Cardiac magnetic resonance (CMR) imaging allows precise non-invasive quantification of cardiac function. It requires reliable image segmentation for myocardial tissue. Clinically used software usually offers automatic approaches for this step. These are, however, designed for segmentation of human images obtained at clinical field strengths. They reach their limits when applied to preclinical data and ultrahigh field strength (such as CMR of pigs at 7 T). In our study, eleven animals (seven with myocardial infarction) underwent four CMR scans each. Short-axis cine stacks were acquired and used for functional cardiac analysis. End-systolic and end-diastolic images were labelled manually by two observers and inter- and intra-observer variability were assessed. Aiming to make the functional analysis faster and more reproducible, an established deep learning (DL) model for myocardial segmentation in humans was re-trained using our preclinical 7 T data (nþinspace=þinspace772 images and labels). We then tested the model on nþinspace=þinspace288 images. Excellent agreement in parameters of cardiac function was found between manual and DL segmentation: For ejection fraction (EF) we achieved a Pearson's r of 0.95, an Intraclass correlation coefficient (ICC) of 0.97, and a Coefficient of variability (CoV) of 6.6\%. Dice scores were 0.88 for the left ventricle and 0.84 for the myocardium.
Description
Cardiac function in a large animal model of myocardial infarction at 7 T: deep learning based automatic segmentation increases reproducibility | Scientific Reports
%0 Journal Article
%1 Kollmann2024
%A Kollmann, Alena
%A Lohr, David
%A Ankenbrand, Markus J.
%A Bille, Maya
%A Terekhov, Maxim
%A Hock, Michael
%A Elabyad, Ibrahim
%A Baltes, Steffen
%A Reiter, Theresa
%A Schnitter, Florian
%A Bauer, Wolfgang R.
%A Hofmann, Ulrich
%A Schreiber, Laura M.
%D 2024
%J Scientific Reports
%K myown cctb bmd markusankenbrand from:iimog
%N 1
%P 11009
%R 10.1038/s41598-024-61417-4
%T Cardiac function in a large animal model of myocardial infarction at 7 T: deep learning based automatic segmentation increases reproducibility
%U https://doi.org/10.1038/s41598-024-61417-4
%V 14
%X Cardiac magnetic resonance (CMR) imaging allows precise non-invasive quantification of cardiac function. It requires reliable image segmentation for myocardial tissue. Clinically used software usually offers automatic approaches for this step. These are, however, designed for segmentation of human images obtained at clinical field strengths. They reach their limits when applied to preclinical data and ultrahigh field strength (such as CMR of pigs at 7 T). In our study, eleven animals (seven with myocardial infarction) underwent four CMR scans each. Short-axis cine stacks were acquired and used for functional cardiac analysis. End-systolic and end-diastolic images were labelled manually by two observers and inter- and intra-observer variability were assessed. Aiming to make the functional analysis faster and more reproducible, an established deep learning (DL) model for myocardial segmentation in humans was re-trained using our preclinical 7 T data (nþinspace=þinspace772 images and labels). We then tested the model on nþinspace=þinspace288 images. Excellent agreement in parameters of cardiac function was found between manual and DL segmentation: For ejection fraction (EF) we achieved a Pearson's r of 0.95, an Intraclass correlation coefficient (ICC) of 0.97, and a Coefficient of variability (CoV) of 6.6\%. Dice scores were 0.88 for the left ventricle and 0.84 for the myocardium.
@article{Kollmann2024,
abstract = {Cardiac magnetic resonance (CMR) imaging allows precise non-invasive quantification of cardiac function. It requires reliable image segmentation for myocardial tissue. Clinically used software usually offers automatic approaches for this step. These are, however, designed for segmentation of human images obtained at clinical field strengths. They reach their limits when applied to preclinical data and ultrahigh field strength (such as CMR of pigs at 7 T). In our study, eleven animals (seven with myocardial infarction) underwent four CMR scans each. Short-axis cine stacks were acquired and used for functional cardiac analysis. End-systolic and end-diastolic images were labelled manually by two observers and inter- and intra-observer variability were assessed. Aiming to make the functional analysis faster and more reproducible, an established deep learning (DL) model for myocardial segmentation in humans was re-trained using our preclinical 7 T data (n{\thinspace}={\thinspace}772 images and labels). We then tested the model on n{\thinspace}={\thinspace}288 images. Excellent agreement in parameters of cardiac function was found between manual and DL segmentation: For ejection fraction (EF) we achieved a Pearson's r of 0.95, an Intraclass correlation coefficient (ICC) of 0.97, and a Coefficient of variability (CoV) of 6.6{\%}. Dice scores were 0.88 for the left ventricle and 0.84 for the myocardium.},
added-at = {2024-05-17T15:13:43.000+0200},
author = {Kollmann, Alena and Lohr, David and Ankenbrand, Markus J. and Bille, Maya and Terekhov, Maxim and Hock, Michael and Elabyad, Ibrahim and Baltes, Steffen and Reiter, Theresa and Schnitter, Florian and Bauer, Wolfgang R. and Hofmann, Ulrich and Schreiber, Laura M.},
biburl = {https://www.bibsonomy.org/bibtex/208c3bce6b92429eaa6f6eb0b643e248c/cctb},
day = 14,
description = {Cardiac function in a large animal model of myocardial infarction at 7 T: deep learning based automatic segmentation increases reproducibility | Scientific Reports},
doi = {10.1038/s41598-024-61417-4},
interhash = {64ac491ec24dfc946f1a777cbf5605cb},
intrahash = {08c3bce6b92429eaa6f6eb0b643e248c},
issn = {2045-2322},
journal = {Scientific Reports},
keywords = {myown cctb bmd markusankenbrand from:iimog},
month = may,
number = 1,
pages = 11009,
timestamp = {2024-05-17T15:13:43.000+0200},
title = {Cardiac function in a large animal model of myocardial infarction at 7 T: deep learning based automatic segmentation increases reproducibility},
url = {https://doi.org/10.1038/s41598-024-61417-4},
volume = 14,
year = 2024
}