CN-122021970-A - Image reconstruction for magnetic resonance imaging

Abstract

Systems and methods for training a machine learning model to generate de-noised and de-aliased image data are provided. Techniques for training a Machine Learning (ML) model to generate de-noised and de-aliased imaging data are provided. The method includes (1) training a first ML model using a first training data set comprising first image data to obtain a second ML model, and (2) training (a) the second ML model or (b) a third ML model using the second training data set to obtain a fourth ML model. The second training data set includes (i) first image data and (ii) training image data obtained by applying at least one of the second ML model and the third ML model to the second image data. The denoising and dealiasing ML model may be or be derived from a fourth ML model.

Inventors

• J. Schlumber

Assignees

• 海珀菲纳股份有限公司

Dates

Publication Date

20260512

Application Date

20240426

Priority Date

20230428

Claims (15)

1. 1. A method of generating a trained machine learning model, i.e., a trained ML model, for image reconstruction, wherein generating the trained ML model comprises: (1) Updating the first ML model using a first training dataset comprising first image data to obtain a second ML model, and (2) Updating the second ML model using a second training data set to obtain the trained ML model, wherein the second training data set comprises: (i) The first image data, and (Ii) Training image data obtained by applying the second ML model to second image data.
2. 2. The method of claim 1, wherein at least one of the first training data set and the second training data set comprises simulated imaging data.
3. 3. The method of claim 2, wherein the simulated imaging data is based on a simulated image of arbitrary contrast.
4. 4. The method of claim 1, wherein the second image data comprises non-independent and non-uniformly distributed noise.
5. 5. The method of claim 1, further comprising applying the trained ML model to a patient image to obtain a reconstructed patient image.
6. 6. The method of claim 5, wherein the patient image is acquired using at least one of a low field magnetic resonance imaging system, i.e., a low field MR imaging system, and a point of care MR imaging system, i.e., a POC MR imaging system.
7. 7. The method of claim 1, wherein the first image data and the second image data belong to separate domains.
8. 8. The method of claim 1, further comprising enhancing the training image data based on an enhancement process prior to step (2).
9. 9. The method of claim 1, wherein the trained second ML model comprises a plurality of convolutional neural network layers, or CNN layers.
10. 10. The method of claim 1, further comprising generating the first training data set by applying raw imaging data to an image reconstruction pipeline.
11. 11. The method of claim 10, further comprising adding simulated image damage to the raw imaging data.
12. 12. A method comprising acquiring a patient image using an imaging system, and applying a trained machine learning model, i.e., a trained ML model, to the patient image to obtain a reconstructed patient image, the trained ML model having been generated by the method of claim 1.
13. 13. The method of claim 12, wherein the patient image is acquired using at least one of a low-field MR imaging system and a POC MR imaging system.
14. 14. A non-transitory computer-readable storage medium comprising instructions that, when executed by one or more processors of a computing system, cause the computing system to: acquiring patient image data using an imaging system, and Obtaining a reconstructed patient image based on the patient image data, wherein obtaining the reconstructed patient image comprises applying a trained machine learning model, i.e., a trained ML model, to the patient image data, the trained ML model having been generated by the method of claim 1.
15. 15. A system comprising an imaging system configured to generate imaging data, and one or more processors configured to cause the imaging system to generate a patient image and apply a trained ML model to the patient image to generate a reconstructed patient image, the trained ML model having been generated by the method of claim 1.

Description

Image reconstruction for magnetic resonance imaging (The present application is a divisional application of application No. 202480041534.0, titled "image reconstruction for magnetic resonance imaging", with application No. 2024, 4, 26 and application No. 202480041534.0.) Technical Field The present application relates generally to reducing noise in medical imaging through machine learning for applications such as Magnetic Resonance Imaging (MRI) where, for example, a low-field MRI system may be used. Example machine learning methods include two-step semi-supervised error correction and/or artifact correction (e.g., denoising and/or dealiasing) methods to reduce noise and/or reduce artifacts in low field diffusion Magnetic Resonance (MR) images. Background Magnetic Resonance Imaging (MRI) systems may be utilized to generate images of the interior of the human body. An MRI system may be used to detect Magnetic Resonance (MR) signals in response to an applied electromagnetic field. The MR signals generated by the MRI system may be processed to generate images, which may enable viewing of internal anatomy for diagnostic or research purposes. It can be said to be challenging to accurately reconstruct MR signals captured by an MRI system while removing sufficient noise to enable adequate viewing of anatomical structures. Disclosure of Invention At least one aspect of the present disclosure relates to a method of generating a trained Machine Learning (ML) model for image reconstruction. Generating the trained ML model may include (1) updating the first ML model using a first training data set to obtain a second ML model, the first training data set including first image data, and (2) updating the second ML model using a second training data set to obtain the trained ML model. The second training data set may include the first image data and training image data. Training image data may be obtained by applying the second ML model to the second image data. In various embodiments, at least one of the first training data set and the second training data set comprises simulated imaging data. In various embodiments, the simulated imaging data is based on a simulated image of arbitrary contrast. In various embodiments, the second image data includes noise that is non-independent and non-uniformly distributed. In various embodiments, the method includes applying the trained ML model to the patient image to obtain a reconstructed patient image. In various embodiments, the patient image is acquired using at least one of a low-field Magnetic Resonance (MR) imaging system and a point-of-care (POC) MR imaging system. In various embodiments, the first image data and the second image data belong to separate domains. In various embodiments, the method includes enhancing the training image data based on an enhancement process prior to step (2). In various embodiments, the trained second ML model includes a plurality of Convolutional Neural Network (CNN) layers. In various embodiments, the method further includes generating a first training data set by applying the raw imaging data to an image reconstruction pipeline. In various embodiments, the method includes adding simulated image impairments to raw imaging data. In another aspect, the present disclosure is directed to a method comprising acquiring a patient image using an imaging system and applying a trained Machine Learning (ML) model to the patient image to obtain a reconstructed patient image. The trained ML model may be generated by any of the methods described above. In various embodiments, the patient image is acquired using at least one of a low field MR imaging system and a POC MR imaging system. In yet another aspect, the present disclosure is directed to a non-transitory computer-readable storage medium comprising instructions that, when executed by one or more processors of a computing system, cause the computing system to acquire patient image data using an imaging system, and obtain a reconstructed patient image based on the patient image data, wherein obtaining the reconstructed patient image comprises applying a trained Machine Learning (ML) model to the patient image data. The trained ML model has been generated by any of the methods described above. In yet another aspect, the present disclosure is directed to a system comprising an imaging system configured to generate imaging data and one or more processors configured to cause the imaging system to generate a patient image and apply a trained ML model to the patient image to generate a reconstructed patient image. The trained ML model may be generated by any of the methods described above. Training the ML model may be performed on one computing system or using multiple computing systems in communication with each other (e.g., via one or more wired or wireless networks). Training of the ML model may be performed on one or more computing systems, and the trained ML model is then sent to one or more other systems