US 2022/0179030 A1 · FAST DIFFUSION TENSOR MRI USING DEEP LEARNING

Abstract

Higher quality diffusion metrics and/or diffusion-weighted images are generated from lower quality input diffusion-weighted images using a suitably trained neural network (or other machine learning algorithm). High-fidelity scalar and orientational diffusion metrics can be extracted using a theoretical minimum of a single non-diffusion-weighted image and six diffusion-weighted images, achieved with data-driven supervised deep learning. As an example, a deep convolutional neural network (“CNN”) is used to map the input non-diffusion-weighted image and diffusion-weighted images sampled along six optimized diffusion-encoding directions to the residuals between the input and output high-quality non-diffusion-weighted image and diffusion-weighted images, which enables residual learning to boost the performance of CNN and full tensor fitting to generate any scalar and orientational diffusion metrics.

Claims23

Dependent claims

A method for generating a diffusion metric from magnetic resonance image data acquired from a subject, the method comprising:

(a) accessing magnetic resonance image data with a computer system, wherein the magnetic resonance image data comprise:
a non-diffusion-weighted image acquired from a subject;
at least six diffusion-weighted images acquired from the subject, each diffusion-weighted image corresponding to one of six different diffusion-encoding directions;
(b) accessing a neural network with the computer system, wherein the neural network has been trained on training data to learn a mapping from lower quality diffusion-weighted images to ground-truth diffusion-weighted images;
(c) inputting at least the at least six diffusion-weighted images contained in the magnetic resonance image data to the neural network using the computer system, generating output as residual image data that represents image differences relative to the ground-truth diffusion-weighted images;
(d) generating updated diffusion-weighted images by combining the at least six diffusion-weighted images and the residual image data with the computer system;
and
(e) generating a diffusion metric from the updated diffusion-weighted images using the computer system.

5 more elements

A method for generating at least one of diffusion metrics or diffusion-weighted images of a subject, the method comprising:

(a) accessing magnetic resonance image data with a computer system, wherein the magnetic resonance image data comprise:
a non-diffusion-weighted image acquired from a subject;
at least six diffusion-weighted images acquired from the subject, each diffusion-weighted image corresponding to one of six different diffusion-encoding directions;
(b) computing initial diffusion tensor elements from the magnetic resonance image data by using the computer system to fit the magnetic resonance image data to a diffusion tensor model;
(c) accessing a neural network with the computer system, wherein the neural network has been trained on training data to learn a mapping from lower quality diffusion tensor elements to ground-truth diffusion tensor elements;
(d) inputting the initial diffusion tensor elements to the neural network using the computer system, generating output as residual diffusion tensor element data that represents diffusion tensor element differences relative to the ground-truth diffusion tensor elements;
(e) generating updated diffusion tensor elements by combining the initial diffusion tensor elements and the residual diffusion tensor element data with the computer system;
and
(f) generating at least one of a diffusion metric or a diffusion-weighted image from the updated diffusion tensor elements using the computer system.

6 more elements

Dependencies are inferred from the claim text.

Fast diffusion tensor mri using deep learning

Life of this patent

Fast diffusion tensor mri using deep learning

Life of this patent

Abstract

Claims23