Neural network-based left ventricle geometry prediction from CMR images with application in biomechanics

作者：

Highlights：

• Automatic CNN-based framework for predicting LV geometries directly from CMR images

• Two-stage approach combines a segmentation network and geometry prediction network.

• Geometry prediction based on principal component analysis for dimensionality reduction

• CNN-predicted LV geometries useful in biomechanical studies of LV stiffness estimation

• Stepping stone towards the development of LV geometry-heterogeneous emulators

摘要

•Automatic CNN-based framework for predicting LV geometries directly from CMR images•Two-stage approach combines a segmentation network and geometry prediction network.•Geometry prediction based on principal component analysis for dimensionality reduction•CNN-predicted LV geometries useful in biomechanical studies of LV stiffness estimation•Stepping stone towards the development of LV geometry-heterogeneous emulators

论文关键词：3D reconstruction,Myocardium,Convolutional neural network,Deep learning,Cardiac Mechanics,Cardiac Magnetic Resonance Imaging

论文评审过程：Received 21 December 2020, Revised 10 June 2021, Accepted 3 August 2021, Available online 11 August 2021, Version of Record 26 August 2021.

论文官网地址：https://doi.org/10.1016/j.artmed.2021.102140