Cardiac Motion Estimation with Temporal Feature Refinement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing machine learning-based methods for cardiac motion estimation lack accuracy in determining the motion of anatomical structures like the myocardium, particularly in medical videos.
Innovation Solution
A machine learning model using a twin neural network and transformer or recurrent neural networks to refine image features from multiple image pairs in medical videos, leveraging temporal information to improve motion estimation accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If machine learning-based methods are used for motion estimation, then automation and processing capability are improved, but measurement precision of cardiac motion is insufficient
Solution Approach 1:
The patent segments the cardiac motion estimation task into multiple pairs of images arranged in temporal sequence. Each image pair is processed independently to extract local motion features, which are then integrated to achieve global motion estimation. This segmentation allows the system to maintain automation while improving precision by analyzing temporal relationships between consecutive frames.
Solution Approach 2:
The patent introduces a temporal dimension by processing multiple image pairs in sequence rather than analyzing single isolated images. The machine learning model leverages temporal information across multiple frames to refine motion estimates, transforming the problem from static image analysis to dynamic temporal sequence analysis, thereby improving measurement precision while maintaining automation.
2Measurement precision
If multiple image pairs are processed via machine learning model, then motion estimation accuracy is improved, but computational complexity and processing time increase
Solution Approach 1:
The patent divides the complex task of analyzing entire medical videos into smaller, manageable image pairs. Each image pair is processed independently through the machine learning model to extract local motion features. This segmentation reduces the computational burden on the system while maintaining high accuracy through temporal integration of multiple pairwise comparisons.
Solution Approach 2:
The patent processes multiple image pairs beyond the minimum single pair to achieve accurate motion estimation. By analyzing successive image pairs and integrating their results, the system obtains more comprehensive motion information than would be possible with a single comparison, improving accuracy while distributing computational load across multiple simpler processing steps.
3Measurement precision
If multiple image pairs are processed via machine learning model, then motion estimation accuracy is improved, but processing time increases
Solution Approach 1:
The patent segments video processing into independent image pair analyses that can be processed in parallel or sequentially. Each image pair requires minimal processing time compared to analyzing entire videos, yet the cumulative effect of multiple pairwise comparisons delivers high accuracy motion estimation. This segmentation enables efficient time management while maintaining precision.
Solution Approach 2:
The patent performs preliminary feature extraction and comparison on individual image pairs before integrating results for final motion estimation. This preliminary action on smaller units allows the system to prepare motion features in advance, reducing the computational burden during final integration and thereby decreasing overall processing time while maintaining accuracy.
Data Source
AI summary
A video of medical scan images associated with an anatomical structure may be arranged into multiple image pairs. The multiple image pairs may be provided to a machine learning (ML) model successively and the ML model may determine respective first sets of image features associated with the multiple image pairs and, for each of the multiple image pairs, refine the first set of image features associated with the image pair based on the respective first sets of image features associated with one or more other image pairs. A motion field associated with the image pair may be determined based at least on the refined first set of image features associated with the image pair and a task may be performed based on the respective motion fields.