Ventricle Segmentation in Contrast-Enhanced Cine MRI
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automatic segmentation methods for ventricles in contrast-enhanced cine MRI datasets face challenges due to increased signal intensity of infarcted myocardium, leading to decreased signal intensity differences with the blood-pool cavity, making accurate segmentation difficult.
Innovation Solution
The method utilizes additional MRI datasets, such as delayed-enhancement and first-pass perfusion datasets, acquired during the same imaging session to provide information that assists in segmenting the ventricle in contrast-enhanced cine MRI datasets by mapping features like infarct and blood-pool cavity information to refine the segmentation process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If contrast-enhanced cine MRI is performed after contrast administration, then time efficiency is improved and patient discomfort is reduced, but signal intensity difference between blood-pool cavity and myocardium decreases making segmentation more difficult
Solution Approach 1:
The patent applies segmentation by dividing the ventricle into distinct regions (endocardial border and epicardial border) and processing them separately through automated segmentation algorithms. This allows precise delineation of ventricular boundaries despite the reduced signal intensity contrast in contrast-enhanced images.
Solution Approach 2:
The patent uses an intermediary approach by incorporating additional image processing techniques and algorithms that act as mediators between the contrast-enhanced image data and the segmentation output. These intermediaries help recover boundary information that is lost due to reduced signal contrast.
2Measurement precision
If manual segmentation is performed to achieve accurate ventricle delineation, then segmentation accuracy is improved, but time consumption increases significantly
Solution Approach 1:
The patent implements self-service through automated segmentation algorithms that perform ventricle delineation without requiring manual intervention. The system uses deformable models and image processing techniques to automatically identify and segment ventricular boundaries, making the process self-sufficient while maintaining accuracy.
Solution Approach 2:
The patent replaces the mechanical manual segmentation process with automated computational methods. Deformable models and image-based algorithms substitute for manual tracing, eliminating the need for clinician time while preserving segmentation accuracy through sophisticated image processing.
3Productivity
If traditional automatic segmentation methods are used on contrast-enhanced cine MRI, then processing speed is improved, but segmentation accuracy deteriorates due to reduced signal intensity difference
Solution Approach 1:
The patent applies dynamics through deformable models that can adapt and change shape dynamically during the segmentation process. These models evolve iteratively to conform to the actual ventricular boundaries, allowing the segmentation to adjust to the reduced signal contrast conditions while maintaining both speed and accuracy.
Solution Approach 2:
The patent utilizes parameter changes by modifying segmentation algorithm parameters specifically for contrast-enhanced images. This includes adjusting threshold values, contrast weights, and model parameters to account for the reduced signal intensity difference, enabling accurate segmentation while maintaining automated processing speed.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for delineating a ventricle from MRI data relating to the heart of a patient, the method comprising: a) providing a contrast-enhanced cine MRI dataset; b) providing one or more additional MRI datasets; c) segmenting one or more features on the additional MRI dataset or datasets; d) mapping the segmented features to the contrast-enhanced cine MRI dataset; e) using the segmented features as mapped in step d) to assist segmentation of the ventricle on the contrast-enhanced cine MRI dataset. A corresponding device and computer program are also disclosed.