Coil mixing error matrix and deep learning for prospective motion assessment

Patent metadata

Field	Value
Publication number	US-11630177-B2
Application number	US-202217719916-A
Country	US
Kind code	B2
Filing date	Apr 13, 2022
Priority date	Apr 14, 2021
Publication date	Apr 18, 2023
Grant date	Apr 18, 2023

Abstract

Systems and Methods that identify the effect of motion during a medical imaging procedure. A neural network is trained to translate motion induced deviations of a coil-mixing matrix relative to a reference acquisition into a motion score. This score can be used for the prospective detection of the most corrupted echo trains for removal or triggering a replacement by reacquisition.

First claim

What is claimed is:

1. A method for prospective or retrospective motion identification during an acquisition of magnetic resonance (MR) images of a patient by an imaging system, the method comprising:
acquiring a motion free reference;
calculating, based on the motion free reference, a first coil mixing matrix representing a linear combination of coils of the imaging system;
applying the first coil mixing matrix to the motion free reference to generate a linearly combined reference data;
acquiring MR data for the patient from the imaging system and applying the first coil mixing matrix to the MR data to generate linearly combined motion data;
determining a second coil mixing matrix for a respective subset of MR data based on the linearly combined motion data;
inputting the second coil mixing matrix into a neural network trained to output a motion assessment for the acquired MR data; and
providing the motion assessment generated by the neural network to an operator.

2. The method of claim 1 , wherein the motion free reference comprises scout data from a scout procedure acquired prior to acquiring the MR data.

3. The method of claim 1 , wherein the first coil mixing matrix is calculated using singular value decomposition.

4. The method of claim 1 , wherein additional information comprising at least one a data consistency error of a current echo train, an object size relative to an image matrix size, or a relative energy of the current echo train to a whole is further input into the neural network.

5. The method of claim 1 , wherein the linearly combined motion data and the second coil mixing matrix are calculated for each respective echo train of the MR data.

6. The method of claim 1 , wherein the motion assessment comprises values for respective degree of freedoms describing a three-dimensional motion state of the patient at a time of an acquisition of a portion of the MR data relative to an initial position.

7. The method of claim 1 , wherein the motion assessment comprises at least a motion score for a respective chunk of the MR data.

8. The method of claim 1 , further comprising:
ranking motion scores for each echo train; and
replacing data for echo train ranked above a certain level.

9. The method of claim 1 , further comprising:
ranking motion scores for each echo train; and
reacquiring data for echo train ranked above a certain level.

10. A method for prospective or retrospective motion identification during an acquisition of magnetic resonance (MR) images of a patient by an imaging system, the method comprising:
acquiring a motion free reference;
calculating, based on the motion free reference, a first coil mixing matrix representing a linear combination of coils of the imaging system;
acquiring MR data for the patient from the imaging system and applying the first coil mixing matrix to the MR data to generate linearly combined motion data;
calculating a second coil mixing matrix for a respective subset of MR data based on the linearly combined motion data;
calculating a difference coil mixing error matrix for a respective subset of MR data based on the difference of the first coil mixing matrix and the second coil mixing matrix;
inputting the difference coil mixing error matrix into a neural network trained to output a motion assessment for the acquired MR data; and
providing the motion assessment generated by the neural network to an operator.

11. The method of claim 10 , wherein the motion free reference comprises scout data from a scout procedure acquired prior to acquiring the MR data.

12. The method of claim 10 , wherein the first coil mixing matrix is calculated using singular value decomposition.

13. The method of claim 10 , wherein additional information comprising at least one a data consistency error of a current echo train, an object size relative to an image matrix size, or a relative energy of the current echo train to a whole is further input into the neural network.

14. The method of claim 10 , wherein the motion assessment comprises values for respective degree of freedoms describing a three-dimensional motion state of the patient at a time of an acquisition of a portion of the MR data relative to an initial position.

15. The method of claim 10 , wherein the motion assessment comprises at least a motion score for a respective chunk of the MR data.

16. The method of claim 10 , further comprising:
ranking motion scores for each echo train; and
replacing data for echo train ranked above a certain level.

17. The method of claim 10 , further comprising:
ranking motion scores for each echo train; and
reacquiring data for echo train ranked above a certain level.

18. A system for prospective or retrospective motion identification during an acquisition of magnetic resonance data of a patient, the system comprising:
a magnetic resonance imaging system configured to acquire motion free reference data and magnetic resonance data;
a processor configured to calculate, based on the motion free reference data, a scout coil mixing matrix and to calculate, based on the scout coil mixing matrix and the magnetic resonance data, a coil mixing error matrix, a coil mixing matrix, or the coil mixing error matrix and the coil mixing matrix;
a neural network configured to output a motion assessment when input the coil mixing error matrix, the coil mixing matrix, or the coil mixing error matrix and the coil mixing matrix; and
an output device configured to output the motion assessment from the neural network during an imaging procedure for the acquisition of the magnetic resonance data.

19. The system of claim 18 , wherein the motion assessment comprises values for respective degree of freedoms describing a three-dimensional motion state of the patient at a time of an acquisition of a portion of the MR data relative to an initial position.

20. The system of claim 18 , wherein the motion assessment comprises at least a motion score for a respective chunk of the MR data.

Assignees

• Siemens Healthcare Gmbh
• Massachusetts Gen Hospital

Inventors

• Splitthoff Daniel Nicolas
• Hossbach Julian
• Polak Daniel
• Cauley Stephen Farman
• Clifford Bryan
• Lo Wei-Ching

Classifications

• G01R33/5608

  Data processing and visualization specially adapted for MR, e.g. for feature analysis and pattern recognition on the basis of measured MR data, segmentation of measured MR data, edge contour detection on the basis of measured MR data, for enhancing measured MR data in terms of signal-to-noise ratio by means of noise filtering or apodization, for enhancing measured MR data in terms of resolution by means for deblurring, windowing, zero filling, or generation of gray-scaled images, colour-coded images or images displaying vectors instead of pixels (image data processing or generation, in general G06T) · CPC title

• G01R33/56509 Primary

  due to motion, displacement or flow, e.g. gradient moment nulling (G01R33/567 takes precedence) · CPC title

• G01R33/5676

  Gating or triggering based on an MR signal, e.g. involving one or more navigator echoes for motion monitoring and correction · CPC title

• G01R33/546 Primary

  Interface between the MR system and the user, e.g. for controlling the operation of the MR system or for the design of pulse sequences · CPC title