Systems and methods for improving magnetic resonance imaging using deep learning

Abstract

A computer-implemented method is provided for improving image quality with shortened acquisition time. The method comprises: determining an accelerated image acquisition scheme for imaging a subject using a medical imaging apparatus; acquiring a medical image of the subject according to the accelerated image acquisition scheme using the medical imaging apparatus; applying a deep network model to the medical image to improve the quality of the medical image; and outputting an improved quality image of the subject, for analysis by a physician.

What is claimed is: 
     
         1 . A computer-implemented method for improving image quality with shortened acquisition time, the method comprising:
 (a) determining an accelerated image acquisition scheme for imaging a subject using a medical imaging apparatus;   (b) acquiring, using the medical imaging apparatus, a medical image of the subject according to the accelerated image acquisition scheme;   (c) applying a deep network model to the medical image to improve the quality of the medical image; and   (d) outputting an improved quality image of the subject for analysis by a physician.   
     
     
         2 . The computer-implemented method of  claim 1 , wherein the medical image includes a magnetic resonance image. 
     
     
         3 . The computer-implemented method of  claim 1 , wherein determining the accelerated image acquisition scheme comprises:
 (i) receiving a target acceleration factor or target acquisition speed via a graphical user interface, and   (ii) selecting the accelerated image acquisition scheme from a plurality of accelerated image acquisition schemes based on the target acceleration factor or the target acquisition speed.   
     
     
         4 . The computer-implemented method of  claim 3 , wherein selecting the accelerated image acquisition scheme comprises applying the plurality of accelerated image acquisition schemes to a portion of the medical image for simulation. 
     
     
         5 . The computer-implemented method of  claim 1 , wherein the accelerated image acquisition scheme is determined based on user input and real-time simulated output images. 
     
     
         6 . The computer-implemented method of  claim 1 , wherein the accelerated image acquisition scheme comprises one or more parameters related to an undersampled k-space, an undersampling pattern, and a reduced number of repetitions. 
     
     
         7 . The computer-implemented method of  claim 6 , wherein the undersampling pattern is selected from a group consisting of a uniform undersampling pattern, a random undersampling pattern, and a variable undersampling pattern. 
     
     
         8 . The computer-implemented method of  claim 1 , wherein the medical image comprises undersampled k-space image or image acquired using reduced number of repetitions. 
     
     
         9 . The computer-implemented method of  claim 1 , wherein the deep learning model is trained with adaptively optimized metrics based on user input and real-time simulated output images. 
     
     
         10 . The computer-implemented method of  claim 1 , wherein the deep learning model is trained using training datasets comprising at least a low quality image and a high quality image. 
     
     
         11 . The computer-implemented method of  claim 10 , wherein the low quality image is generated by applying one or more filters or adding synthetic noise to the high quality image to create noise or undersampling artifacts. 
     
     
         12 . The computer-implemented method of  claim 1 , wherein the deep learning model is trained using image patches that comprise a portion of at least a low quality image and a high quality image. 
     
     
         13 . The computer-implemented method of  claim 12 , wherein the image patches are selected based on one or more metrics quantifying an image similarity. 
     
     
         14 . The computer-implemented method of  claim 1 , wherein the deep learning model is a deep residual learning model. 
     
     
         15 . The computer-implemented method of  claim 1 , wherein the deep learning model is trained by adaptively tuning one or more model parameters to approximate a reference image. 
     
     
         16 . The computer-implemented method of  claim 1 , wherein the improved quality image of the subject has greater SNR, higher resolution, or less aliasing compared with the medical image acquired using the medical imaging apparatus. 
     
     
         17 . A non-transitory computer-readable storage medium including instructions that, when executed by one or more processors, cause the one or more processors to perform operations comprising:
 (a) determining an accelerated image acquisition scheme for imaging a subject using a medical imaging apparatus;   (b) acquiring, using the medical imaging apparatus, a medical image of the subject according to the accelerated image acquisition scheme;   (c) applying a deep network model to the medical image to improve the quality of the medical image; and   (d) outputting an improved quality image of the subject for analysis by a physician.   
     
     
         18 . The non-transitory computer-readable storage medium of  claim 17 , wherein the medical image includes a magnetic resonance image. 
     
     
         19 . The non-transitory computer-readable storage medium of  claim 17 , wherein the accelerated image acquisition scheme is determined based on user input and real-time simulated output images. 
     
     
         20 . The non-transitory computer-readable storage medium of  claim 17 , wherein the accelerated image acquisition scheme comprises one or more parameters related to an undersampled k-space, an undersampling pattern, and a reduced number of repetitions.