Development of an Artificial Intelligence Analysis Method for Magnetic Particle Imaging Using a Machine Learning Algorithm
Development of an artificial intelligence analysis method for magnetic particle imaging using a machine learning algorithm.
A magnetic particle imaging (MPI) device uses a changing magnetic field to cause superparamagnetic iron oxide (SPIO) nanoparticles to produce a signal upon exposure. The MPI device then detects these signals and produces a high contrast image. These images are useful in tracking SPIO throughout the body. This is useful in tracking transplanted cells modified with SPIO or with cells capable of SPIO uptake such as cancer cells. Image processing is done by selecting a region of interest and estimating the amount of SPIO nanoparticles in the area from pixel brightness in the images. The subjective nature of this process means that results can vary by user.
Description of Technology
This technology incorporates software to complete the processing of MPI images. The software uses machine learning to identify the regions of interest, then estimate the amount of SPIO nanoparticles in the area. The technology uses the intensity of each pixel as a data point. This data point is then compared to reference images to differentiate useful signals from noise. The addition of a convolutional neural network (CNN) improves the estimation and provides more advanced predictive capabilities.
- Reduced User Error: With an automated process, results become more consistent. This allows multiple users to compare findings without user input.
- Adaptable: This technology can be used with various other methods of image processing.
- MPI image processing
- PET/MRI scans
LICENSING RIGHTS AVAILABLE
Full licensing rights available
Developer: Dr. Ping Wang
Tech ID: TEC2021-0022
For Information, Contact:
Michigan State University