MRI Image Classification Using Metadata Grouping and Label Voting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current MRI imaging systems lack an automated process for determining sequence type, imaging parameters, and other information, often requiring manual intervention and being inconsistent due to variability in scanner manufacturers, protocols, and metadata inconsistencies.
Innovation Solution
A system and method for automatically classifying MRI images using metadata from DICOM headers, sorting images into groups, and applying machine learning classifiers to identify sequence type, view, and anatomical regions, reducing manual effort and enhancing consistency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual verification of MRI image metadata is performed, then accuracy of sequence type identification is improved, but productivity and workflow efficiency deteriorate
Solution Approach 1:
The system performs self-service by automatically classifying MRI images using machine learning models that analyze image content and metadata. The classifier independently determines sequence type, anatomy, and view type without requiring manual verification, thereby maintaining accuracy while eliminating the productivity loss associated with manual review.
Solution Approach 2:
The manual mechanical process of technician review is replaced with an automated electronic classification system. The machine learning classifier processes images and metadata computationally, substituting human manual verification with an automated algorithmic system that achieves both high accuracy and improved workflow efficiency.
2Reliability
If manual verification of MRI image metadata is performed, then reliability of image classification is improved, but device complexity and operational complexity increase
Solution Approach 1:
The classification system serves itself by automatically generating reliable classifications through machine learning models. The system self-validates its outputs by analyzing both metadata and image content, eliminating the need for complex manual verification workflows and reducing operational complexity while maintaining high reliability.
Solution Approach 2:
The system performs preliminary classification actions automatically before any potential manual review. By pre-classifying images with high reliability through trained machine learning models, the system reduces the need for subsequent manual verification steps, thereby simplifying the overall operational process while maintaining classification reliability.
3Productivity
If automated classification is implemented, then productivity and workflow efficiency are improved, but measurement precision and classification accuracy may deteriorate
Solution Approach 1:
The automated system replaces manual classification with machine learning-based electronic analysis. The classifier processes both metadata and image content using trained models, achieving high accuracy automatically without requiring manual intervention, thus improving productivity while maintaining measurement precision.
Solution Approach 2:
The system incorporates feedback mechanisms where classification results are continuously refined based on analysis of both metadata and image content. The machine learning models learn from training data and can be retrained to improve accuracy, ensuring that automated classification maintains high precision while delivering improved workflow efficiency.
4Measurement precision
If comprehensive manual review of all images is performed, then classification accuracy is improved, but loss of time and operational duration increase
Solution Approach 1:
The system performs self-service classification by automatically analyzing all images through machine learning models. This eliminates the time-consuming manual review process while maintaining classification accuracy, as the automated classifier processes images rapidly without the time losses associated with human review.
Solution Approach 2:
The automated classification system operates continuously without interruption, processing images in sequence without the breaks, fatigue, and time losses inherent in manual review. The machine learning classifier maintains consistent accuracy across all images while eliminating the time loss associated with comprehensive manual examination.
Data Source
AI summary
Systems and methods for performing image classification are disclosed. The methods include receiving a plurality of magnetic resonance imaging (MM) images that each include metadata. The plurality of images are sorted into one or more groups using the metadata. The methods further include, for each of the one or more groups: identifying a subset of images, generating for a classification label for each image in the subset of images using a classifier, identifying a first classification label that is associated with a maximum number of images in the subset of images, and assigning the first classification label to each image in that group.