Automatic Lung Segmentation in MRI Videos Using Key Landmarks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Automatic lung segmentation in MRI videos is challenging due to large variations in lung shape and size among patients and variations in contrast between MRI scans, requiring efficient and robust methods to accelerate diagnosis.
Innovation Solution
A progressive two-stage lung segmentation method is employed, detecting key landmarks and deforming a statistically obtained shape using a thin plate splines transform, with key landmarks detected through discriminative joint contexts and a probabilistic boosting tree classifier, and refining the lung boundary using a trained boundary detector.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automatic lung segmentation is performed in MRI videos, then diagnosis time is reduced and productivity is improved, but the complexity of the segmentation system increases
Solution Approach 1:
The segmentation process is divided into two distinct stages: key landmark detection and lung boundary segmentation. This segmentation allows each stage to be optimized independently, with landmark detection handling the complex task of identifying anatomical points and the boundary segmentation using the detected landmarks to produce the final lung mask, thereby reducing overall system complexity while maintaining high productivity
Solution Approach 2:
Key landmarks are detected in advance before performing the actual lung boundary segmentation. These pre-detected landmarks serve as constraints and guides for the subsequent boundary segmentation process, enabling the system to achieve accurate segmentation faster by using the preliminary landmark information to constrain the segmentation algorithm
2Measurement precision
If key landmarks are detected using discriminative joint contexts, then lung segmentation accuracy is improved, but computational time and processing complexity increase
Solution Approach 1:
The accuracy-time tradeoff is resolved by segmenting the computation into two phases: first detecting key landmarks using discriminative joint contexts (which is computationally intensive but only needs to be done once per image), then using these pre-detected landmarks to constrain the boundary segmentation (which is faster because the complex landmark detection is already complete). This segmentation of computation allows high accuracy without proportional increase in total processing time
Solution Approach 2:
The computationally intensive key landmark detection is performed as a preliminary action before the main segmentation task. By pre-detecting the landmarks and using them as constraints, the subsequent boundary segmentation can proceed much faster, as it only needs to refine the boundary rather than detect all anatomical features from scratch, thus achieving high accuracy with reduced total computational time
Data Source
AI summary
A method and system for automatic lung segmentation in magnetic resonance imaging (MRI) images and videos is disclosed. A plurality of predetermined key landmarks of a lung are detected in an MRI image. The key landmarks may be detected using discriminative joint contexts representing combinations of multiple key landmarks. A lung boundary is segmented in the MRI image based on the detected key landmarks. The landmark detection and the lung boundary segmentation can be repeated in each frame of an MRI video.