evolution are noted and compared to theory. We observe ADAM optimizer and Glorot initialization consistently performs better, while RMSProp outperforms SGDM as the latter appears to have been stuck at a local minima. Therefore, this autoML BayesOpt framework provides a means to choose the best hyper-parameter settings for a given dataset.

Download Full-text

Emotion Detection using Social Media and Machine Learning

International Journal for Research in Applied Science and Engineering Technology ◽

10.22214/ijraset.2021.36117 ◽

2021 ◽

Vol 9 (VI) ◽

pp. 4491-4494

Author(s):

Mr. Bhavar Shivam S.

Keyword(s):

Machine Learning ◽

Social Networking ◽

Social Networking Sites ◽

Training Dataset ◽

Emotion Detection ◽

Pos Tagging ◽

Testing Dataset ◽

Svm Algorithm ◽

The Right ◽

Negative Sentiment

Today we do a lot of things online from shopping to data sharing on social networking sites. Social networking (SNS) is good for releasing stress and depression by sharing one’s thoughts. Thus, emotion detection has become a hot trend to day. But there is a problem in analyzing emotions on a SNS like twitter as it generates lakhs of tweets each day and it is hard to keep track of the emotion behind each tweet as it is impossible for a human being to read and decide the emotions behind tweets. So, to help understand behind the texts in a SNS site we thought of designing a project which will keep track of the tweets and predict the right emotion behind the tweets whether they have a positive or a negative sentiment behind them. This thought of project can be achieved by a integration of SNS with NLP and machine learning together. For SNS we will use Twitter as it generates a lot of data which is accessible freely using an API. First, we will enter a keyword and fetch tweets from the twitter. Then stop words will be removed from these tweets using NLTK stop words database. Then the tweets will be passed for POS tagging and only right form of grammatical words will be kept and others will be removed. Then we create a training dataset with two types positive and negative. Then SVM algorithm will be trained using this training dataset. Then each tweet will be passed to the SVM as testing dataset which in turn will return classification of each tweet as a whole in two classes positive and negative. Thus, our application will be helpful in recognizing emotion behind a tweet.

Download Full-text

A Two-Layer Learning Architecture for Multi-Class Protein Folds Classification

Bioinformatics ◽

10.4018/978-1-4666-3604-0.ch041 ◽

2013 ◽

pp. 786-797

Author(s):

Ruofei Wang ◽

Xieping Gao

Keyword(s):

Query Protein ◽

Training Dataset ◽

Support Vector ◽

K Nearest Neighbors ◽

Protein Folds ◽

Testing Dataset ◽

Ensemble Strategy ◽

Component Classifier ◽

Independent Testing Dataset

Classification of protein folds plays a very important role in the protein structure discovery process, especially when traditional sequence alignment methods fail to yield convincing structural homologies. In this chapter, we have developed a two-layer learning architecture, named TLLA, for multi-class protein folds classification. In the first layer, OET-KNN (Optimized Evidence-Theoretic K Nearest Neighbors) is used as the component classifier to find the most probable K-folds of the query protein. In the second layer, we use support vector machine (SVM) to build the multi-class classifier just on the K-folds, generated in the first layer, rather than on all the 27 folds. For multi-feature combination, ensemble strategy based on voting is selected to give the final classification result. The standard percentage accuracy of our method at ~63% is achieved on the independent testing dataset, where most of the proteins have <25% sequence identity with those in the training dataset. The experimental evaluation based on a widely used benchmark dataset has shown that our approach outperforms the competing methods, implying our approach might become a useful vehicle in the literature.

Download Full-text

Impact of Body Mass Index (BMI) on the Effect of a Lactobacillus Rhamnosus GG (LGG)/Bifidobacterium Animalis Subspecies Lactis BB-12 (BB-12) Combination on Gut Microbiota (P20-023-19)

Current Developments in Nutrition ◽

10.1093/cdn/nzz040.p20-023-19 ◽

2019 ◽

Vol 3 (Supplement_1) ◽

Author(s):

Debra Poutsiaka ◽

Lori Stern ◽

Virginia Riquelme ◽

Emily Hollister ◽

Julia Cope ◽

...

Keyword(s):

Community Structure ◽

Gut Microbiota ◽

Lactobacillus Rhamnosus ◽

Detailed Examination ◽

Global Changes ◽

Shotgun Metagenomics ◽

Gene Abundance ◽

Funding Sources ◽

And Function ◽

Relative Gene

Abstract Objectives This exploratory study builds upon an earlier study of probiotic supplementation1 to assess the effects of a probiotic combination (P) of LGG and BB-12 on human gut microbiota composition and function, and to uncover an association with BMI. Methods Healthy subjects ingested P for 21 days (n = 18, P group) or did not (n = 7, C group). Fecal samples obtained at baseline (D_0) and after 21 days of supplementation (D_21) underwent 16S ribosomal RNA gene and shotgun metagenomics sequencing to characterize the bacterial and archaeal communities to the genus/species level and identify functional community genes. Results Following P ingestion, no global differences in microbiota community structure or relative gene abundance were detected. In targeted analyses, the abundances of LGG and BB-12 in the P group at D_21 increased in a statistically significant manner as the BMI decreased (Spearman correlation, P = 0.04 and P = 0.01, respectively). The relative abundance of LGG but not BB-12 appeared increased in P subjects at D_21 with BMI < 25 compared to BMI > 25 (P = 0.09). P group subjects with BMI < 25 demonstrated trends toward or statistically significant increases in the relative abundances of 5 genes involved with flagellar structure (KEGG orthologs K02422, P = 0.04; K03406, P = 0.06; K02407, P = 0.08; K02397, P = 0.08; K02396, P = 0.09) at D_21 compared to those with BMI > 25. No such differences were observed for the C group nor were there differences in relative gene abundance at D_0 in the P group with BMI < 25 vs BMI > 25. Conclusions We observed no global changes in the fecal microbial community structure or function with P ingestion in this sample of healthy persons. However, we did observe patterns suggestive of a potential link between BMI and the response of the gut microbiota to P. Although our results are based on a small number of subjects, they are in line with previous findings related to LGG supplementation and the expression of flagellar genes2. We agree with other recent reports that future studies would benefit from a detailed examination of the transcriptome, proteome and/or metabolome to better understand the potential impact of probiotics on the gut microbiota, and the mechanism of the effect of BMI. Funding Sources Pfizer Inc.

Download Full-text

Cross-domain gearbox diagnostics under variable working conditions with deep convolutional transfer learning

Journal of Vibration and Control ◽

10.1177/1077546320933793 ◽

2020 ◽

pp. 107754632093379

Author(s):

Moslem Azamfar ◽

Jaskaran Singh ◽

Xiang Li ◽

Jay Lee

Keyword(s):

Fault Diagnosis ◽

Transfer Learning ◽

High Performance ◽

Training Sample ◽

Training Dataset ◽

Target Domain ◽

Training Sample Size ◽

Vibration Data ◽

Cross Domain ◽

Testing Dataset

This study proposes a novel 1D deep convolutional transfer learning method that is able to learn the high-dimensional domain-invariant feature from the labeled training dataset and perform diagnosis tasks on the unlabeled testing dataset subjected to a domain shift. To obtain the domain-invariant features, the cross-entropy loss in the source domain classifier and the maximum mean discrepancies between the source and target domain data are minimized simultaneously. To evaluate the performance of the proposed method, an experimental study is conducted on a gearbox under significant speed variation. Because of inherent limitations of the vibration data, in this research, the effectiveness of torque measurement signals has been explored for gearbox fault diagnosis. Comprehensive studies on network parameters and the training sample size are performed to illustrate the robustness and effectiveness of the proposed method. A comparison study is performed on similar techniques to illustrate the superiority and high performance of the proposed diagnosis method. The achieved results illustrate the effectiveness of torque signal in multiclass cross-domain fault diagnosis of gearboxes.

Download Full-text