Deep Learning Multi-Mask Segmentation for MRI Anatomical Stratification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current MRI systems face challenges in efficiently performing whole body scanning and spine scanning due to the need for multiple localizer images and anatomical stratification, which requires significant technologist effort and involves complex image processing techniques like cropping, pasting, and stitching.
Innovation Solution
A computer-implemented method and system utilizing a deep learning-based multi-mask segmentation network to perform region stratification on calibration or low-resolution images from an MRI scanner. This involves obtaining calibration scan data, inputting it into the network, and outputting labeled mask images to determine the extent of anatomical landmarks, thereby automating the process of localizer setup and image acquisition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple localizer images are acquired for whole body or spine scanning, then anatomical stratification accuracy is improved, but technologist effort and processing time increase significantly
Solution Approach 1:
The system performs self-service by automatically acquiring multiple localizer images and performing anatomical stratification without requiring technologist intervention. The processor autonomously determines anatomical boundaries and generates station-specific imaging protocols, eliminating manual processing while maintaining high accuracy.
Solution Approach 2:
The system performs preliminary action by acquiring multiple localizer images and performing anatomical stratification before the actual diagnostic scanning begins. This preliminary processing establishes all necessary anatomical boundaries and station definitions in advance, so that subsequent high-resolution scanning can proceed efficiently without interruptions for manual adjustments.
2Adaptability or versatility
If manual image processing techniques (cropping, pasting, stitching) are used for anatomical stratification, then flexibility in handling different anatomical regions is improved, but process complexity increases
Solution Approach 1:
The system replaces manual mechanical image processing operations (cropping, pasting, stitching) with automated computational image processing. The processor automatically performs all necessary image manipulation operations based on detected anatomical landmarks, maintaining the flexibility to handle various anatomical regions while eliminating the complexity of manual procedures.
Solution Approach 2:
The system applies segmentation by automatically dividing the imaged volume into distinct anatomical stations based on detected landmarks. The processor identifies boundaries between stations (e.g., cervical, thoracic, lumbar spine regions) and automatically segments the imaging protocol accordingly, providing flexibility without manual intervention.
3Area of stationary object
If multiple stations are involved in spine scanning, then complete coverage of all spinal regions is improved, but scanning efficiency deteriorates
Solution Approach 1:
The system performs preliminary action by automatically determining the extent of each spinal region (cervical, thoracic, lumbar) from localizer images before the diagnostic scan begins. This preliminary anatomical stratification allows the system to pre-configure station boundaries and scanning parameters, enabling complete coverage of all spinal regions to be achieved efficiently in a single coordinated scan rather than through multiple separate scanning sessions.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The solution enables automatic patient-specific extent estimation for correct localizer scans, reduces the time required for setting up subsequent scans, and simplifies the process of anatomical region stratification, ensuring accurate and efficient MRI scanning.
Implementation Method 1
inputting the calibration data or the low resolution images into a trained deep learning-based multi-mask segmentation network
Implementation Method 2
The resulting set of received nuclear magnetic resonance (NMR) signals are digitized and processed to reconstruct the image
Data Source
AI summary
A method and a system include obtaining calibration scan data or low resolution images of a subject acquired with a magnetic resonance (MR) scanner of an MR imaging system. The method and the system also include inputting the calibration data or the low resolution images into a trained deep learning-based multi-mask segmentation network. The method and the system further include outputting labeled mask images for different anatomical stations, wherein a respective mask of a respective labeled mask image highlights an anatomical landmark of interest in each respective anatomical station of the different anatomical stations. The method and the system even further include determining an extent of a respective anatomical landmark of interest in each respective anatomical station of the different anatomical stations for a respective localizer scan for each respective anatomical station based at least on a respective label mask image for each respective anatomical station.