Deep Learning Techniques for Automatic MRI Cardiac Multi-Structures Segmentation and Diagnosis: Is the Problem Solved?

  title={Deep Learning Techniques for Automatic MRI Cardiac Multi-Structures Segmentation and Diagnosis: Is the Problem Solved?},
  author={Olivier Bernard and Alain Lalande and Cl{\'e}ment Zotti and Fr{\'e}d{\'e}ric Cervenansky and Xin Yang and Pheng-Ann Heng and Irem Cetin and Karim Lekadir and Oscar Camara and Miguel A. Gonz{\'a}lez Ballester and Gerard Sanrom{\'a} and Sandy Napel and Steffen Erhard Petersen and Georgios Tziritas and Elias Grinias and Mahendra Khened and Varghese Alex Kollerathu and Ganapathy Krishnamurthi and Marc-Michel Roh{\'e} and Xavier Pennec and Maxime Sermesant and Fabian Isensee and Paul F. J{\"a}ger and Klaus Maier-Hein and Peter M. Full and Ivo Wolf and Sandy Engelhardt and Christian F. Baumgartner and Lisa M. Koch and Jelmer M. Wolterink and Ivana I{\vs}gum and Yeonggul Jang and Yoonmi Hong and Jay Patravali and Shubham Jain and Olivier Humbert and Pierre-Marc Jodoin},
  journal={IEEE Transactions on Medical Imaging},
Delineation of the left ventricular cavity, myocardium, and right ventricle from cardiac magnetic resonance images (multi-slice 2-D cine MRI) is a common clinical task to establish diagnosis. The automation of the corresponding tasks has thus been the subject of intense research over the past decades. In this paper, we introduce the “Automatic Cardiac Diagnosis Challenge” dataset (ACDC), the largest publicly available and fully annotated dataset for the purpose of cardiac MRI (CMR) assessment… 

Deep Learning for Segmentation Using an Open Large-Scale Dataset in 2D Echocardiography

Results show that encoder-decoder-based architectures outperform state-of-the-art non-deep learning methods and faithfully reproduce the expert analysis for the end-diastolic and end-systolic left ventricular volumes, which open the door for accurate and fully-automatic analysis of 2D echocardiographic images.

End-To-End Diagnosis And Segmentation Learning From Cardiac Magnetic Resonance Imaging

This paper proposes a learning method to train diagnosis models based on multi-task learning that jointly trains for the tasks of segmentation and diagnosis classification, and hypothesizes that segmentation has a regularizing effect on the learning of features relevant for diagnosis.

Left ventricle segmentation in LGE-MRI using multiclass learning

A multi-scale fully convolutional neural network with skip-connections (U-Net) and residual units is applied to solve the multiclass segmentation problem in LGE-MRI and segments the endocardium and epicardium of the left ventricle fully automatically with a high accuracy.

Deep Learning Based Cardiac MRI Segmentation: Do We Need Experts?

This paper gauged the performance of three segmentation models, namely U-Net, Attention U- net, and ENet, trained with different loss functions on expert and non-expert groundtruth for cardiac cine-MRI segmentation to explore whether expert knowledge is a strict requirement for the creation of annotated datasets that machine learning can successfully train on.

Improving the generalization of deep learning methods to segment the left ventricle in short axis MR images

This work improves the result between the cross-dataset evaluation of the left ventricle segmentation, reducing the constraints for daily clinical adoption of a fully-automatic segmentation method.

Multi-Disease, Multi-View & Multi-Center Right Ventricular Segmentation in Cardiac MRI (M&Ms-2)

Despite recent advances in deep learning, robust segmentation of the RV continues to pose challenges in practice and there is a need for new methods to handle the inherent geometrical and textural complexities, in particular in the presence of RV related pathologies.

DR-Unet++: An Approach for Left Ventricle Segmentation from Magnetic Resonance Images

The DR-Unet++ model, which utilizes advanced Deep Learning techniques, provides better feature extraction for the segmentation of desired segmented objects and outperforms other modern models in terms of Dice coefficient and IoU metric.

A deep learning approach for the segmentation of myocardial diseases

A SegU-Net model based on the fusion of two deep learning segmentation techniques (U- net and Seg-Net) for automated LGE-MRI (Late gadolinium enhanced magnetic resonance imaging) myocardial disease quantification in a new multifield expert annotated dataset outperforming seven previously proposed methods.

A deep learning pipeline for automatic analysis of multi-scan cardiovascular magnetic resonance

A fully automatic multi-scan CMR image analysis pipeline allows for fast and robust analysis of large CMR studies while guaranteeing reproducibility, hence potentially improving patient’s diagnosis as well as clinical studies outcome.



Automatic Cardiac Disease Assessment on cine-MRI via Time-Series Segmentation and Domain Specific Features

This paper uses an ensemble of UNet inspired architectures for segmentation of cardiac structures such as the left and right ventricular cavity (LVC, RVC) and the left ventricular myocardium (LVM) and a random forest classifier to predict the pathologic target class.

An Exploration of 2D and 3D Deep Learning Techniques for Cardiac MR Image Segmentation

A fully automated framework for segmentation of the left and right ventricular cavities and the myocardium on short-axis cardiac MR images is presented and it is found that processing the images in a slice-by-slice fashion using 2D networks is beneficial due to a relatively large slice thickness.

Human-level CMR image analysis with deep fully convolutional networks

By combining FCN with a large-scale annotated dataset, this work shows for the first time that an automated method achieves a performance on par with human experts in analysing CMR images and deriving clinical measures.

A Fully Convolutional Neural Network for Cardiac Segmentation in Short-Axis MRI

This work proposes to tackle the problem of automated left and right ventricle segmentation through the application of a deep fully convolutional neural network architecture for pixel-wise labeling in cardiac magnetic resonance imaging.

Recurrent Fully Convolutional Neural Networks for Multi-slice MRI Cardiac Segmentation

A recurrent fully-convolutional network (RFCN) that learns image representations from the full stack of 2D slices and has the ability to leverage inter-slice spatial dependences through internal memory units is proposed.

Automated Quality Assessment of Cardiac MR Images Using Convolutional Neural Networks

Inspired by the success of deep learning methods, Convolutional Neural Networks are trained to construct a set of discriminative features for automatic detection of missing slices in Cardiac Magnetic Resonance Imaging scans, which is currently performed by tedious visual assessment.

Cardiac left ventricle segmentation using convolutional neural network regression

This paper designs two convolutional neural networks (CNN), one for localization of the LV, and the other for determining the endocardial radius, and demonstrates that CNN regression is a viable and highly promising method for automated LVendocardial segmentation at ED and ES phases, and is capable of generalizing learning between highly distinct training and testing data sets.

Automatic localization of the left ventricle in cardiac MRI images using deep learning

  • Omar EmadI. YassineA. Fahmy
  • Computer Science
    2015 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC)
  • 2015
A new approach based on deep Convolutional Neural Network to localize the LV in cardiac MRI in short axis views by employing six-layer CNN with different kernel sizes, followed by Softmax fully connected layer for classification.

Class-Balanced Deep Neural Network for Automatic Ventricular Structure Segmentation

This paper proposes a general and fully automatic solution to concurrently segment three important ventricular structures from cardiovascular MR scan and investigates the capacity of different loss functions and proposes a Multi-class Dice Similarity Coefficient (mDSC) based loss function to re-weight the training for all classes.