Cardiac MR Motion Parameter Detection via Neural Network Landmark Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for determining motion parameters of the heart, especially for heart failure with preserved ejection fraction, are time-consuming and do not provide satisfactory results due to the complexity of the disease and its early subtle effects on heart motion.
Innovation Solution
A method using two trained convolutional neural networks to analyze cardiac MR images, where the first network identifies anatomical landmarks and aligns images, and the second network extracts time-resolved motion parameters based on these landmarks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional methods are used to extract motion parameters from echo-doppler acquisitions and MR images, then measurement precision can be achieved, but the process becomes time-consuming and productivity decreases
Solution Approach 1:
The patent replaces traditional mechanical/image processing methods with a neural network-based system. The neural network automatically extracts motion parameters from cardiac MR images, substituting the manual or algorithmic image processing pipeline with an intelligent system that learns optimal feature extraction, thereby improving both speed and accuracy
Solution Approach 2:
The neural network performs self-learning and automatic parameter extraction without requiring manual intervention or complex preprocessing steps. The system autonomously identifies anatomical landmarks, tracks motion, and determines parameters, making the process self-service and significantly reducing time consumption
2Measurement precision
If traditional image analysis methods are applied to detect subtle heart motion effects, then measurement precision is maintained, but device complexity and difficulty of detection increase
Solution Approach 1:
The patent replaces complex traditional image analysis algorithms with a neural network that automatically learns detection patterns. The neural network handles the complexity of detecting subtle motion effects by learning from training data, eliminating the need for manual tuning of detection parameters and reducing the difficulty of measurement
Solution Approach 2:
The neural network transforms the detection problem by learning optimal parameter representations from training data. It automatically adapts to detect subtle motion effects by adjusting its internal parameters during training, making the detection process simpler while maintaining high precision
3Measurement precision
If manual analysis of mitral valve annulus motion is performed, then measurement precision can be achieved, but the extent of automation decreases and time consumption increases
Solution Approach 1:
The patent replaces manual landmark identification and motion analysis with a neural network system that automatically performs these tasks. The neural network identifies anatomical landmarks and tracks their motion through the cardiac cycle, achieving full automation while maintaining or improving detection accuracy
Solution Approach 2:
The neural network system performs self-service by automatically extracting features, identifying landmarks, and determining motion parameters without human intervention. The system learns from training data and independently applies this knowledge to analyze new cardiac images, achieving complete automation
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention relates A method for determining a motion parameter of a heart, including determining a sequence of cardiac MR images (61, 62) showing a time resolved motion of the heart. A subset of the sequence of cardiac MR images is applied as a first input to a first trained convolutional neural network (40) configured to determine, as first output a probability distribution of at least 2 anatomical landmarks in the subset. The sequence of cardiac MR images is cropped and realigned based on the at least 2 anatomical landmarks, in order to determine a reframed and aligned sequence of new cardiac MR images (64) showing the same orientation of the heart. The reframed and aligned sequence of new cardiac MR images (64)is applied to a second trained convolutional neural network (41) configured to determine, as second output, a further probability distribution of the at least 2 anatomical landmarks in each new MR image of the reframed and aligned sequence, the motion parameter of the heart is determined based on the second output.