3D MRI Contrast Enhancement With Reduced Contrast Agent Dose
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing MRI techniques face challenges in reducing contrast agent administration without perceptible loss of information, particularly due to noise transfer and false-positive contrast enhancement signals in AI-based approaches.
Innovation Solution
A method and device using image registration, subtraction, and a trained neural network for contrast enhancement and artifact reduction in MRI images, employing three-dimensional rigid body registration and nonlinear transformations to amplify contrast signals while minimizing noise and artifacts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If AI-based approaches are used to predict T1-weighted images with full contrast agent administration, then image prediction accuracy is improved, but false-positive contrast enhancement signals and noise transfer occur
Solution Approach 1:
The patent extracts and removes noise components from the image prediction process by using denoising autoencoders and conditional random fields. The system separates true contrast enhancement signals from noise artifacts, allowing accurate prediction without false-positive signals. The denoising autoencoder specifically extracts and eliminates noise while preserving genuine contrast agent accumulation patterns.
Solution Approach 2:
The patent introduces intermediary models including denoising autoencoders and conditional random fields as mediators between the input images and final predictions. These intermediary components process and refine the image data, filtering out noise and artifacts while maintaining the essential contrast enhancement information, thereby preventing direct noise transfer to the output.
2Loss of time
If processing is carried out at slice image level to compensate for low processing speed, then computational time is reduced, but artifacts occur and quality results are limited to slice level
Solution Approach 1:
The patent transitions from processing individual slice images to processing three-dimensional volumetric data. By operating in the spatial volume dimension rather than just the slice dimension, the system achieves both computational efficiency and superior image quality. The volumetric processing allows the neural network to capture spatial relationships across multiple slices simultaneously, reducing artifacts while maintaining processing speed.
Solution Approach 2:
The patent merges multiple slice images into a unified three-dimensional volume for processing. This combination allows the system to handle the entire imaging dataset as a single processing unit, improving computational efficiency while preserving the quality information from all slices. The volumetric approach consolidates processing operations that would otherwise be performed separately on each slice.
3Object-affected harmful factors
If contrast agents are administered at reduced doses to reduce harm and cost, then patient safety and healthcare costs are improved, but signal intensity and diagnostic quality deteriorate
Solution Approach 1:
The patent uses neural network models to create synthetic copies of high-contrast images from low-contrast input images. The denoising autoencoder and conditional random field models learn to generate realistic T1-weighted images with full contrast agent administration characteristics from reduced-dose input images. This copying process produces high-quality output images that mimic the appearance of full-dose scans without requiring actual full-dose administration.
Solution Approach 2:
The patent applies parameter transformations through neural network models that map low-contrast image parameters to high-contrast image parameters. The denoising autoencoder and conditional random fields perform non-linear parameter transformations, adjusting signal intensity, contrast characteristics, and noise properties to produce images that resemble full-dose scans. This parameter change approach enables signal enhancement without additional contrast agent administration.
Data Source
AI summary
The invention relates to a method and a device (1) for providing MRI images relating to at least one part of a patient's body with reduced contrast medium administration. The device (1) has input means for obtaining MRI images relating to at least one part of a patient's body without contrast medium administration, ⋅input means for obtaining MRI images with respect to at least the part of the patient's body with contrast agent administration, wherein the contrast agent administration allowed for the body part is reduced by at least 50% compared to a conventional MRI image, ⋅means for image registration of the MRI images without contrast medium administration and of the MRI images with contrast medium administration, wherein the means for image registration operate in three-dimensional space, ⋅means for producing at least one subtraction image from the comparison of image-registered MRI images without contrast medium administration and image-registered MRI images with contrast medium administration, ⋅means for contrast enhancement based on the subtraction image and the image-registered MRI images without contrast agent administration or with reduced contrast agent administration, wherein the means for contrast enhancement based on a trained neural network provide both enhancement and artefact reduction by means of a non-linear transformation, wherein the means for contrast enhancement are arranged to produce a de-noised contrast enhanced difference image, ⋅wherein three-dimensional data are used throughout, and the image registration means provide rigid body registration in three dimensions. The invention also relates to a computer program product and the use of the method and devices according to the invention.