Systems and methods for medical imaging incorporating prior knowledge
Aspects of the present disclosure relate to systems and methods for medical imaging that incorporate prior knowledge. Some aspects relate to incorporating prior knowledge using a non-local means filter. Some aspects relate to incorporating prior knowledge for improved perfusion imaging, such as those incorporating arterial spin labeling.
1. A method comprising:
acquiring raw complex-valued perfusion image data directly from a magnetic resonance imaging (MRI) scanner in a first scan covering an area of interest of a subject, wherein the acquisition uses arterial spin labeling (ASL) in k-space;
acquiring anatomical image data corresponding to the area of interest, by a second scan different from the first scan; and
applying a cost function to the raw complex-valued perfusion image data acquired directly from the MRI scanner, as part of an iterative reconstruction to form one or more enhanced perfusion images, wherein the cost function comprises a non-local means (NLM) filter trained with filter weights based on the anatomical image data.
2. The method of claim 1 , wherein the one or more enhanced perfusion images has a higher signal-to-noise ratio than an image generated from the raw complex-valued perfusion image data by an inverse Fourier transform.
3. The method of claim 1 , wherein the cost function comprises a differencing matrix.
4. The method of claim 1 , wherein the raw complex-valued perfusion image data has a lower signal-to-noise ratio than the anatomical image data.
5. The method of claim 1 , wherein the anatomical image data is acquired concurrently with the acquisition of the raw complex-valued perfusion image data.
6. The method of claim 1 , wherein the area of interest comprises at least one area of the brain of the subject and the at least one physiological activity comprises cerebral blood flow (CBF) in the brain of the subject.
7. The method of claim 1 , further comprising training, from the anatomical image data, a model to produce the trained model.
8. A system comprising
a magnetic resonance imaging (MRI) device;
a memory device;
at least one processor coupled to the memory device and the MRI device and storing computer-readable instructions which, when executed by the at least one processor, cause the system to perform functions that include:
acquiring raw complex-valued perfusion image data directly from the magnetic resonance image device scanner in a first scan covering an area of interest of a subject, wherein the acquisition uses arterial spin labeling (ASL) in k-space;
acquiring anatomical image data corresponding to the area of interest, by a second scan different from the first scan; and
acquiring a cost function to the raw complex-valued perfusion image data acquired directly from the MRI scanner, as part of an iterative reconstruction to form one or more enhanced perfusion images, wherein the cost function comprises a non-local means (NLM) filter trained with filter weights based on the anatomical image data.
9. The system of claim 8 , wherein the one or more enhanced perfusion images has a higher signal-to-noise ratio than an image generated from the raw complex-valued perfusion image data by an inverse Fourier transform.
10. The system of claim 8 , wherein the cost function comprises a differencing matrix.
11. The system of claim 8 , wherein the raw complex-valued perfusion image data has a lower signal-to-noise ratio than the anatomical image data.
12. The system of claim 8 , wherein the anatomical image data is acquired concurrently with the acquisition of the raw complex-valued perfusion image data.
13. The system of claim 8 , wherein the area of interest comprises at least one area of the brain of the subject and the at least one physiological activity comprises cerebral blood flow (CBF) in the brain of the subject.
14. A computer-readable medium storing instructions that, when executed by one or more processors, cause a computer to perform functions that comprise:
acquiring raw complex-valued perfusion image data directly from a magnetic resonance imaging (MRI) scanner in a first scan covering an area of interest of a subject, wherein the acquisition uses arterial spin labeling (ASL) in k-space;
acquiring anatomical image data corresponding to the area of interest, by a second scan different from the first scan; and
applying a cost function to the raw complex-valued perfusion image data acquired directly from the MRI scanner, as part of an iterative reconstruction to form one or more enhanced perfusion images, wherein the cost function comprises a non-local means (NLM) filter trained with filter weights based on the anatomical image data.
15. The computer readable medium of claim 14 , wherein the one or more enhanced perfusion images has a higher signal-to-noise ratio than an image generated from the raw complex-valued perfusion image data by an inverse Fourier transform.
16. The computer readable medium of claim 14 , wherein the cost function comprises a differencing matrix.
17. The computer readable medium of claim 14 , wherein the raw complex-valued perfusion image data has a lower signal-to-noise ratio than the anatomical image data.
18. The computer readable medium of claim 14 , wherein the anatomical image data is acquired concurrently with the acquisition of the raw complex-valued perfusion image data.
19. The computer readable medium of claim 14 , wherein the area of interest comprises at least one area of the brain of the subject and the at least one physiological activity comprises cerebral blood flow (CBF) in the brain of the subject.