scholarly journals Deep-learning-based membranous nephropathy classification and Monte-Carlo dropout uncertainty estimation

2021 ◽  
Author(s):  
Paulo Chagas ◽  
Luiz Souza ◽  
Izabelle Pontes ◽  
Rodrigo Calumby ◽  
Michele Angelo ◽  
...  
Keyword(s):  
Monte Carlo ◽  
Nephrotic Syndrome ◽  
Deep Learning ◽  
Membranous Nephropathy ◽  
Uncertainty Estimation ◽  
Glomerular Diseases ◽  
Uncertainty Estimates ◽  
Model Predictions

Membranous Nephropathy (MN) is one of the most common glomerular diseases that cause adult nephrotic syndrome. To assist pathologists on MN classification, we evaluated three deep-learning-based architectures, namely, ResNet-18, DenseNet and Wide-ResNet. In addition, to accomplish more reliable results, we applied Monte-Carlo Dropout for uncertainty estimation. We achieved average F1-Scores above 92% for all models, with Wide-ResNet obtaining the highest average F1-Score (93.2%). For uncertainty estimation on Wide-ResNet, the uncertainty scores showed high relation with incorrect classifications, proving that these uncertainty estimates can support pathologists on the analysis of model predictions.

scholarly journals Quantifying the uncertainty of deep learning-based computer-aided diagnosis for patient safety

2019 ◽  
Vol 5 (1) ◽  
pp. 223-226
Author(s):  
Max-Heinrich Laves ◽  
Sontje Ihler ◽  
Tobias Ortmaier ◽  
Lüder A. Kahrs
Keyword(s):  
Monte Carlo ◽  
Patient Safety ◽  
Deep Learning ◽  
Epistemic Uncertainty ◽  
Computer Aided Diagnosis ◽  
Uncertainty Estimation ◽  
Test Time ◽  
Prediction Uncertainty ◽  
Computer Aided ◽  
Aided Diagnosis

AbstractIn this work, we discuss epistemic uncertainty estimation obtained by Bayesian inference in diagnostic classifiers and show that the prediction uncertainty highly correlates with goodness of prediction. We train the ResNet-18 image classifier on a dataset of 84,484 optical coherence tomography scans showing four different retinal conditions. Dropout is added before every building block of ResNet, creating an approximation to a Bayesian classifier. Monte Carlo sampling is applied with dropout at test time for uncertainty estimation. In Monte Carlo experiments, multiple forward passes are performed to get a distribution of the class labels. The variance and the entropy of the distribution is used as metrics for uncertainty. Our results show strong correlation with ρ = 0.99 between prediction uncertainty and prediction error. Mean uncertainty of incorrectly diagnosed cases was significantly higher than mean uncertainty of correctly diagnosed cases. Modeling of the prediction uncertainty in computer-aided diagnosis with deep learning yields more reliable results and is therefore expected to increase patient safety. This will help to transfer such systems into clinical routine and to increase the acceptance of machine learning in diagnosis from the standpoint of physicians and patients.

scholarly journals Ranking Information Extracted from Uncertainty Quantification of the Prediction of a Deep Learning Model on Medical Time Series Data

Mathematics ◽  
2020 ◽  
Vol 8 (7) ◽  
pp. 1078
Author(s):  
Ruxandra Stoean ◽  
Catalin Stoean ◽  
Miguel Atencia ◽  
Roberto Rodríguez-Labrada ◽  
Gonzalo Joya
Keyword(s):  
Time Series ◽  
Monte Carlo ◽  
Deep Learning ◽  
Uncertainty Quantification ◽  
Time Series Data ◽  
Short Term Memory ◽  
Learning Model ◽  
Series Data ◽  
Real World Problem ◽  
Uncertainty Estimates

Uncertainty quantification in deep learning models is especially important for the medical applications of this complex and successful type of neural architectures. One popular technique is Monte Carlo dropout that gives a sample output for a record, which can be measured statistically in terms of average probability and variance for each diagnostic class of the problem. The current paper puts forward a convolutional–long short-term memory network model with a Monte Carlo dropout layer for obtaining information regarding the model uncertainty for saccadic records of all patients. These are next used in assessing the uncertainty of the learning model at the higher level of sets of multiple records (i.e., registers) that are gathered for one patient case by the examining physician towards an accurate diagnosis. Means and standard deviations are additionally calculated for the Monte Carlo uncertainty estimates of groups of predictions. These serve as a new collection where a random forest model can perform both classification and ranking of variable importance. The approach is validated on a real-world problem of classifying electrooculography time series for an early detection of spinocerebellar ataxia 2 and reaches an accuracy of 88.59% in distinguishing between the three classes of patients.

scholarly journals Horseshoe kidney with PLA2R-positive membranous nephropathy

2021 ◽  
Vol 22 (1) ◽  
Author(s):  
Shuai-Shuai Shi ◽  
Xian-Zu Yang ◽  
Xiao-Ye Zhang ◽  
Hui-Dan Guo ◽  
Wen-Feng Wang ◽  
...  
Keyword(s):  
Nephrotic Syndrome ◽  
Renal Biopsy ◽  
Membranous Nephropathy ◽  
Horseshoe Kidney ◽  
Glomerular Disease ◽  
Urinary Stones ◽  
Abdominal Ultrasonography ◽  
Glomerular Diseases ◽  
Structural Abnormalities ◽  
Diagnosis Rate

Abstract Background Horseshoe kidney (HSK) is a common congenital defect of the urinary system. The most common complications are urinary tract infection, urinary stones, and hydronephrosis. HSK can be combined with glomerular diseases, but the diagnosis rate of renal biopsy is low due to structural abnormalities. There are only a few reports on HSK with glomerular disease. Here, we have reported a case of PLA2R-positive membranous nephropathy occurring in a patient with HSK. Case presentation After admission to the hospital due to oedema of both the lower extremities, the patient was diagnosed with nephrotic syndrome due to abnormal 24-h urine protein (7540 mg) and blood albumin (25 g/L) levels. Abdominal ultrasonography revealed HSK. The patient’s brother had a history of end-stage renal disease due to nephrotic syndrome. Therefore, the patient was diagnosed with PLA2R-positive stage II membranous nephropathy through renal biopsy under abdominal ultrasonography guidance. He was administered adequate prednisone and cyclophosphamide, and after 6 months of treatment, urinary protein excretion levels significantly decreased. Conclusion The risk and difficulty of renal biopsy in patients with HSK are increased due to structural abnormalities; however, renal biopsy can be accomplished through precise positioning with abdominal ultrasonography. In the literature, 20 cases of HSK with glomerular disease have been reported thus far. Because of the small number of cases, estimating the incidence rate of glomerular diseases in HSK is impossible, and the correlation between HSK and renal pathology cannot be stated. Further studies should be conducted and cases should be accumulated to elucidate this phenomenon.

scholarly journals Topological Analysis in Monte Carlo Simulation for Uncertainty Estimation

2019 ◽  
Author(s):  
Evren Pakyuz-Charrier ◽  
Mark Jessell ◽  
Jérémie Giraud ◽  
Mark Lindsay ◽  
Vitaliy Ogarko
Keyword(s):  
Monte Carlo Simulation ◽  
Monte Carlo ◽  
Input Data ◽  
Topological Analysis ◽  
Uncertainty Estimation ◽  
Population Heterogeneity ◽  
Underlying Assumption ◽  
3D Geological Modeling ◽  
Plausible Model ◽  
Uncertainty Estimates

Abstract. This paper proposes and demonstrates improvements for the Monte Carlo simulation for Uncertainty Estimation (MCUE) method. MCUE is a type of Bayesian Monte Carlo aimed at input data uncertainty propagation in implicit 3D geological modeling. In the Monte Carlo process, a series of statistically plausible models are built from the input data set which uncertainty is to be propagated to a final probabilistic geological model (PGM) or uncertainty index model (UIM). Significant differences in terms of topology are observed in the plausible model suite that is generated as an intermediary step in MCUE. These differences are interpreted as analogous to population heterogeneity. The source of this heterogeneity is traced to be the non-linear relationship between plausible datasets’ variability and plausible model’s variability. Non-linearity is shown to arise from the effect of the geometrical ruleset on model building which transforms lithological continuous interfaces into discontinuous piecewise ones. Plausible model heterogeneity induces geological incompatibility and challenges the underlying assumption of homogeneity which global uncertainty estimates rely on. To address this issue, a method for topological analysis applied to the plausible model suite in MCUE is introduced. Boolean topological signatures recording lithological units’ adjacency are used as n-dimensional points to be considered individually or clustered using the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm. The proposed method is tested on two challenging synthetic examples with varying levels of confidence in the structural input data. Results indicate that topological signatures constitute a powerful discriminant to address plausible model heterogeneity. Basic topological signatures appear to be a reliable indicator of the structural behavior of the plausible models and provide useful geological insights. Moreover, ignoring heterogeneity was found to be detrimental to the accuracy and relevance of the PGMs and UIMs.

scholarly journals Morphologic lesions of membranous nephropathy in association with various demographic and laboratory parameters of patients; a single center study

2019 ◽  
Vol 9 (1) ◽  
pp. e09-e09
Author(s):  
Sara Zamani ◽  
Hamid Nasri
Keyword(s):  
Nephrotic Syndrome ◽  
Serum Creatinine ◽  
Renal Biopsy ◽  
Membranous Nephropathy ◽  
Interstitial Fibrosis ◽  
Tubular Atrophy ◽  
Glomerular Diseases ◽  
Laboratory Parameters ◽  
Positive Correlation ◽  
End Stage

Introduction: Nephrotic syndrome is an important clinical presentation of glomerular diseases that is classified into several types based on the findings of renal biopsy. Membranous neuropathy is the most common cause of nephrotic syndrome, especially in adults over 40 years of age, which may lead to end-stage renal failure. Objectives: The present study aimed to assess the association of morphologic lesions of membranous nephropathy (MN) on renal biopsy with various demographic and laboratory parameters of the patients. Patients and Methods: This study was performed on renal biopsies, which were referred to the laboratory with the diagnosis of MN. To reach a definite diagnosis of MN, an immunofluorescence study (IgG, IgA, IgM, C1q and C3 antibody deposits) was conducted for all patients. Light microscopy was conducted to categorize the morphologic lesions of the glomeruli and interstitial area. The percentage of interstitial fibrosis/tubular atrophy was assessed too. Additionally, age, gender, and 24- hour urinary protein and serum creatinine were recorded. Results: Among 175 idiopathic MN patients, 98 were male (56%). The patients’ age was between 14 and 84 years (mean; 42±15 years). The mean of serum creatinine and 24-hour urine protein were 1.05 ± 0.31 mg/dL and 2779.56± 1495.80 mg/d, respectively. We found a significant correlation between gender and serum creatinine level, which was higher in men (P<0.001). Moreover, there was a significant, positive correlation between serum creatinine and age of patients (P<0.001, r=0.25). Additionally, there was a significant correlation between serum creatinine and interstitial fibrosis (P=0.001). We found a significant correlation between serum creatinine and the pathologic stage of glomeruli (P=0.003). The stages of glomeruli were also associated with the proportion of interstitial fibrosis (P=0.001) and C3 deposition rate (P=0.002). IgG deposition score was also significantly different in age ranges over and under 40 years of age (P=0.001). The 24-hours proteinuria had no correlation with other laboratory parameters and microscopic findings. Conclusion: In accordance with other studies, we found that MN is more common among male patients. The positive correlation between serum creatinine and proportion of interstitial fibrosis is in concordance with previous studies. We found a positive correlation between serum creatinine and glomerular morphologic stages. It may show the importance of glomerular damage intensity in prognosis and survival of patients.

scholarly journals Late Relapses of Membranous Nephropathy: A Case Series

Kidney360 ◽  
2021 ◽  
pp. 10.34067/KID.0007712020
Author(s):  
Yonatan Peleg ◽  
Andrew S. Bomback ◽  
Pietro A. Canetta ◽  
Jai Radhakrishnan ◽  
Gerald B. Appel ◽  
...  
Keyword(s):  
Nephrotic Syndrome ◽  
Membranous Nephropathy ◽  
Clinical Symptoms ◽  
Case Series ◽  
Renal Outcome ◽  
Late Relapse ◽  
Sustained Remission ◽  
Glomerular Diseases ◽  
Disease Remission

Background: Relapse of the nephrotic syndrome is common among primary membranous nephropathy (MN) patients. Relapses of MN typically occur within a few years of achieving disease remission. There is limited description to date regarding MN patients who have late relapse of MN, i.e. after more than five years of sustained disease remission. The objective of this case series was to report the clinical course of MN patients with late relapse. Methods: We analyzed the patient database of the Center for Glomerular Diseases at Columbia University to identify patients seen at our center who had relapse of biopsy-proven MN at least five years after achieving sustained disease remission. Results: We identified 16 patients with late MN relapse. The median time in sustained remission prior to relapse was 10.2 (range 7-29.0) years. Ten (62.5%) patients were diagnosed with late relapse based on laboratory monitoring alone without clinical symptoms of the nephrotic syndrome. Fourteen (87.5%) patients received immunosuppression during their initial presentation and late relapse. Patients had favorable long term renal outcomes over a median 21 (range 12-56) year follow-up period with 14 (87.5%) patients in remission at study conclusion and median decline in eGFR per year -0.63 (range -6.3 - 17.5) ml/min/1.73m2/year. Conclusions: This case series highlights a previously under-appreciated and likely rare outcome of MN, namely late relapse. Late relapse patients, having a longer time in sustained remission, may have a more favorable long-term renal outcome.

Protein deep profile and model predictions for identifying the causal genes of male infertility based on deep learning

2021 ◽  
Vol 75 ◽  
pp. 70-89
Author(s):  
Fang Xu ◽  
Ganggang Guo ◽  
Feida Zhu ◽  
Xiaojun Tan ◽  
Liqing Fan
Keyword(s):  
Deep Learning ◽  
Male Infertility ◽  
Model Predictions ◽  
Causal Genes ◽  
Deep Profile

scholarly journals Uncertainty-Aware Deep Learning-Based Cardiac Arrhythmias Classification Model of Electrocardiogram Signals

Computers ◽  
2021 ◽  
Vol 10 (6) ◽  
pp. 82
Author(s):  
Ahmad O. Aseeri
Keyword(s):  
Deep Learning ◽  
Cardiac Arrhythmias ◽  
Large Scale ◽  
Clinical Decision Making ◽  
Probabilistic Approach ◽  
Classification Model ◽  
Gating Mechanism ◽  
Uncertainty Estimates ◽  
Wide Range

Deep Learning-based methods have emerged to be one of the most effective and practical solutions in a wide range of medical problems, including the diagnosis of cardiac arrhythmias. A critical step to a precocious diagnosis in many heart dysfunctions diseases starts with the accurate detection and classification of cardiac arrhythmias, which can be achieved via electrocardiograms (ECGs). Motivated by the desire to enhance conventional clinical methods in diagnosing cardiac arrhythmias, we introduce an uncertainty-aware deep learning-based predictive model design for accurate large-scale classification of cardiac arrhythmias successfully trained and evaluated using three benchmark medical datasets. In addition, considering that the quantification of uncertainty estimates is vital for clinical decision-making, our method incorporates a probabilistic approach to capture the model’s uncertainty using a Bayesian-based approximation method without introducing additional parameters or significant changes to the network’s architecture. Although many arrhythmias classification solutions with various ECG feature engineering techniques have been reported in the literature, the introduced AI-based probabilistic-enabled method in this paper outperforms the results of existing methods in outstanding multiclass classification results that manifest F1 scores of 98.62% and 96.73% with (MIT-BIH) dataset of 20 annotations, and 99.23% and 96.94% with (INCART) dataset of eight annotations, and 97.25% and 96.73% with (BIDMC) dataset of six annotations, for the deep ensemble and probabilistic mode, respectively. We demonstrate our method’s high-performing and statistical reliability results in numerical experiments on the language modeling using the gating mechanism of Recurrent Neural Networks.

Sign in / Sign up

Export Citation Format

Share Document