System and method for improving image quality of susceptibility weighted imaging utilizing deep learning

1 . A computer-implemented method for improving image quality of susceptibility weighted imaging (SWI), comprising:

acquiring, via a processor, a magnetic resonance imaging (MRI) complex signal having a plurality of complex echoes of a region of interest via an MRI scanner from a coil during an SWI sequence;

phase filtering, via the processor, each complex echo of the plurality of complex echoes;

generating, via the processor, a respective phase image and a respective magnitude image from each phase filtered complex echo, wherein the respective phase image is generated only from imaginary channel data and the respective magnitude image is generated only from real channel data from each phase filtered complex echo;

combining, via the processor, separately the respective magnitude images of the plurality of complex echoes with each other to generate a combined magnitude image and the respective phase images of the plurality of complex echoes with each other to generate a combined phase image;

generating, via the processor, a complex image from both the combined magnitude image and the combined phase image; and

utilizing, via the processor, a deep learning-based denoising network to denoise the complex image to generate a denoised complex image.

2 . The computer-implemented method of claim 1 , further comprising applying, via the processor, super-resolution to the denoised complex image to generate a higher resolution denoised complex image that has had any blur removed.

3 . The computer-implemented method of claim 2 , further comprising generating, via the processor, a phase image from the higher resolution denoised complex image utilizing phase filtering.

4 . The computer-implemented method of claim 3 , further comprising generating, via the processor, a phase mask based solely on the phase image.

5 . The computer-implemented method of claim 4 , further comprising generating, via the processor, a SWI processed image of the region of interest based on the phase mask and the respective magnitude image for each complex echo of the plurality of complex echoes.

6 . The computer-implemented method of claim 3 , further comprising generating, via the processor, a magnitude image from the higher resolution denoised complex image, wherein the magnitude image is partially denoised.

7 . The computer-implemented method of claim 6 , further comprising generating, via the processor, a phase mask based on phase image and the magnitude image with a percent residue of the combined magnitude image added.

8 . The computer-implemented method of claim 7 , further comprising generating, via the processor, a SWI processed image of the region of interest based on the phase mask and the magnitude image with the percent residue of the combined magnitude image added.

9 . The computer-implemented method of claim 1 , wherein combining the respective magnitude images of the plurality of complex echoes with each other comprises combining the respective magnitude images with each other utilizing root mean square to generate the combined magnitude image.

10 . The computer-implemented method of claim 1 , wherein combining the respective phase images of the plurality of complex echoes with each other comprises combining the respective phase images with each other utilizing averaging to generate the combined phase image.

11 . A system for improving image quality of susceptibility weighted imaging (SWI), comprising:

a memory encoding processor-executable routines; and

a processor configured to access the memory and to execute the processor-executable routines, wherein the routines, when executed by the processor, cause the processor to:

acquire a magnetic resonance imaging (MRI) complex signal having a plurality of complex echoes of a region of interest via an MRI scanner from a coil during an SWI sequence;

phase filter each complex echo of the plurality of complex echoes;

generate a respective phase image and a respective magnitude image from each phase filtered complex echo, wherein the respective phase image is generated only from imaginary channel data and the respective magnitude image is generated only from real channel data from each phase filtered complex echo;

combine separately the respective magnitude images of the plurality of complex echoes with each other to generate a combined magnitude image and the respective phase images of the plurality of complex echoes with each other to generate a combined phase image;

generate a complex image from both the combined magnitude image and the combined phase image; and

utilize a deep learning-based denoising network to denoise the complex image to generate a denoised complex image.

12 . The system of claim 11 , wherein the routines, when executed by the processor, cause the processor to apply super-resolution to the denoised complex image to generate a higher resolution denoised complex image that has had any blur removed.

13 . The system of claim 12 , wherein the routines, when executed by the processor, cause the processor to generate a phase image from the higher resolution denoised complex image.

14 . The system of claim 13 , wherein the routines, when executed by the processor, cause the processor to generate a phase mask based solely on the phase image.

15 . The system of claim 14 , wherein the routines, when executed by the processor, cause the processor to generate a SWI processed image of the region of interest based on the phase mask and the respective magnitude image for each complex echo of the plurality of complex echoes.

16 . The system of claim 13 , wherein the routines, when executed by the processor, cause the processor to generate a magnitude image from the higher resolution denoised complex image, wherein the magnitude image is partially denoised.

17 . The system of claim 16 , wherein the routines, when executed by the processor, cause the processor to generate a phase mask based on phase image and the magnitude image with a percent residue of the combined magnitude image added.

18 . The system of claim 17 , wherein the routines, when executed by the processor, cause the processor to generate a SWI processed image of the region of interest based on the phase mask and the magnitude image with the percent residue of the combined magnitude image added.

19 . The system of claim 11 , wherein combining the respective magnitude images of the plurality of complex echoes with each other comprises combining the respective magnitude images with each other utilizing root mean square to generate the combined magnitude image, and wherein combining the respective phase images of the plurality of complex echoes with each other comprises combining the respective phase images with each other utilizing averaging to generate the combined phase image.

20 . A non-transitory computer-readable medium, the computer-readable medium comprising processor-executable code that when executed by a processor, causes the processor to:

acquire a magnetic resonance imaging (MRI) complex signal having a plurality of echoes of a region of interest via an MRI scanner from a coil during a multi-echo acquisition imaging sequence;

phase filter each complex echo of the plurality of complex echoes;

generate a respective phase image and a respective magnitude image from each phase filtered complex echo, wherein the respective phase image is generated only from imaginary channel data and the respective magnitude image is generated only from real channel data from each phase filtered complex echo;

combine separately the respective magnitude images of the plurality of complex echoes with each other to generate a combined magnitude image and the respective phase images of the plurality of complex echoes with each other to generate a combined phase image;

generate a complex image from both the combined magnitude image and the combined phase image;

utilize a deep learning-based denoising network to denoise the complex image to generate a denoised complex image; and

apply super-resolution to the denoised complex image to generate a higher resolution denoised complex image that has had any blur removed.