System and method for automated angiography

Abstract

A method for analyzing computed tomography angiography (CTA) data is provided. The method includes receiving, at a processor, three-dimensional (3D) CTA data. The method also includes automatically, via the processor, detecting objects of interest within the 3D CTA data. The method further includes generating, via the processor, a CTA image volume that only includes the objects of interest.

1 . A method for analyzing computed tomography angiography (CTA) data, comprising:
 receiving, at a processor, three-dimensional (3D) CTA data;   automatically, via the processor, detecting objects of interest within the 3D CTA data, wherein automatically detecting the objects of interest within the CTA data comprises applying, via the processor, a trained convolutional neural network to the 3D CTA data to segment the objects of interest from the 3D CTA data; and   generating, via the processor, a CTA image volume that only includes the objects of interest.   
     
     
         2 . The method of  claim 1 , wherein the objects of interest comprise arteries, veins, soft tissue, or bone. 
     
     
         3 . The method of  claim 1 , comprising:
 receiving, at the processor, four-dimensional (4D) CTA data;   generating, via the processor, non time-resolved CTA data from the 4D CTA data;   generating, via the processor, a first set of 4D images including veins only from the 4D CTA data; and   generating, via the processor, a second set of 4D images including arteries only from the 4D CTA data.   
     
     
         4 . The method of  claim 1 , comprising training, via the processor, a convolutional neural network utilizing the non time-resolved CTA data, the first set of 4D images, and the second set of 4D images to generate the trained convolutional neural network. 
     
     
         5 . The method of  claim 4 , comprising training, via the processor, a convolutional neural network utilizing the non time-resolved CTA data, the first set of 4D images, or the second set of 4D images to generate the trained convolutional neural network. 
     
     
         6 . The method of  claim 4 , wherein generating the non time-resolved CTA data comprises applying, via the processor, a weighted average to the 4D CTA data. 
     
     
         7 . The method of  claim 4 , wherein generating the first and second sets of 4D images comprises performing, via the processor, 4D segmentation on the 4D CTA data. 
     
     
         8 . The method of  claim 1 , comprising automatically, via the processor, reformatting the CTA image volume to generate one or more two-dimensional (2D) CTA images. 
     
     
         9 . The method of  claim 8 , wherein automatically reformatting the CTA image volume comprises applying, via the processor, the trained convolutional neural network to the CTA image volume to reformat the CTA image volume. 
     
     
         10 . The method of  claim 9 , wherein the trained convolutional neural network, via the processor, in reformatting the CTA image volume identifies an anatomical location of the objects of interest within the CTA image volume and determines a desired orientation of the one or more 2D CTA images. 
     
     
         11 . One or more non-transitory computer-readable media encoding one or more processor-executable routines, wherein the one or more routines, when executed by a processor, cause acts to be performed comprising:
 receiving three-dimensional (3D) CTA data;   automatically detecting objects of interest within the 3D CTA data, wherein automatically detecting the objects of interest within the CTA data comprises applying, via the processor, a trained convolutional neural network to the 3D CTA data to segment the objects of interest from the 3D CTA data; and   generating a CTA image volume that only includes the objects of interest.   
     
     
         12 . The one or more non-transitory computer-readable media of  claim 11 , wherein the objects of interest comprise arteries, veins, soft tissue, or bone. 
     
     
         13 . The one or more non-transitory computer-readable media of  claim 11 , wherein the one or more routines, when executed by the processor, cause acts to be performed comprising automatically reformatting the CTA image volume to generate one or more two-dimensional (2D) CTA images. 
     
     
         14 . The one or more non-transitory computer-readable media of  claim 13 , wherein automatically reformatting the CTA image volume comprises applying the trained convolutional neural network to the CTA image volume to reformat the CTA image volume. 
     
     
         15 . The one or more non-transitory computer-readable media of  claim 14 , wherein the trained convolutional neural network in reformatting the CTA image volume identifies an anatomical location of the objects of interest within the CTA image volume and determines a desired orientation of the one or more 2D CTA images. 
     
     
         16 . A processor-based system, comprising:
 a memory structure encoding one or more processor-executable routines, wherein the routines, when executed cause acts to be performed comprising:
 receiving three-dimensional (3D) CTA data; 
 automatically detecting objects of interest within the 3D CTA data, wherein automatically detecting the objects of interest within the CTA data comprises applying, via the processor, a trained convolutional neural network to the 3D CTA data to segment the objects of interest from the 3D CTA data; and 
 generating a CTA image volume that only includes the objects of interest; and 
   a processing component configured to access and execute the one or more routines encoded by the memory structure.   
     
     
         17 . The processor-based system of  claim 16 , wherein the objects of interest comprise arteries, veins, soft tissue, or bone. 
     
     
         18 . The processor-based system of  claim 16 , wherein the one or more routines, when executed by the processing component, cause acts to be performed comprising automatically reformatting the CTA image volume to generate one or more two-dimensional (2D) CTA images. 
     
     
         19 . The processor-based system of  claim 18 , wherein automatically reformatting the CTA image volume comprises applying the trained convolutional neural network to the CTA image volume to reformat the CTA image volume. 
     
     
         20 . The processor-based system of  claim 19 , wherein the trained convolutional neural network in reformatting the CTA image volume identifies an anatomical location of the objects of interest within the CTA image volume and determines a desired orientation of the one or more 2D CTA images.