UCL Discovery
UCL home » Library Services » Electronic resources » UCL Discovery

Landmark Detection in Cardiac MRI Using a Convolutional Neural Network

Xue, H; Artico, J; Fontana, M; Moon, JC; Davies, RH; Kellman, P; (2021) Landmark Detection in Cardiac MRI Using a Convolutional Neural Network. Radiology: Artificial Intelligence , Article e200197. 10.1148/ryai.2021200197. (In press). Green open access

[thumbnail of ryai.2021200197.pdf]
Preview
Text
ryai.2021200197.pdf - Accepted Version

Download (3MB) | Preview

Abstract

Purpose: To develop a convolutional neural network (CNN) solution for landmark detection in cardiac MRI. / Materials and Methods: This retrospective study included cine, late-gadolinium enhancement (LGE), and T1 mapping scans from two hospitals. The training set included 2329 patients (34019 images; mean age 54.1 years; 1471 men; December 2017-March 2020). A hold-out test set included 531 patients (7723 images; mean age 51.5 years, 323 men; May 2020-July 2020). CNN models were developed to detect two mitral valve plane and apical points on long-axis images. On short-axis images, anterior and posterior right ventricular insertion points and left ventricle center were detected. Model outputs were compared with manual labels by two readers. The trained model was deployed to MR scanners. / Results: For the long-axis images, successful detection of cardiac landmarks ranged from 99.7% to 100% for cine images and from 99.2% to 99.5% for LGE images. For the short-axis, detection rates was 96.6% for cine, 97.6% for LGE, and 98.9% for T1-mapping. The Euclidean distances between model and manual labels ranged from 2 to 3.5 mm for different landmarks, indicating close agreement between model landmarks to manual labels. No differences were found for the anterior right ventricular insertion angle and left ventricle length by the models and readers for all views and imaging sequences. Model inference on MR scanner took 610 msec on the graphics processing unit and 5.6 sec on central processing unit, respectively, for a typical cardiac cine series. / Conclusion: A CNN was developed for landmark detection in both long and short-axis cardiac MR images for cine, LGE and T1 mapping sequences, with the accuracy comparable to the interreader variation.

Type: Article
Title: Landmark Detection in Cardiac MRI Using a Convolutional Neural Network
Open access status: An open access version is available from UCL Discovery
DOI: 10.1148/ryai.2021200197
Publisher version: https://doi.org/10.1148/ryai.2021200197
Language: English
Additional information: This is an Open Access article published under a Creative Commons Attribution 4.0 International (CC BY 4.0) Licence (https://creativecommons.org/licenses/by/4.0/).
UCL classification: UCL
UCL > Provost and Vice Provost Offices > School of Life and Medical Sciences
UCL > Provost and Vice Provost Offices > School of Life and Medical Sciences > Faculty of Medical Sciences
UCL > Provost and Vice Provost Offices > School of Life and Medical Sciences > Faculty of Medical Sciences > Div of Medicine
UCL > Provost and Vice Provost Offices > School of Life and Medical Sciences > Faculty of Medical Sciences > Div of Medicine > Inflammation
UCL > Provost and Vice Provost Offices > School of Life and Medical Sciences > Faculty of Population Health Sciences > Institute of Cardiovascular Science
UCL > Provost and Vice Provost Offices > School of Life and Medical Sciences > Faculty of Population Health Sciences > Institute of Cardiovascular Science > Clinical Science
URI: https://discovery.ucl.ac.uk/id/eprint/10131998
Downloads since deposit
60Downloads
Download activity - last month
Download activity - last 12 months
Downloads by country - last 12 months

Archive Staff Only

View Item View Item