Machine learning phenotyping of scarred myocardium from cine in hypertrophic cardiomyopathy

2021 ◽  
Author(s):  
Jennifer Mancio ◽  
Farhad Pashakhanloo ◽  
Hossam El-Rewaidy ◽  
Jihye Jang ◽  
Gargi Joshi ◽  
...  
Keyword(s):  
Machine Learning ◽  
Hypertrophic Cardiomyopathy ◽  
Wall Thickness ◽  
Model Development ◽  
Area Under The Curve ◽  
False Negatives ◽  
Regional Wall ◽  
Texture Size ◽  
Myocardial Signal Intensity ◽  
Rule Out

Abstract Aims Cardiovascular magnetic resonance (CMR) with late-gadolinium enhancement (LGE) is increasingly being used in hypertrophic cardiomyopathy (HCM) for diagnosis, risk stratification, and monitoring. However, recent data demonstrating brain gadolinium deposits have raised safety concerns. We developed and validated a machine-learning (ML) method that incorporates features extracted from cine to identify HCM patients without fibrosis in whom gadolinium can be avoided. Methods and results An XGBoost ML model was developed using regional wall thickness and thickening, and radiomic features of myocardial signal intensity, texture, size, and shape from cine. A CMR dataset containing 1099 HCM patients collected using 1.5T CMR scanners from different vendors and centres was used for model development (n=882) and validation (n=217). Among the 2613 radiomic features, we identified 7 features that provided best discrimination between +LGE and −LGE using 10-fold stratified cross-validation in the development cohort. Subsequently, an XGBoost model was developed using these radiomic features, regional wall thickness and thickening. In the independent validation cohort, the ML model yielded an area under the curve of 0.83 (95% CI: 0.77–0.89), sensitivity of 91%, specificity of 62%, F1-score of 77%, true negatives rate (TNR) of 34%, and negative predictive value (NPV) of 89%. Optimization for sensitivity provided sensitivity of 96%, F2-score of 83%, TNR of 19% and NPV of 91%; false negatives halved from 4% to 2%. Conclusion An ML model incorporating novel radiomic markers of myocardium from cine can rule-out myocardial fibrosis in one-third of HCM patients referred for CMR reducing unnecessary gadolinium administration.

Layer-specific strain and the degree of left ventricular thickness in patients with hypertrophic cardiomyopathy

2020 ◽  
Vol 41 (Supplement_2) ◽  
Author(s):  
T Tsugu ◽  
Y Nagatomo ◽  
R Dulgheru ◽  
S Marchetta ◽  
A Postolache ◽  
...  
Keyword(s):  
Hypertrophic Cardiomyopathy ◽  
Wall Thickness ◽  
Myocardial Dysfunction ◽  
Interventricular Septum ◽  
Area Under The Curve ◽  
Cellular Level ◽  
Left Ventricular ◽  
Myocardial Tissue ◽  
Characteristic Analysis ◽  
Specific Strain

Abstract Background Left ventricular (LV) wall thickness is an important parameter for the diagnosis of hypertrophic cardiomyopathy (HCM) and is also associated with long-term clinical outcome in HCM patients. However, conventional tools have failed to analyze the mechanisms of structural and functional abnormalities that occur at the cellular level in hypertrophied myocardial tissue. Recently, technological progression of 2D-speckle tracking echocardiography (2D-STE) has enabled the estimation of layer-specific strain (LSS), such as epicardial, mid-myocardial, and endocardial longitudinal strain, respectively. LSS may have the potential to elucidate the detailed mechanisms of myocardial dysfunction. Purpose The aim of this study was (i) to clarify the detailed mechanisms of structural and functional abnormalities of myocardial tissue in HCM using LSS (ii) to investigate the diagnostic accuracy of LSS for HCM. Methods Forty-one patients with HCM and preserved LV ejection fraction (LVEF) (66% male, 52±18 years, LVEF 62.9±3.7%) and 41 controls matched for age and sex (66% male, 52±20 years, LVEF 63.5±8.2%) underwent 2D-STE (GE-Healthcare, Vivid-E9). Quantitative strain values of epicardial, mid-myocardial, and endocardial layers were measured. Results LV wall thickness including interventricular septum thickness (HCM vs. Controls; 18.9±5.0 vs. 9.1±1.8, p<0.001), posterior wall thickness (11.5±2.5 vs. 8.8±1.9, p<0.001), and maximum wall thickness (20.1±4.3 vs. 9.4±0.4, p<0.001) were significantly lower in HCM than in Controls. Absolute values of LSS for all layers were lower in HCM than in Controls (HCM vs. Controls; epicardial; −13.1±3.3 vs. −19.5±1.6, p<0.001; mid-myocardial; −15.8±3.3 vs. −21.4±1.7, p<0.001; endocardial; −18.9±3.9 vs. −23.6±1.9, p<0.001). End/Epi ratio was higher in HCM than in Controls (HCM vs. Controls; 1.5±0.2 vs. 1.2±0.0, p<0.001). Next, we investigated the echocardiographic parameters that correlated with LV maximal wall thickness (MWT). End/Epi ratio was an independent predictor of LV MWT (β=0.96, p<0.001). Receiver operating characteristic analysis revealed that a higher End/Epi ratio (≥1.3) was the strongest predictor of diagnostic criteria for HCM (LV wall thickness ≥15 mm) (area under the curve 0.99, p<0.001, sensitivity 98%, specificity 97%). Conclusions In HCM patients with preserved LVEF, (i) LSS was lower and End/Epi ratio was higher than in controls. (ii) End/Epi ratio (≥1.3) was the strongest predictor of abnormal wall thickness of HCM. The mechanism of higher End/Epi ratio in HCM might be attributable to the more common myofibrillar disarray in mid- and epicardial layers. Variations of LSS represented by End/Epi ratio might have the potential to accurately detect HCM and to elucidate the pathophysiology of impaired LV wall motion at cellular level in HCM. Funding Acknowledgement Type of funding source: None

scholarly journals Novel Morphological Features on CMR for the Prediction of Pathogenic Sarcomere Gene Variants in Subjects Without Hypertrophic Cardiomyopathy

2021 ◽  
Vol 8 ◽  
Author(s):  
Nikki van der Velde ◽  
Roy Huurman ◽  
H. Carlijne Hassing ◽  
Ricardo P. J. Budde ◽  
Marjon A. van Slegtenhorst ◽  
...  
Keyword(s):  
Genetic Testing ◽  
Hypertrophic Cardiomyopathy ◽  
Wall Thickness ◽  
Area Under The Curve ◽  
Posterior Wall ◽  
Left Ventricular ◽  
Morphological Features ◽  
Gene Variants ◽  
Healthy Controls ◽  
Score System

Background: Carriers of pathogenic DNA variants (G+) causing hypertrophic cardiomyopathy (HCM) can be identified by genetic testing. Several abnormalities have been brought forth as pre-clinical expressions of HCM, some of which can be identified by cardiovascular magnetic resonance (CMR). In this study, we assessed morphological differences between G+/left ventricular hypertrophy-negative (LVH-) subjects and healthy controls and examined whether CMR-derived variables are useful for the prediction of sarcomere gene variants.Methods: We studied 57 G+ subjects with a maximal wall thickness (MWT) < 13 mm, and compared them to 40 healthy controls matched for age and sex on a group level. Subjects underwent CMR including morphological, volumetric and function assessment. Logistic regression analysis was performed for the determination of predictive CMR characteristics, by which a scoring system for G+ status was constructed.Results: G+/LVH- subjects were subject to alterations in the myocardial architecture, resulting in a thinner posterior wall thickness (PWT), higher interventricular septal wall/PWT ratio and MWT/PWT ratio. Prominent hook-shaped configurations of the anterobasal segment were only observed in this group. A model consisting of the anterobasal hook, multiple myocardial crypts, right ventricular/left ventricular ratio, MWT/PWT ratio, and MWT/left ventricular mass ratio predicted G+ status with an area under the curve of 0.92 [0.87–0.97]. A score of ≥3 was present only in G+ subjects, identifying 56% of the G+/LVH- population.Conclusion: A score system incorporating CMR-derived variables correctly identified 56% of G+ subjects. Our results provide further insights into the wide phenotypic spectrum of G+/LVH- subjects and demonstrate the utility of several novel morphological features. If genetic testing for some reason cannot be performed, CMR and our purposed score system can be used to detect possible G+ carriers and to aid planning of the control intervals.

Detection of Falsely Elevated Point-of-Care Potassium Results Due to Hemolysis Using Predictive Analytics

2020 ◽  
Vol 154 (2) ◽  
pp. 242-247
Author(s):  
Robert C Benirschke ◽  
Thomas J Gniadek
Keyword(s):  
Machine Learning ◽  
Predictive Value ◽  
Predictive Analytics ◽  
Point Of Care ◽  
Area Under The Curve ◽  
Plasma Potassium ◽  
Data Set ◽  
Laboratory Errors ◽  
Rule Out

Abstract Objectives Preanalytical factors, such as hemolysis, affect many components of a test panel. Machine learning can be used to recognize these patterns, alerting clinicians and laboratories to potentially erroneous results. In particular, machine learning might identify which cases of elevated potassium from a point-of-care (POC) basic metabolic panel are likely erroneous. Methods Plasma potassium concentrations were compared between POC and core laboratory basic metabolic panels to identify falsely elevated POC results. A logistic regression model was created using these labels and the other analytes on the POC panel. Results This model has high predictive power in classifying POC potassium as falsely elevated or not (area under the curve of 0.995 when applied to the test data set). A rule-in and rule-out approach further improves the model’s applicability with a positive predictive value of around 90% and a negative predictive value near 100%. Conclusions Machine learning has the potential to detect laboratory errors based on the recognition of patterns in commonly requested multianalyte panels. This could be used to alert providers at the POC that a result is suspicious or used to monitor the quality of POC results.

scholarly journals Predicting Kidney Graft Survival using Machine Learning Methods: Prediction Model Development and Feature Significance Analysis Study (Preprint)

2021 ◽  
Author(s):  
Syed Asil Ali Naqvi ◽  
Karthik Tennankore ◽  
Amanda Vinson ◽  
Patrice C. Roy ◽  
Syed Sibte Raza Abidi
Keyword(s):  
Machine Learning ◽  
Kidney Transplantation ◽  
Graft Survival ◽  
Graft Failure ◽  
Prediction Models ◽  
Model Development ◽  
Area Under The Curve ◽  
Kidney Graft ◽  
Applied Machine Learning ◽  
Importance Analysis

BACKGROUND Kidney transplantation is the optimal treatment for patients with end-stage kidney disease. Short and long term kidney graft survival is influenced by a number of donor-and recipient factors. Predicting the success of kidney transplantation is important to optimize kidney allocation. OBJECTIVE To predict the risk of kidney graft failure across three temporal cohorts – within 1 year, 5 years, and more than 5 years after transplantation, based on donor and recipient characteristics. We analyzed a large dataset comprising over 50000 kidney transplants covering an approximate 20-year period. METHODS We applied machine learning based classification algorithms to develop prediction models to predict the risk of graft failure for the three different temporal cohorts. Deep learning based autoencoders were applied for data dimensionality reduction, which improved the prediction performance. Feature influence towards graft survival for each cohort was studied by investigating a new non-overlapping patient stratification approach. RESULTS Our models predicted graft survival with area under the curve (AUC) scores of 82% within 1 year, 69% within 5 years, and 81% within 17 years. The feature importance analysis elucidated the varying influence of clinical features towards graft survival across the three different temporal cohorts. CONCLUSIONS We developed machine learning models to predict kidney graft survival for three temporal cohorts and analyzed the changing relevance of features over time.

scholarly journals Blastocyst Prediction of Day-3 Cleavage-Stage Embryos Using Machine Learning

2021 ◽  
pp. 1-6
Author(s):  
Dung P. Nguyen ◽  
Quan T. Pham ◽  
Thanh L. Tran ◽  
Lan N. Vuong ◽  
Tuong M. Ho
Keyword(s):  
Machine Learning ◽  
Model Development ◽  
Area Under The Curve ◽  
Training Dataset ◽  
Validation Dataset ◽  
Blastocyst Formation ◽  
Final Model ◽  
Vitro Fertilization ◽  
Day 3 Embryo

Background:Embryo selection plays an important role in the success of in vitro fertilization (IVF). However, morphological embryo assessment has a number of limitations, including the time required, lack of accuracy, and inconsistency. This study determined whether a machine learning-based model could predict blastocyst formation using day-3 embryo images. Methods:Day-3 embryo images from IVF/intracytoplasmic sperm injection (ICSI) cycles performed at My Duc Phu Nhuan Hospital between August 2018 and June 2019 were retrospectively analyzed to inform model development. Day-3 embryo images derived from two-pronuclear (2PN) zygotes with known blastocyst formation data were extracted from the CCM-iBIS time-lapse incubator (Astec, Japan) at 67 hours post ICSI, and labeled as blastocyst/non-blastocyst based on results at 116 hours post ICSI. Images were used as the input dataset to train (85%) and validate (15%) the convolutional neural network (CNN) model, then model accuracy was determined using the training and validation dataset. The performance of 13 experienced embryologists for predicting blastocyst formation based on 100 day-3 embryo images was also evaluated. Results:A total of 1,135 images were allocated into training ([Formula: see text] = 967) and validation ([Formula: see text] = 168) sets, with an even distribution for blastocyst formation outcome. The accuracy of the final model for blastocyst formation was 97.72% in the training dataset and 76.19% in the validation dataset. The final model predicted blastocyst formation from day-3 embryo images in the validation dataset with an area under the curve of 0.75 (95% confidence interval [CI] 0.69–0.81). Embryologists predicted blastocyst formation with the accuracy of 70.07% (95% CI 68.12%–72.03%), sensitivity of 87.04% (95% CI 82.56%–91.52%), and specificity of 30.93% (95% CI 29.35%–32.51%). Conclusions:The CNN-based machine learning model using day-3 embryo images predicted blastocyst formation more accurately than experienced embryologists. The CNN-based model is a potential tool to predict additional IVF outcomes.

scholarly journals Integration of transcriptomic data identifies key hallmark genes in hypertrophic cardiomyopathy

2021 ◽  
Vol 21 (1) ◽  
Author(s):  
Jing Xu ◽  
Xiangdong Liu ◽  
Qiming Dai
Keyword(s):  
Machine Learning ◽  
Hypertrophic Cardiomyopathy ◽  
Heart Diseases ◽  
Expression Patterns ◽  
Support Vector ◽  
Rna Seq ◽  
Ppi Network ◽  
Learning Methods ◽  
Transcriptomic Data ◽  
Machine Learning Methods

Abstract Background Hypertrophic cardiomyopathy (HCM) represents one of the most common inherited heart diseases. To identify key molecules involved in the development of HCM, gene expression patterns of the heart tissue samples in HCM patients from multiple microarray and RNA-seq platforms were investigated. Methods The significant genes were obtained through the intersection of two gene sets, corresponding to the identified differentially expressed genes (DEGs) within the microarray data and within the RNA-Seq data. Those genes were further ranked using minimum-Redundancy Maximum-Relevance feature selection algorithm. Moreover, the genes were assessed by three different machine learning methods for classification, including support vector machines, random forest and k-Nearest Neighbor. Results Outstanding results were achieved by taking exclusively the top eight genes of the ranking into consideration. Since the eight genes were identified as candidate HCM hallmark genes, the interactions between them and known HCM disease genes were explored through the protein–protein interaction (PPI) network. Most candidate HCM hallmark genes were found to have direct or indirect interactions with known HCM diseases genes in the PPI network, particularly the hub genes JAK2 and GADD45A. Conclusions This study highlights the transcriptomic data integration, in combination with machine learning methods, in providing insight into the key hallmark genes in the genetic etiology of HCM.

scholarly journals Early Prediction of Seven-Day Mortality in Intensive Care Unit Using a Machine Learning Model: Results from the SPIN-UTI Project

2021 ◽  
Vol 10 (5) ◽  
pp. 992
Author(s):  
Martina Barchitta ◽  
Andrea Maugeri ◽  
Giuliana Favara ◽  
Paolo Marco Riela ◽  
Giovanni Gallo ◽  
...  
Keyword(s):  
Machine Learning ◽  
Intensive Care ◽  
Intensive Care Units ◽  
Learning Algorithm ◽  
Area Under The Curve ◽  
Support Vector ◽  
Icu Admission ◽  
Risk Of Death ◽  
Saps Ii ◽  
Svm Algorithm

Patients in intensive care units (ICUs) were at higher risk of worsen prognosis and mortality. Here, we aimed to evaluate the ability of the Simplified Acute Physiology Score (SAPS II) to predict the risk of 7-day mortality, and to test a machine learning algorithm which combines the SAPS II with additional patients’ characteristics at ICU admission. We used data from the “Italian Nosocomial Infections Surveillance in Intensive Care Units” network. Support Vector Machines (SVM) algorithm was used to classify 3782 patients according to sex, patient’s origin, type of ICU admission, non-surgical treatment for acute coronary disease, surgical intervention, SAPS II, presence of invasive devices, trauma, impaired immunity, antibiotic therapy and onset of HAI. The accuracy of SAPS II for predicting patients who died from those who did not was 69.3%, with an Area Under the Curve (AUC) of 0.678. Using the SVM algorithm, instead, we achieved an accuracy of 83.5% and AUC of 0.896. Notably, SAPS II was the variable that weighted more on the model and its removal resulted in an AUC of 0.653 and an accuracy of 68.4%. Overall, these findings suggest the present SVM model as a useful tool to early predict patients at higher risk of death at ICU admission.

Assessment of Machine Learning-Based Audiovisual Quality Predictors

2021 ◽  
Vol 17 (2) ◽  
pp. 1-22
Author(s):  
Mythili K. ◽  
Manish Narwaria
Keyword(s):  
Machine Learning ◽  
Signal Processing ◽  
Quality Assessment ◽  
Computational Models ◽  
Model Development ◽  
Multimedia Communication ◽  
Current Approach ◽  
Learning Ability ◽  
Data Uncertainty ◽  
Attractive Alternative

Quality assessment of audiovisual (AV) signals is important from the perspective of system design, optimization, and management of a modern multimedia communication system. However, automatic prediction of AV quality via the use of computational models remains challenging. In this context, machine learning (ML) appears to be an attractive alternative to the traditional approaches. This is especially when such assessment needs to be made in no-reference (i.e., the original signal is unavailable) fashion. While development of ML-based quality predictors is desirable, we argue that proper assessment and validation of such predictors is also crucial before they can be deployed in practice. To this end, we raise some fundamental questions about the current approach of ML-based model development for AV quality assessment and signal processing for multimedia communication in general. We also identify specific limitations associated with the current validation strategy which have implications on analysis and comparison of ML-based quality predictors. These include a lack of consideration of: (a) data uncertainty, (b) domain knowledge, (c) explicit learning ability of the trained model, and (d) interpretability of the resultant model. Therefore, the primary goal of this article is to shed some light into mentioned factors. Our analysis and proposed recommendations are of particular importance in the light of significant interests in ML methods for multimedia signal processing (specifically in cases where human-labeled data is used), and a lack of discussion of mentioned issues in existing literature.

Sign in / Sign up

Export Citation Format

Share Document