Cardiac MRI Segmentation with Strong Anatomical Guarantees

Abstract

Recent publications have shown that the segmentation accuracy of modern-day convolutional neural networks (CNN) applied on cardiac MRI can reach the inter-expert variability, a great achievement in this area of research. However, despite these successes, CNNs still produce anatomically inaccurate segmentations as they provide no guarantee on the anatomical plausibility of their outcome, even when using a shape prior. In this paper, we propose a cardiac MRI segmentation method which always produces anatomically plausible results. At the core of the method is an adversarial variational autoencoder (aVAE) whose latent space encodes a smooth manifold on which lies a large spectrum of valid cardiac shapes. This aVAE is used to automatically warp anatomically inaccurate cardiac shapes towards a close but correct shape. Our method can accommodate any cardiac segmentation method and convert its anatomically implausible results to plausible ones without affecting its overall geometric and clinical metrics. With our method, CNNs can now produce results that are both within the inter-expert variability and always anatomically plausible.

arXiv comment9 pages, accepted for MICCAI 2019; camera ready corrections, acknowledgments

References in corpus (2)

• Auto-Encoding Variational Bayes
• Anatomically Constrained Neural Networks (ACNN): Application to Cardiac Image Enhancement and Segmentation

Cited by in corpus (16)

• A Survey on Incorporating Domain Knowledge into Deep Learning for Medical Image Analysis
• Improving the generalizability of convolutional neural network-based segmentation on CMR images
• Data Augmentation in High Dimensional Low Sample Size Setting Using a Geometry-Based Variational Autoencoder
• Filtering in tractography using autoencoders (FINTA)
• A persistent homology-based topological loss for CNN-based multi-class segmentation of CMR
• On the effectiveness of GAN generated cardiac MRIs for segmentation
• PC-U Net: Learning to Jointly Reconstruct and Segment the Cardiac Walls in 3D from CT Data
• How Can CNNs Use Image Position for Segmentation?
• Learning Directional Feature Maps for Cardiac MRI Segmentation
• Myocardial Segmentation of Cardiac MRI Sequences with Temporal Consistency for Coronary Artery Disease Diagnosis
• Automatic view plane prescription for cardiac magnetic resonance imaging via supervision by spatial relationship between views
• Re-using Adversarial Mask Discriminators for Test-time Training under Distribution Shifts
• Cooperative Training and Latent Space Data Augmentation for Robust Medical Image Segmentation
• Anatomically-Informed Deep Learning on Contrast-Enhanced Cardiac MRI for Scar Segmentation and Clinical Feature Extraction
• Shape Constrained CNN for Cardiac MR Segmentation with Simultaneous Prediction of Shape and Pose Parameters
• Training Automatic View Planner for Cardiac MR Imaging via Self-Supervision by Spatial Relationship between Views