Deformable Image Registration for Self-supervised Cardiac Phase Detection in Multi-View Multi-Disease Cardiac Magnetic Resonance Images

2025-10-07

AI Summary

Cardiovascular magnetic resonance (CMR) temporal analysis relies on accurate detection of cardiac-phase keyframes, yet conventional methods identify end-systole (ES) and end-diastole (ED) solely from left ventricular volume curves—failing to capture myocardial motion dynamics. This work proposes a self-supervised deep learning framework that jointly leverages dense deformable image registration and 1D motion curve modeling to automatically detect five clinically relevant keyframes from multi-view CMR acquisitions (5 short-axis + 4 long-axis slices). The method eliminates dependence on volumetric curves and instead incorporates rule-guided motion feature identification. It is trained and validated on multi-center, multi-disease data. Quantitatively, it improves ED/ES detection accuracy by 30–51% (short-axis) and 11–47% (long-axis) over conventional volume-based approaches, achieving mean frame errors of 1.31 (short-axis) and 1.73 (long-axis). Moreover, it demonstrates markedly enhanced generalizability across diverse cardiac pathologies.

Technology Category

• Computer Vision: Motion & Tracking
• Machine Learning: Multi-instance/Multi-view Learning
• Data Mining & Knowledge Management: Mining of Visual, Multimedia & Multimodal Data

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Cardiovascular magnetic resonance (CMR) is the gold standard for assessing cardiac function, but individual cardiac cycles complicate automatic temporal comparison or sub-phase analysis. Accurate cardiac keyframe detection can eliminate this problem. However, automatic methods solely derive end-systole (ES) and end-diastole (ED) frames from left ventricular volume curves, which do not provide a deeper insight into myocardial motion. We propose a self-supervised deep learning method detecting five keyframes in short-axis (SAX) and four-chamber long-axis (4CH) cine CMR. Initially, dense deformable registration fields are derived from the images and used to compute a 1D motion descriptor, which provides valuable insights into global cardiac contraction and relaxation patterns. From these characteristic curves, keyframes are determined using a simple set of rules. The method was independently evaluated for both views using three public, multicentre, multidisease datasets. M&Ms-2 (n=360) dataset was used for training and evaluation, and M&Ms (n=345) and ACDC (n=100) datasets for repeatability control. Furthermore, generalisability to patients with rare congenital heart defects was tested using the German Competence Network (GCN) dataset. Our self-supervised approach achieved improved detection accuracy by 30% - 51% for SAX and 11% - 47% for 4CH in ED and ES, as measured by cyclic frame difference (cFD), compared with the volume-based approach. We can detect ED and ES, as well as three additional keyframes throughout the cardiac cycle with a mean cFD below 1.31 frames for SAX and 1.73 for LAX. Our approach enables temporally aligned inter- and intra-patient analysis of cardiac dynamics, irrespective of cycle or phase lengths. GitHub repository: https://github.com/Cardio-AI/cmr-multi-view-phase-detection.git

Problem

• Detecting cardiac keyframes beyond standard ES/ED for detailed motion analysis
• Enabling temporal alignment of cardiac dynamics across patients and cycles
• Providing self-supervised phase detection in multi-view multi-disease CMR images

Innovation

• Self-supervised deep learning detects cardiac keyframes
• Dense deformable registration fields compute motion descriptor
• Characteristic curves determine keyframes using simple rules