Neural Network Medical Image Reconstruction for Artifact Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current medical imaging methods often produce high-quality images overall but can have severe artifacts, noises, or poor contrasts in small regions, making it difficult for doctors to obtain required information, thus reducing the image's availability for medical use.
Innovation Solution
The use of pre-trained neural networks, specifically a first neural network and a second neural network, for image reconstruction and processing to improve image quality and align it with specific medical imaging tasks, such as grey matter and white matter segmentation, water and fat separation, or lung tissue segmentation, by performing reconstruction on down-sampling data and subsequent image processing to enhance image quality and relevance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional medical imaging methods are used to obtain high-quality images overall, then image resolution and signal-to-noise ratio are improved, but severe artifacts, noises, or poor contrasts appear in small regions reducing medical availability
Solution Approach 1:
The patent applies local quality by training the neural network to focus on specific regions of interest within the medical image. The system identifies small regions with artifacts or poor quality and applies targeted reconstruction and enhancement operations to those specific areas rather than uniformly processing the entire image, thereby improving local image quality where it matters most for medical diagnosis
Solution Approach 2:
The patent introduces a neural network as an intermediary between the raw medical imaging data and the final diagnostic image. This neural network mediator performs reconstruction and enhancement operations to eliminate artifacts and improve contrast in problematic regions, acting as a bridge that transforms low-quality regional data into high-quality diagnostic information
2Loss of time
If down-sampling data is used for image reconstruction, then scanning time and radiation dose are reduced, but image quality may deteriorate
Solution Approach 1:
The patent applies preliminary action by performing image reconstruction using down-sampled data first, then using the neural network to identify and correct quality issues in subsequent processing steps. The system prepares the image through initial reconstruction with reduced data, then applies targeted enhancement operations to achieve final high-quality results without requiring full-sampling data from the start
Solution Approach 2:
The patent applies parameter changes by dynamically adjusting image processing parameters based on the quality assessment of different regions. The neural network analyzes the reconstructed image from down-sampled data and modifies processing parameters locally to enhance specific regions that suffered from down-sampling, thereby recovering image quality despite using reduced data
Data Source
AI summary
Methods, devices, systems and apparatus for medical imaging, e.g., Magnetic Resonance (MR) imaging or Computed Tomography (CT) imaging, using neural networks are provided. In one aspect, an imaging method includes: determining a first neural network and a second neural network corresponding to a target imaging task, the first neural network including a first neural network parameter and a first neural network model, the second neural network including a second neural network parameter and a second neural network model, obtaining a reconstructed image by performing reconstruction for down-sampling data of a tissue under test using the first neural network, the target imaging task corresponding to the tissue under test, and obtaining an image output by a second neural network as a target image of the tissue under test by performing an image processing operation corresponding to the target imaging task for the reconstructed image using the second neural network.