Using deep learning method to identify left ventricular hypertrophy on echocardiography

Enhanced classification of left ventricular hypertrophy in cardiac patients using extended Siamese CNN

Objective.Left ventricular hypertrophy (LVH) is the thickening of the left ventricle wall of the heart. The objective of this study is to develop a novel approach for the accurate assessment of LVH) severity, addressing the limitations of traditional manual grading systems.Approach.We propose the Multi-purpose Siamese Weighted Euclidean Distance Model (MSWED), which utilizes convolutional Siamese neural networks and zero-shot/few-shot learning techniques. Unlike traditional methods, our model introduces a cutoff distance-based approach for zero-shot learning, enhancing accuracy. We also incorporate a weighted Euclidean distance targeting informative regions within echocardiograms.Main results.We collected comprehensive datasets labeled by experienced echocardiographers, including Normal heart and various levels of LVH severity. Our model outperforms existing techniques, demonstrating significant precision enhancement, with improvements of up to 13% for zero-shot and few-shot learning approaches.Significance.Accurate assessment of LVH severity is crucial for clinical prognosis and treatment decisions. Our proposed MSWED model offers a more reliable and efficient solution compared to traditional grading systems, reducing subjectivity and errors while providing enhanced precision in severity classification.

Advance of echocardiography in cardiac amyloidosis

Cardiac amyloidosis (CA) occurs when the insoluble fibrils formed by misfolded precursor proteins deposit in cardiac tissues. The early clinical manifestations of CA are not evident, but it is easy to progress to refractory heart failure with an inferior prognosis. Echocardiography is the most commonly adopted non-invasive modality of imaging to visualize cardiac structures and functions, and the preferred modality in the evaluation of patients with cardiac symptoms and suspected CA, which plays a vital role in the diagnosis, prognosis, and long-term management of CA. The present review summarizes the echocardiographic manifestations of CA, new echocardiographic techniques, and the application of multi-parametric echocardiographic models in CA diagnosis.

A data-efficient zero-shot and few-shot Siamese approach for automated diagnosis of left ventricular hypertrophy

Left ventricular hypertrophy (LVH) is a life-threatening condition in which the muscle of the left ventricle thickens and enlarges. Echocardiography is a test performed by cardiologists and echocardiographers to diagnose this condition. The manual interpretation of echocardiography tests is time-consuming and prone to errors. To address this issue, we have developed an automated LVH diagnosis technique using deep learning. However, the availability of medical data is a significant challenge due to varying industry standards, privacy laws, and legal constraints. To overcome this challenge, we have proposed a data-efficient technique for automated LVH classification using echocardiography. Firstly, we collected our own dataset of normal and LVH echocardiograms from 70 patients in collaboration with a clinical facility. Secondly, we introduced novel zero-shot and few-shot algorithms based on a modified Siamese network to classify LVH and normal images. Unlike traditional zero-shot learning approaches, our proposed method does not require text vectors, and classification is based on a cutoff distance. Our model demonstrates superior performance compared to state-of-the-art techniques, achieving up to 8% precision improvement for zero-shot learning and up to 11% precision improvement for few-shot learning approaches. Additionally, we assessed the inter-observer and intra-observer reliability scores of our proposed approach against two expert echocardiographers. The results revealed that our approach achieved better inter-observer and intra-observer reliability scores compared to the experts.

Challenges and solutions of echocardiography generalization for deep learning: a study in patients with constrictive pericarditis

Purpose

The inherent characteristics of transthoracic echocardiography (TTE) images such as low signal-to-noise ratio and acquisition variations can limit the direct use of TTE images in the development and generalization of deep learning models. As such, we propose an innovative automated framework to address the common challenges in the process of echocardiography deep learning model generalization on the challenging task of constrictive pericarditis (CP) and cardiac amyloidosis (CA) differentiation.

Approach

Patients with a confirmed diagnosis of CP or CA and normal cases from Mayo Clinic Rochester and Arizona were identified to extract baseline demographics and the apical 4 chamber view from TTE studies. We proposed an innovative preprocessing and image generalization framework to process the images for training the ResNet50, ResNeXt101, and EfficientNetB2 models. Ablation studies were conducted to justify the effect of each proposed processing step in the final classification performance.

Results

The models were initially trained and validated on 720 unique TTE studies from Mayo Rochester and further validated on 225 studies from Mayo Arizona. With our proposed generalization framework, EfficientNetB2 generalized the best with an average area under the curve (AUC) of 0.96 (± 0.01 ) and 0.83 (± 0.03 ) on the Rochester and Arizona test sets, respectively.

Conclusions

Leveraging the proposed generalization techniques, we successfully developed an echocardiography-based deep learning model that can accurately differentiate CP from CA and normal cases and applied the model to images from two sites. The proposed framework can be further extended for the development of echocardiography-based deep learning models.

Multi-channel deep learning model-based myocardial spatial-temporal morphology feature on cardiac MRI cine images diagnoses the cause of LVH

BACKGROUND

To develop a fully automatic framework for the diagnosis of cause for left ventricular hypertrophy (LVH) via cardiac cine images.

METHODS

A total of 302 LVH patients with cine MRI images were recruited as the primary cohort. Another 53 LVH patients prospectively collected or from multi-centers were used as the external test dataset. Different models based on the cardiac regions (Model 1), segmented ventricle (Model 2) and ventricle mask (Model 3) were constructed. The diagnostic performance was accessed by the confusion matrix with respect to overall accuracy. The capability of the predictive models for binary classification of cardiac amyloidosis (CA), hypertrophic cardiomyopathy (HCM) or hypertensive heart disease (HHD) were also evaluated. Additionally, the diagnostic performance of best Model was compared with that of 7 radiologists/cardiologists.

RESULTS

Model 3 showed the best performance with an overall classification accuracy up to 77.4% in the external test datasets. On the subtasks for identifying CA, HCM or HHD only, Model 3 also achieved the best performance with AUCs yielding 0.895-0.980, 0.879-0.984 and 0.848-0.983 in the validation, internal test and external test datasets, respectively. The deep learning model showed non-inferior diagnostic capability to the cardiovascular imaging expert and outperformed other radiologists/cardiologists.

CONCLUSION

The combined model based on the mask of left ventricular segmented from multi-sequences cine MR images shows favorable and robust performance in diagnosing the cause of left ventricular hypertrophy, which could be served as a noninvasive tool and help clinical decision.

Developing an Echocardiography-Based, Automatic Deep Learning Framework for the Differentiation of Increased Left Ventricular Wall Thickness Etiologies

AIMS

Increased left ventricular (LV) wall thickness is frequently encountered in transthoracic echocardiography (TTE). While accurate and early diagnosis is clinically important, given the differences in available therapeutic options and prognosis, an extensive workup is often required to establish the diagnosis. We propose the first echo-based, automated deep learning model with a fusion architecture to facilitate the evaluation and diagnosis of increased left ventricular (LV) wall thickness.

METHODS AND RESULTS

Patients with an established diagnosis of increased LV wall thickness (hypertrophic cardiomyopathy (HCM), cardiac amyloidosis (CA), and hypertensive heart disease (HTN)/others) between 1/2015 and 11/2019 at Mayo Clinic Arizona were identified. The cohort was divided into 80%/10%/10% for training, validation, and testing sets, respectively. Six baseline TTE views were used to optimize a pre-trained InceptionResnetV2 model. Each model output was used to train a meta-learner under a fusion architecture. Model performance was assessed by multiclass area under the receiver operating characteristic curve (AUROC). A total of 586 patients were used for the final analysis (194 HCM, 201 CA, and 191 HTN/others). The mean age was 55.0 years, and 57.8% were male. Among the individual view-dependent models, the apical 4-chamber model had the best performance (AUROC: HCM: 0.94, CA: 0.73, and HTN/other: 0.87). The final fusion model outperformed all the view-dependent models (AUROC: HCM: 0.93, CA: 0.90, and HTN/other: 0.92).

CONCLUSION

The echo-based InceptionResnetV2 fusion model can accurately classify the main etiologies of increased LV wall thickness and can facilitate the process of diagnosis and workup.

Differential diagnosis of common etiologies of left ventricular hypertrophy using a hybrid CNN-LSTM model

Differential diagnosis of left ventricular hypertrophy (LVH) is often obscure on echocardiography and requires numerous additional tests. We aimed to develop a deep learning algorithm to aid in the differentiation of common etiologies of LVH (i.e. hypertensive heart disease [HHD], hypertrophic cardiomyopathy [HCM], and light-chain cardiac amyloidosis [ALCA]) on echocardiographic images. Echocardiograms in 5 standard views (parasternal long-axis, parasternal short-axis, apical 4-chamber, apical 2-chamber, and apical 3-chamber) were obtained from 930 subjects: 112 with HHD, 191 with HCM, 81 with ALCA and 546 normal subjects. The study population was divided into training (n = 620), validation (n = 155), and test sets (n = 155). A convolutional neural network-long short-term memory (CNN-LSTM) algorithm was constructed to independently classify the 3 diagnoses on each view, and the final diagnosis was made by an aggregate network based on the simultaneously predicted probabilities of HCM, HCM, and ALCA. Diagnostic performance of the algorithm was evaluated by the area under the receiver operating characteristic curve (AUC), and accuracy was evaluated by the confusion matrix. The deep learning algorithm was trained and verified using the training and validation sets, respectively. In the test set, the average AUC across the five standard views was 0.962, 0.982 and 0.996 for HHD, HCM and CA, respectively. The overall diagnostic accuracy was significantly higher for the deep learning algorithm (92.3%) than for echocardiography specialists (80.0% and 80.6%). In the present study, we developed a deep learning algorithm for the differential diagnosis of 3 common LVH etiologies (HHD, HCM and ALCA) by applying a hybrid CNN-LSTM model and aggregate network to standard echocardiographic images. The high diagnostic performance of our deep learning algorithm suggests that the use of deep learning can improve the diagnostic process in patients with LVH.

Heterogeneous Distribution of Genetic Mutations in Myosin Binding Protein-C Paralogs

Myosin binding protein-C (MyBP-C) is a sarcomeric protein which regulates the force of contraction in striated muscles. Mutations in the MYBPC family of genes, including slow skeletal (MYBPC1), fast skeletal (MYBPC2) and cardiac (MYBPC3), can result in cardiac and skeletal myopathies. Nonetheless, their evolutionary pattern, pathogenicity and impact on MyBP-C protein structure remain to be elucidated. Therefore, the present study aimed to systematically assess the evolutionarily conserved and epigenetic patterns of MYBPC family mutations. Leveraging a machine learning (ML) approach, the Genome Aggregation Database (gnomAD) provided variants in MYBPC1, MYBPC2, and MYBPC3 genes. This was followed by an analysis with Ensembl’s variant effect predictor (VEP), resulting in the identification of 8,618, 3,871, and 3,071 variants in MYBPC1, MYBPC2, and MYBPC3, respectively. Missense variants comprised 61%–66% of total variants in which the third nucleotide positions in the codons were highly altered. Arginine was the most mutated amino acid, important because most disease-causing mutations in MyBP-C proteins are arginine in origin. Domains C5 and C6 of MyBP-C were found to be hotspots for most mutations in the MyBP-C family of proteins. A high percentage of truncated mutations in cMyBP-C cause cardiomyopathies. Arginine and glutamate were the top hits in fMyBP-C and cMyBP-C, respectively, and tryptophan and tyrosine were the most common among the three paralogs changing to premature stop codons and causing protein truncations at the carboxyl terminus. A heterogeneous epigenetic pattern was identified among the three MYBP-C paralogs. Overall, it was shown that databases using computational approaches can facilitate diagnosis and drug discovery to treat muscle disorders caused by MYBPC mutations.