DCNN Segmentation of Myocardial Perfusion MRI
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for automatic segmentation of myocardial first-pass perfusion images in MRI face challenges such as varied image contrast and poor contrast before contrast arrival, limiting the accuracy and robustness of pixel-to-pixel correspondence across frames, especially in the presence of cardiac and respiratory motion.
Innovation Solution
A deep convolutional neural network (DCNN) is developed and trained using augmented MRI data to segment the endocardium and epicardium layers within images, allowing for accurate calculation of myocardial blood flow by learning to suppress artifact-generating signals and improve segmentation quality across all frames.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual segmentation is performed on single frames, then the process is simple and quick, but the accuracy and robustness against motion are poor
Solution Approach 1:
The patent applies preliminary action by performing segmentation on multiple pre-acquired frames before contrast arrival and during the first-pass perfusion phase. This allows the system to establish pixel-to-pixel correspondence across frames in advance, creating a robust foundation for subsequent blood flow quantification that is resistant to cardiac and respiratory motion.
Solution Approach 2:
The patent merges multiple segmentation results from different frames to create a composite segmentation. By combining information from pre-contrast frames and first-pass perfusion frames, the system achieves more accurate and motion-robust segmentation than would be possible with single-frame manual segmentation.
2Measurement precision
If automatic segmentation is attempted on all frames, then pixel-to-pixel correspondence accuracy improves, but the complexity of handling varied image contrast increases
Solution Approach 1:
The system performs segmentation on pre-contrast frames before the complex first-pass perfusion phase begins. This preliminary segmentation establishes the anatomical framework and pixel correspondence that can be applied throughout the subsequent perfusion frames, simplifying the overall processing complexity.
Solution Approach 2:
The patent employs dynamic adaptation by adjusting segmentation parameters and strategies based on the specific characteristics of each frame type (pre-contrast vs. first-pass perfusion). The system dynamically handles contrast arrival by using different segmentation approaches for different phases of the perfusion study.
3Measurement precision
If multiple acquisition steps are used to improve segmentation quality, then segmentation accuracy improves, but scan time increases
Solution Approach 1:
The patent combines segmentation information from pre-contrast frames with first-pass perfusion frames in a unified processing framework. This merging approach allows the system to achieve high segmentation quality using data already acquired during the standard first-pass perfusion protocol, without requiring additional acquisition steps.
Solution Approach 2:
The segmentation system is designed to handle multiple functions using the same acquired data: it performs both anatomical segmentation and perfusion quantification from the same first-pass perfusion images. This multi-functionality eliminates the need for separate acquisition steps dedicated solely to segmentation.
4Measurement precision
If segmentation is performed before contrast arrival, then the contrast arrival timing can be accurately identified, but the image contrast is very poor
Solution Approach 1:
The system performs preliminary segmentation on pre-contrast frames to establish anatomical structures and pixel correspondence before contrast arrival. This preliminary action enables accurate identification of contrast arrival timing and location by providing a reference framework against which subsequent contrast-enhanced frames can be compared.
Solution Approach 2:
The patent employs dynamic contrast enhancement techniques that adapt to the varying image contrast conditions. The system dynamically adjusts segmentation parameters and processing strategies based on whether the current frame is pre-contrast or post-contrast, optimizing performance for each phase despite the poor contrast before contrast arrival.
Data Source
AI summary
A computerized system and method of modeling myocardial tissue perfusion can include acquiring a plurality of original frames of magnetic resonance imaging (MRI) data representing images of a heart of a subject and developing a manually segmented set of ground truth frames from the original frames. Applying training augmentation techniques to a training set of the originals frame of MRI data can prepare the data for training at least one convolutional neural network (CNN). The CNN can segment the training set of frames according to the ground truth frames. Applying the respective input test frames to a trained CNN can allow for segmenting an endocardium layer and an epicardium layer within the respective images of the input test frames. The segmented images can be used in calculating myocardial blood flow into the myocardium from segmented images of the input test frames.