Standard Echocardiographic View Recognition in Diagnosis of Congenital Heart Defects in Children Using Deep Learning Based on Knowledge Distillation

Wu, Lihui; Dong, Bin; Liu, Xiaoqing; Hong, Wenjing; Chen, Lijun; Kunlun, Gao; Sheng, Qiuyang; Yu, Yizhou; Zhao, Liebin; Zhang, Yuqi

doi:10.3389/fped.2021.770182

Cited by 13 publications

(7 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Dice Similarity Coefficient (DSC) is used as the evaluation process metric during cardiac anatomical segmentation. These are their definitions, EQU (7), EQU (8), EQU (9), EQU (10), EQU (11), and EQU ( 12…”

Section: A Experimental Resultsmentioning

confidence: 99%

“…It is now the most frequently used non-invasive investigation procedure for CRDs. The heterogeneity of ECG diagnosis is caused by various causes, including the heart's natural pulse variability, speckled noise and artefacts, and variations in typical ECG views within and between classes [8]. Accurate diagnosis of TTECG is complex and requires minimum time consumption, and it severely depends on the accurate interpretation of each ECG view by professional heart specialists.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Echocardiographic Image Segmentation for Diagnosing Fetal Cardiac Rhabdomyoma During Pregnancy Using Deep Learning

et al. 2022

View full text Add to dashboard Cite

Automated interpretation of cardiac images has the potential to change clinical practice in many ways. For example, it could make it possible for non-experts in primary care and rural settings to test the heart's function over time. In this paper, we tested the research hypothesis that recent developments in computer vision would make it possible to create a fully automated, scalable analysis for echocardiogram interpretation, covering all of the steps from view identification and Medical Image Segmentation (MIS) to structure and function quantification and Fetal Cardiac RhabDomyomas (FCRD) detection. Even though they are rare, FCRDs are the most frequent cause of Fetal Cardiac Tumor (FCT). When it comes to diagnosing and monitoring fetuses with an injured circulatory system, imaging (particularly echocardiography (ECG)) has proven helpful in the field of fetal cardiology. Because of the severe lack of qualified and experienced sonographers, it is very challenging t o diagnose Cardiac RhabDomyomas (CRD). Prior to delivery, accurate segmentation of the FC to identify structural cardiac defects is critical for minimizing the illness among newborns. To automate the process of segmenting the cardiac chamber for the CRD, we propose a novel Attention-Residual Network-based V-Net architecture (ARVNet). In this study, examinations were performed on Fetal Rhabdomyomas noted in the Right Ventricle (FRRV), Fetal Rhabdomyomas noted in the Left Ventricle (FRLV), Fetal Rhabdomyomas noted in the Right Atrium (FRRA), Fetal Rhabdomyomas noted in the Left Atrium (FRLA), Fetal Rhabdomyomas noted in the Tricuspid Valve (FRTV). Images without Rhabdomyoma mean "Normal Condition (NC)" at Selvam Hospital in Melapalayam, Tirunelveli, Tamil Nadu, India. Even with a relatively small number of datasets, the proposed technique possesses high CRD detection performance, as evidenced by the results. The results showed that the proposed model did a good job segmenting all the views, with a specificity of 99.7% and a Dice coefficient similarit y of 99.8%. It also did well at finding CRDs, with an average mean accuracy of around 99.85%.

show abstract

Section: A Experimental Resultsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Echocardiographic Image Segmentation for Diagnosing Fetal Cardiac Rhabdomyoma During Pregnancy Using Deep Learning

et al. 2022

View full text Add to dashboard Cite

show abstract

“…Standard echocardiographic view recognition is a prerequisite for clinical diagnosis of heart disease. Our standard view identification model is based on our previous work ( 31 ), where we recognized 24 classes of standard views with high accuracy. Since the purpose of this study is to detect ASD, we only focus on four target views (i.e., subAS, A4C, LPS4C, and sax-basal) and refer to all other views as “other.” As shown in Figure 2 , a knowledge distillation ( 32 ) method was applied to train the standard view identification model, in which we applied ResNet-34 ( 33 ) as the student model and ResNeSt-200 ( 34 ) as the teacher model.…”

Section: Methodsmentioning

confidence: 99%

Automatic Detection of Secundum Atrial Septal Defect in Children Based on Color Doppler Echocardiographic Images Using Convolutional Neural Networks

Hong

Sheng

Dong

et al. 2022

Front. Cardiovasc. Med.

Self Cite

View full text Add to dashboard Cite

Secundum atrial septal defect (ASD) is one of the most common congenital heart diseases (CHDs). This study aims to evaluate the feasibility and accuracy of automatic detection of ASD in children based on color Doppler echocardiographic images using convolutional neural networks. In this study, we propose a fully automatic detection system for ASD, which includes three stages. The first stage is used to identify four target echocardiographic views (that is, the subcostal view focusing on the atrium septum, the apical four-chamber view, the low parasternal four-chamber view, and the parasternal short-axis view). These four echocardiographic views are most useful for the diagnosis of ASD clinically. The second stage aims to segment the target cardiac structure and detect candidates for ASD. The third stage is to infer the final detection by utilizing the segmentation and detection results of the second stage. The proposed ASD detection system was developed and validated using a training set of 4,031 cases containing 370,057 echocardiographic images and an independent test set of 229 cases containing 203,619 images, of which 105 cases with ASD and 124 cases with intact atrial septum. Experimental results showed that the proposed ASD detection system achieved accuracy, recall, precision, specificity, and F1 score of 0.8833, 0.8545, 0.8577, 0.9136, and 0.8546, respectively on the image-level averages of the four most clinically useful echocardiographic views. The proposed system can automatically and accurately identify ASD, laying a good foundation for the subsequent artificial intelligence diagnosis of CHDs.

show abstract

“…Kusunose 23 proposed four crucial steps for the application of AI in echocardiography: image quality preservation during acquisition, view classification, measurement and quantification, and anomaly detection. Wu et al 24 introduced a deep learning‐based neural network approach for the automated and effective recognition of standard echocardiographic views, achieving F1 scores exceeding 0.90 for the majority of views. Zhang et al 25 developed, trained, and evaluated Convolutional Neural Network (CNN) models for various tasks, including automatic recognition of 23 viewpoints and left ventricle segmentation for five common views.…”

Section: Introductionmentioning

confidence: 99%

Intelligent detection of left ventricular hypertrophy from pediatric echocardiography videos

Liu,

Han,

Liang

et al. 2024

Int J Imaging Syst Tech

View full text Add to dashboard Cite

Left ventricular hypertrophy (LVH) is a common clinical manifestation associated with cardiovascular adverse events. Relying solely on the subjective judgment, it is challenging to promptly and accurately diagnose mild LVH. Quantitatively measuring the interventricular septum and left ventricular posterior wall to diagnose mild LVH is a time‐consuming and labor‐intensive process that is prone to errors. To propose a novel method for rapid and automatic end‐to‐end diagnosis of mild LVH. We propose a novel end‐to‐end automated method for detecting mild LVH. This method achieves rapid end‐to‐end detection of mild LVH in echocardiographic videos without the need for quantitative measurements. Initially, representative frames are extracted from echocardiographic videos, and these frames are then automatically diagnosed by the proposed network to detect LVH. The network architecture primarily consists of three key components: a feature extractor, bidirectional LSTM, and attention module. The Vit‐b model achieved 88% video classification accuracy in the experiment where 32 frames are extracted, and the ViT‐l model achieves 92% video classification accuracy in the experiment where 16 frames are extracted. It is shown experimentally that extracting fewer video frames can diagnose LVH more accurately. The experiments illustrate the superior performance and competitiveness of this method compared to other approaches, potentially applicable to the clinic.

show abstract

Standard Echocardiographic View Recognition in Diagnosis of Congenital Heart Defects in Children Using Deep Learning Based on Knowledge Distillation

Cited by 13 publications

References 24 publications

Echocardiographic Image Segmentation for Diagnosing Fetal Cardiac Rhabdomyoma During Pregnancy Using Deep Learning

Echocardiographic Image Segmentation for Diagnosing Fetal Cardiac Rhabdomyoma During Pregnancy Using Deep Learning

Automatic Detection of Secundum Atrial Septal Defect in Children Based on Color Doppler Echocardiographic Images Using Convolutional Neural Networks

Intelligent detection of left ventricular hypertrophy from pediatric echocardiography videos

Contact Info

Product

Resources

About