Ultrasound Video Transformers for Cardiac Ejection Fraction Estimation

  title={Ultrasound Video Transformers for Cardiac Ejection Fraction Estimation},
  author={Hadrien Reynaud and Athanasios Vlontzos and Benjamin Hou and Arian Beqiri and Paul Leeson and Bernhard Kainz},
Cardiac ultrasound imaging is used to diagnose various heart diseases. Common analysis pipelines involve manual processing of the video frames by expert clinicians. This suffers from intraand interobserver variability. We propose a novel approach to ultrasound video analysis using a transformer architecture based on a Residual AutoEncoder Network and a BERT model adapted for token classification. This enables videos of any length to be processed. We apply our model to the task of End-Systolic… 

Hierarchical Vision Transformers for Cardiac Ejection Fraction Estimation

A deep learning approach, based on hierarchical vision Transformers, to estimate the ejection fraction from echocardiogram videos, which can estimate ejections fraction without the need for left ventrice segmentation first, make it more efficient than other methods.

EchoCoTr: Estimation of the Left Ventricular Ejection Fraction from Spatiotemporal Echocardiography

The algorithm proposed (EchoCoTr) utilizes the strength of vision transformers and CNNs to tackle the problem of estimating the left ventricular ejection fraction (LVEF) on ultrasound videos and outperforms state-of-the-art work to-date on the EchoNet-Dynamic dataset.

Light-weight spatio-temporal graphs for segmentation and ejection fraction prediction in cardiac ultrasound

A new automated method for predicting ejection fraction and segmenting the left ventricle by detecting anatomical keypoints using Graph Convolutional Networks, which shows accurate segmentation and improvements in robustness and inference runtime.

Cyclical Self-Supervision for Semi-Supervised Ejection Fraction Prediction from Echocardiogram Videos

A novel Cyclical Self-Supervision (CSS) method for learning video-based LV segmentation, motivated by the observation that the heartbeat is a cyclical process with temporal repetition is proposed.

EchoGNN: Explainable Ejection Fraction Estimation with Graph Neural Networks

EchoGNN, a model based on graph neural networks (GNNs) to estimate EF from echo videos achieves prediction performance that is on par with state of the art and provides explainability, which is crucial given the high inter-observer variability inherent in this task.

Artificial Intelligence-Enhanced Echocardiography for Systolic Function Assessment

This review attempts to thoroughly explain the latest progress of AI in assessing left ventricular systolic function and differential diagnosis of heart diseases by echocardiography and discusses the challenges and promises of this new field.

BabyNet: Residual Transformer Module for Birth Weight Prediction on Fetal Ultrasound Video

The Residual Transformer Module is proposed which extends a 3D ResNet-based network for analysis of 2 D + t spatio-temporal ultrasound video scans and shows that BabyNet outperforms several state-of-the-art methods and estimates the weight at birth with accuracy comparable to human experts.

Adaptive Contrast for Image Regression in Computer-Aided Disease Assessment

This paper proposes the first contrastive learning framework for deep image regression, namely AdaCon, which consists of a feature learning branch via a novel adaptive-margin contrastive loss and a regression prediction branch that can be used as a plug-and-play module to improve performance of existing regression methods.

Transformers in Medical Image Analysis: A Review

D'ARTAGNAN: Counterfactual Video Generation

This work combines deep neural networks, twin causal networks and generative adversarial methods for the first time to build D’ARTAGNAN (Deep ARtificial Twin-Architecture GeNerAtive Networks), a novel causal generative model.



Video-based AI for beat-to-beat assessment of cardiac function

A video-based deep learning algorithm that surpasses the performance of human experts in the critical tasks of segmenting the left ventricle, estimating ejection fraction and assessing cardiomyopathy is presented.

Recognizing End-Diastole and End-Systole Frames via Deep Temporal Regression Network

A novel deep learning architecture is proposed, named as temporal regression network (TempReg-Net), to accurately identify specific frames from MRI sequences, by integrating the Convolutional Neural Network (CNN) with the Recurrent Neural network (RNN).

Fully Automatic Real-Time Ejection Fraction and MAPSE Measurements in 2D Echocardiography Using Deep Neural Networks

It is concluded that deep learning can be used to fully automate cardiac ultrasound measurements in real-time while scanning, however more work remains to improve the accuracy.

Estimation of End-Diastole in Cardiac Spectral Doppler Using Deep Learning

A deep learning-based method for automatic detection of the end-diastoles in spectral Doppler spectrograms using a combination of a convolutional neural network for extracting features and an RNN for modeling temporal relations is presented.

Joint Learning of Motion Estimation and Segmentation for Cardiac MR Image Sequences

Experimental results show that the joint learning of both tasks is complementary and the proposed models outperform the competing methods significantly in terms of accuracy and speed.

Automatic detection of end-diastole and end-systole from echocardiography images using manifold learning

The manifold learning algorithm is applied on 2D echocardiography images to find out the relationship between the frames of one cycle of heart motion and the results on six healthy volunteers depict the usefulness of the presented method.

Comparing cardiac ejection fraction estimation algorithms without a gold standard.

The Segmentation of the Left Ventricle of the Heart From Ultrasound Data Using Deep Learning Architectures and Derivative-Based Search Methods

A new supervised learning model designed for the automatic segmentation of the left ventricle of the heart in ultrasound images is presented and it is shown that efficient search methods reduce up to tenfold the complexity of the method while still producing competitive segmentations.

SonoNet: Real-Time Detection and Localisation of Fetal Standard Scan Planes in Freehand Ultrasound

A novel method based on convolutional neural networks is proposed, which can automatically detect 13 fetal standard views in freehand 2-D ultrasound data as well as provide a localization of the fetal structures via a bounding box while providing optimal output for the localization task.

U-Net: Convolutional Networks for Biomedical Image Segmentation

It is shown that such a network can be trained end-to-end from very few images and outperforms the prior best method (a sliding-window convolutional network) on the ISBI challenge for segmentation of neuronal structures in electron microscopic stacks.