Apparatus, methods and articles for four dimensional (4d) flow magnetic resonance imaging

Abstract

An MRI image processing and analysis system may identify instances of structure in MRI flow data, e.g., coherency, derive contours and/or clinical markers based on the identified structures. The system may be remotely located from one or more MRI acquisition systems, and perform: perform error detection and/or correction on MRI data sets (e.g., phase error correction, phase aliasing, signal unwrapping, and/or on other artifacts); segmentation; visualization of flow (e.g., velocity, arterial versus venous flow, shunts) superimposed on anatomical structure, quantification; verification; and/or generation of patient specific 4-D flow protocols. An asynchronous command and imaging pipeline allows remote image processing and analysis in a timely and secure manner even with complicated or large 4-D flow MRI data sets.

1 . (canceled) 
     
     
         2 . A method of operation for use with magnetic resonance imaging (MRI) based medical imaging systems, the method comprising:
 receiving a set of MRI data by at least one processor-based device, the set of MRI data comprising respective anatomical structure and blood flow information for each of a plurality of voxels; and   applying a first filter to isolate blood flow based on directional coherence to at least a portion of the received set of MRI data by the at least one processor-based device.   
     
     
         3 . The method of  claim 2  wherein applying a first filter to isolate blood flow based on directional coherence to at least a portion of the received set of MRI data includes:
 for each of a number of voxels, calculating a directional coherence for the respective voxel. 
 
     
     
         4 . The method of  claim 3  wherein calculating a directional coherence for a respective voxel includes:
 summing a set of weighted directional coherence scores between the respective voxel and a plurality of neighboring voxels which are neighbors of the respective voxel; and 
 dividing a result of the summation by a summation of all weights applied. 
 
     
     
         5 . The method of  claim 4 , further comprising:
 determining the weighted directional coherence scores between the respective voxel and the plurality of neighboring voxels.   
     
     
         6 . The method of  claim 5  wherein determining the weighted directional coherence scores between the respective voxel and the plurality of neighboring voxels includes:
 determining a dot product of normalized velocity vectors, 
 applying the trigonometric function ACOS to a result of the dot product to determine an angle difference; 
 scaling the angle difference between 0 and Pi to get a result between 0 and 1; and 
 multiplying the result of the scaling by a respective weight which is indicative of a distance between the respective voxel and a respective one of the neighboring voxels. 
 
     
     
         7 . The method of  claim 6 , further comprising:
 determining the respective weight.   
     
     
         8 . The method of  claim 7  wherein determining the respective weight includes:
 finding a minimum spacing for all three dimensions, and 
 dividing that minimum spacing by a distance between the voxels. 
 
     
     
         9 . The method of  claim 8  wherein the first filter is applied in one volume per time point. 
     
     
         10 . The method of  claim 8  wherein the first filter is applied in one volume averaged over all time points per time point. 
     
     
         11 . The method of  claim 8 , further comprising:
 applying a second filter to further remove random noise to the at least a portion of the received set of MRI data by the at least one processor-based device.   
     
     
         12 . A processor-based device, comprising:
 at least one nontransitory processor-readable storage medium storing at least one of instructions or data; and   at least one processor communicatively coupled to the at least one nontransitory processor-readable storage medium, in operation, the at least one processor:   receives a set of MRI data by at least one processor-based device, the set of MRI data comprising respective anatomical structure and blood flow information for each of a plurality of voxels; and   applies a first filter to isolate blood flow based on directional coherence to at least a portion of the received set of MRI data by the at least one processor-based device.   
     
     
         13 . The device of  claim 12 , wherein the at least one processor applies the first filter to isolate blood flow based on directional coherence to at least a portion of the received set of MRI data at least in part by:
 for each of a number of voxels, calculating a directional coherence for the respective voxel.   
     
     
         14 . The device of  claim 13 , wherein calculating a directional coherence for a respective voxel includes:
 summing a set of weighted directional coherence scores between the respective voxel and a plurality of neighboring voxels which are neighbors of the respective voxel; and   dividing a result of the summation by a summation of all weights applied.   
     
     
         15 . The device of  claim 14 , wherein in operation, the at least one processor further determines the weighted directional coherence scores between the respective voxel and the plurality of neighboring voxels. 
     
     
         16 . The device of  claim 15 , wherein the at least one processor determines the weighted directional coherence scores between the respective voxel and the plurality of neighboring voxels at least in part by:
 determining a dot product of normalized velocity vectors,   applying the trigonometric function ACOS to a result of the dot product to determine an angle difference;   scaling the angle difference between 0 and Pi to get a result between 0 and 1; and   multiplying the result of the scaling by a respective weight which is indicative of a distance between the respective voxel and a respective one of the neighboring voxels.   
     
     
         17 . The device of  claim 16 , wherein in operation, the at least one processor further determines the respective weight. 
     
     
         18 . The device of  claim 17 , wherein the at least one processor determines the respective weight at least in part by:
 finding a minimum spacing for all three dimensions, and   dividing that minimum spacing by a distance between the voxels.   
     
     
         19 . The device of  claim 18 , wherein the first filter is applied in one volume per time point. 
     
     
         20 . The device of  claim 18 , wherein the first filter is applied in one volume averaged over all time points per time point. 
     
     
         21 . The device of  claim 18 , wherein in operation, the at least one processor further applies a second filter to further remove random noise to the at least a portion of the received set of MRI data by the at least one processor-based device.