Deep Learning-based Automated Segmentation of Left Ventricular Trabeculations and Myocardium on Cardiac MR Images: A Feasibility Study

Axel Bartoli; Joris Fournel; Zakarya Bentatou; Gilbert Habib; Alain Lalande; Monique Bernard; Loïc Boussel; François Pontana; Jean-Nicolas Dacher; Badih Ghattas; Alexis Jacquier

doi:10.1148/ryai.2020200021

Deep Learning-based Automated Segmentation of Left Ventricular Trabeculations and Myocardium on Cardiac MR Images: A Feasibility Study

Radiol Artif Intell. 2020 Nov 25;3(1):e200021. doi: 10.1148/ryai.2020200021. eCollection 2021 Jan.

Affiliation

¹ Departments of Radiology (A.B., A.J.) and Cardiology (G.H.), Hôpital de la Timone Adultes, AP-HM, 264, rue Saint-Pierre 13385 Marseille Cedex 05, France; CRMBM-UMR CNRS 7339, Medical Faculty, Aix-Marseille University, Marseille, France (A.B., J.F., Z.B., M.B., A.J.); I2M-UMR CNRS 7373, Aix-Marseille University, Centrale Marseille, Marseille, France (J.F., B.G.); ImVia Laboratory and University Hospital of Dijon, Bourgogne-Franche Comté University, Dijon, France (A.L.); Department of Radiology, Hôpital de la Croix-Rousse, Hospices Civils de Lyon, Lyon, France (L.B.); Department of Cardiovascular Imaging, Lille University Hospital, Lille, France (F.P.); and Department of Diagnostic Imaging, Rouen University Hospital, Rouen, France (J.N.D.).

Abstract

Purpose: To develop and evaluate a complete deep learning pipeline that allows fully automated end-diastolic left ventricle (LV) cardiac MRI segmentation, including trabeculations and automatic quality control of the predicted segmentation.

Materials and methods: This multicenter retrospective study includes training, validation, and testing datasets of 272, 27, and 150 cardiac MR images, respectively, collected between 2012 and 2018. The reference standard was the manual segmentation of four LV anatomic structures performed on end-diastolic short-axis cine cardiac MRI: LV trabeculations, LV myocardium, LV papillary muscles, and the LV blood cavity. The automatic pipeline was composed of five steps with a DenseNet architecture. Intraobserver agreement, interobserver agreement, and interaction time were recorded. The analysis includes the correlation between the manual and automated segmentation, a reproducibility comparison, and Bland-Altman plots.

Results: The automated method achieved mean Dice coefficients of 0.96 ± 0.01 (standard deviation) for LV blood cavity, 0.89 ± 0.03 for LV myocardium, and 0.62 ± 0.08 for LV trabeculation (mean absolute error, 3.63 g ± 3.4). Automatic quantification of LV end-diastolic volume, LV myocardium mass, LV trabeculation, and trabeculation mass-to-total myocardial mass (TMM) ratio showed a significant correlation with the manual measures (r = 0.99, 0.99, 0.90, and 0.83, respectively; all P < .01). On a subset of 48 patients, the mean Dice value for LV trabeculation was 0.63 ± 0.10 or higher compared with the human interobserver (0.44 ± 0.09; P < .01) and intraobserver measures (0.58 ± 0.09; P < .01). Automatic quantification of the trabeculation mass-to-TMM ratio had a higher correlation (0.92) compared with the intra- and interobserver measures (0.74 and 0.39, respectively; both P < .01).

Conclusion: Automated deep learning framework can achieve reproducible and quality-controlled segmentation of cardiac trabeculations, outperforming inter- and intraobserver analyses.Supplemental material is available for this article.© RSNA, 2020.

2021 by the Radiological Society of North America, Inc.