What does this research mean for the field?

A fully automatic pipeline combining a convolutional neural network for cardiac segmentation and a Random Forest classifier achieves 91% accuracy in diagnosing cardiac diseases from cine MR images. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

August 3, 2017Open Access

Automatic Segmentation and Disease Classification Using Cardiac Cine MR Images

Q: What is the clinical evidence from this study?

Study design: Other. Population: Cardiac diseases (DCM, HCM, MINF, RVA, normal) (n=100). Intervention: Convolutional neural network and Random Forest classifier vs. Manual reference standard. Primary outcome: Overall disease classification accuracy.

Key Result

A fully automatic method using a convolutional neural network and Random Forest classifier achieved an overall disease classification accuracy of 91% using cardiac cine MR images.

Structured PICO

Does a fully automatic CNN and Random Forest classifier accurately segment and classify cardiac diseases using cine MR images?

Population

100 patients from the MICCAI 2017 automated cardiac diagnosis challenge (ACDC) dataset, balanced across 5 classes (normal, dilated cardiomyopathy, hypertrophic cardiomyopathy, heart failure following myocardial infarction, right ventricular abnormality).

Intervention

Fully automatic method for segmentation and disease classification using a convolutional neural network (CNN) and Random Forest classifier on cardiac cine MR images.

Comparator

Manual reference segmentations and reference clinical diagnosis.

Outcome

Segmentation accuracy (Dice similarity coefficient and Hausdorff distance) and disease classification accuracy.surrogate

A fully automatic deep learning and Random Forest pipeline can accurately segment cardiac cine MR images and classify patients into five disease categories with 91% accuracy.

Limitations

Small training dataset
Errors in the segmentation stage may affect performance in the classification stage
Only end-diastole and end-systole images were used, rather than the complete sequence of images

Abstract

Segmentation of the heart in cardiac cine MR is clinically used to quantify cardiac function. We propose a fully automatic method for segmentation and disease classification using cardiac cine MR images. A convolutional neural network (CNN) was designed to simultaneously segment the left ventricle (LV), right ventricle (RV) and myocardium in end-diastole (ED) and end-systole (ES) images. Features derived from the obtained segmentations were used in a Random Forest classifier to label patients as suffering from dilated cardiomyopathy, hypertrophic cardiomyopathy, heart failure following myocardial infarction, right ventricular abnormality, or no cardiac disease. The method was developed and evaluated using a balanced dataset containing images of 100 patients, which was provided in the MICCAI 2017 automated cardiac diagnosis challenge (ACDC). The segmentation and classification pipeline were evaluated in a four-fold stratified cross-validation. Average Dice scores between reference and automatically obtained segmentations were 0.94, 0.88 and 0.87 for the LV, RV and myocardium. The classifier assigned 91% of patients to the correct disease category. Segmentation and disease classification took 5 s per patient. The results of our study suggest that image-based diagnosis using cine MR cardiac scans can be performed automatically with high accuracy.

KI fragen

Bookmark

View Full Paper

Cite This Study

Wolterink et al. (Thu,) conducted a other in Cardiac diseases (DCM, HCM, MINF, RVA, normal) (n=100). Convolutional neural network and Random Forest classifier vs. Manual reference standard was evaluated on Overall disease classification accuracy. A fully automatic method using a convolutional neural network and Random Forest classifier achieved an overall disease classification accuracy of 91% using cardiac cine MR images.

synapsesocial.com/papers/6a1d7c841024216094056d63 https://doi.org/https://doi.org/10.48550/arxiv.1708.01141

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

KI fragen

Bookmark

View Full Paper