Deriving MRI Tissue Eigenvalues from CT Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for acquiring MRI images are limited, especially for patients with contraindications to MRI, and require constructing multiple conversion models to derive MRI images with various representation formats from CT images.
Innovation Solution
An image processing device that uses a machine learning-based derivation model to derive tissue eigenvalues from CT images, allowing the generation of MRI images with desired representation formats without the need for multiple conversion models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If a conversion model is constructed by machine learning to convert CT images into MRI images, then it is possible to acquire MRI images only by performing imaging with the CT apparatus, but the conversion model can only derive MRI images with a single representation format, requiring multiple conversion models for different formats
Solution Approach 1:
The conversion model is designed to derive multiple types of MRI images (T1-weighted, T2-weighted, and other representation formats) from a single CT image input. This multi-functional approach eliminates the need for separate conversion models for each MRI representation format, thereby reducing the number of models required while maintaining the ability to generate various MRI types.
Solution Approach 2:
The conversion model utilizes different parameters and processing techniques to derive various MRI representation formats from the same CT image. By changing parameters such as tissue eigenvalue thresholds and calculation methods, the model can generate different MRI types (T1-weighted, T2-weighted, etc.) without requiring separate models for each format.
2Adaptability or versatility
If multiple conversion models are constructed for different MRI representation formats, then various MRI images can be derived from CT images, but the load for constructing the conversion models becomes large
Solution Approach 1:
A single universal conversion model is developed that can derive multiple MRI representation formats (T1-weighted, T2-weighted, and others) from CT images. This eliminates the need to construct and train multiple separate conversion models, significantly reducing the time and computational resources required for model construction while maintaining the ability to generate various MRI types.
Solution Approach 2:
The patent merges the functionality of multiple conversion models into a single integrated conversion model. By combining the capabilities of what would have been separate models for different MRI formats into one unified model, the system reduces the overall construction load and training time while preserving the ability to generate diverse MRI representation formats.
3Measurement precision
If MRI imaging is performed to derive tissue eigenvalues, then accurate MRI images can be acquired, but imaging with the MRI apparatus is contraindicated in patients with implantable pacemakers and claustrophobia
Solution Approach 1:
The patent uses CT images as an intermediary to derive tissue eigenvalues that would normally require direct MRI imaging. By using the conversion model to translate CT images into tissue eigenvalue data, the system provides an alternative pathway for patients who cannot undergo MRI imaging due to contraindications such as pacemakers or claustrophobia, while still achieving accurate tissue characterization.
Solution Approach 2:
The conversion model creates a computational copy of MRI image data from CT images by deriving tissue eigenvalues. This copying approach allows the system to generate MRI-like information without requiring actual MRI imaging, providing accurate tissue eigenvalue data for patients who cannot undergo MRI while maintaining measurement precision through the machine learning-based conversion process.
Data Source
AI summary
An image acquisition unit acquires at least one target medical image having a representation format different from an MRI image. A tissue eigenvalue derivation unit has a derivation model constructed by machine learning using a plurality of teacher data to, in a case in which at least one medical image having the representation format different from the MRI image is input, output a tissue eigenvalue of MRI for the input medical image, and inputs the target medical image to the derivation model to derive the tissue eigenvalue for the target medical image.