Automatic alignment of magnetic resonance imaging (MRI) brain scan by anatomic landmarks

1. A method to automatically align magnetic resonance (MR) brain scans for diagnostic scan planning, comprising:

acquiring a three-dimensional (3D) localizer image of a patient;

selecting a two-dimensional (2D) coronal view and a 2D transverse view from the localizer image;

identifying a mid-sagittal plane (MSP) line in each of the coronal and transverse views and calculating a 3D MSP based on the MSP lines;

reconstructing the localizer image based on an equation for the 3D MSP to obtain an image of the MSP of the patient's brain;

identifying crista galli (CG) and tip of the occipital bone (TOB) in the image of the MSP of the patient's brain;

calculating a transformation matrix based on the MSP, CG and TOB in the image and using the transformation matrix to obtain a scan plan for the patient; and

outputting the scan plan for the patient,

wherein the step of identifying an MSP line in each of the coronal and transverse views comprises:

detecting a position of the patient's head by fitting an ellipse to each of the coronal and transverse views that maximizes an image gradient magnitude in a boundary region of the ellipse;

establishing a symmetrical axis which minimizes a difference between the coronal and transverse views on either side of the symmetrical axis;

characterizing low intensity pixels in the coronal and transverse views by using smoothed local differential operators;

finding pixels in the coronal and transverse views that form a space between hemispheres in each of the coronal and transverse views; and

running a linear regression with robust weights to define a line in each of the coronal and transverse views that best separates the two hemispheres, wherein the line is the MSP line.

2. The method of claim 1 , wherein the CG is identified by:

applying an active shape model (ASM) to segment the patient's skull and other anatomic structures around the patient's brain.

3. The method of claim 2 , wherein the TOB is identified by:

searching for low intensity voxels in the segmented skull, wherein the low intensity voxels identify a most convex part of the OB; and

tracing boundaries on both sides of the skull to an end of the OB, which is the TOB.

4. The method of claim 1 , wherein the scan plan for the patient is obtained by multiplying the transformation matrix by a standard slice package.

5. The method of claim 4 , wherein the standard slice package includes a set of image slices with geometry descriptions to be acquired from the patient in a standard coordinate system.

6. The method of claim 1 , further comprising:

executing the scan plan for the patient.

7. A system to automatically align magnetic resonance (MR) brain scans for diagnostic scan planning, comprising:

a memory device for storing a program;

a processor in communication with the memory device, the processor operative with the program to:

acquire a three-dimensional (3D) localizer image of a patient;

select a two-dimensional (2D) coronal view and a 2D transverse view from the localizer image;

identify a mid-sagittal plane (MSP) line in each of the coronal and transverse views and calculate a 3D MSP based on the MSP lines;

reconstruct the localizer image based on an equation for the 3D MSP to obtain an image of the MSP of the patient's brain;

identify crista galli (CG) and tip of the occipital bone (TOB) in the image of the MSP of the patient's brain;

calculate a transformation matrix based on the MSP, CO and TOB in the image and using the transformation matrix to obtain a scan plan for the patient; and

output the scan plan for the patient,

wherein the processor is further operative with the program when identifying an MSP line in each of the coronal and transverse views to:

detect a position of the patient's head by fitting an eclipse to each of the coronal and transverse views that maximizes an image gradient magnitude in a boundary region of the ellipse;

establish a symmetrical axis which minimizes a difference between the coronal and transverse views on either side of the symmetrical axis;

characterize low intensity pixels in the coronal and transverse views by using smoothed local differential

find pixels in the coronal and transverse views that form a space between hemispheres in each of the coronal and transverse views; and

run a linear regression with robust weights to define a line in each of the coronal and transverse views that best separates the two hemispheres, wherein the line is the MSP line.

8. The system of claim 7 , wherein the CG is identified by applying an active shape model (ASM) to segment the patient's skull and other anatomic structures around the patient's brain.

9. The system of claim 8 , wherein the TOB is identified by searching for low intensity voxels in the segmented skull, wherein the low intensity voxels identify a most convex part of the OB; and

tracing boundaries on both sides of the skull to an end of the OB, which is the TOB.

10. The system of claim 7 , wherein the scan plan for the patient is obtained by multiplying the transformation matrix by a standard slice package.

11. The system of claim 10 , wherein the standard slice package includes a set of image slices with geometry descriptions to be acquired from the patient in a standard coordinate system.

12. The system of claim 7 , wherein the processor is further operative with the program to:

execute the scan plan for the patient.

13. A non-transitory computer readable medium tangibly embodying a program of instructions executable by a processor to perform method steps to automatically align magnetic resonance (MR) brain scans for diagnostic scan planning, the method steps comprising:

acquiring a three-dimensional (3D) localizer image of a patient;

selecting a two-dimensional (2D) coronal view and a 2D transverse view from the localizer image;

identifying a mid-sagittal plane (MSP) line in each of the coronal and transverse views and calculating a 3D MSP based on the MSP lines;

reconstructing the localizer image based on an equation for the 3D MSP to obtain an image of the MSP of the patient's brain;

identifying crista galli (CG) and tip of the occipital bone (TOB) in the image of the MSP of the patient's brain;

calculating a transformation matrix based on the MSP, CG and TOB in the image and using the transformation matrix to obtain a scan plan for the patient; and

outputting the scan plan for the patient,

wherein the step of identifying an MSP line in each of the coronal and transverse views comprises:

detecting a position of the patient's head by fitting an ellipse to each of the coronal and transverse views that maximizes an image gradient magnitude in a boundary region of the ellipse;

establishing a symmetrical axis which minimizes a difference between the coronal and transverse views on either side of the symmetrical axis;

characterizing low intensity pixels in the coronal and transverse views by using smoothed local differential operators;

finding pixels in the coronal and transverse views that form a space between hemispheres in each of the coronal and transverse views; and

running a linear regression with robust weights to define a line in each of the coronal and transverse views that best separates the two hemispheres, wherein the line is the MSP line.

14. The computer readable medium of claim 13 , wherein the CG is identified by:

applying an active shape model (ASM) to segment the patient's skull and other anatomic structures around the patient's brain.

15. The computer readable medium of claim 14 , wherein the TOB is identified by:

searching for low intensity voxels in the segmented skull, wherein the low intensity voxels identify a most convex part of the OB; and

tracing boundaries on both sides of the skull to an end of the OB, which is the TOB.

16. The computer readable medium of claim 13 , wherein the scan plan for the patient is obtained by multiplying the transformation matrix by a standard slice package.

17. The computer readable medium of claim 16 , wherein the standard slice package includes a set of image slices with geometry descriptions to be acquired from the patient in a standard coordinate system.

18. The computer readable medium of claim 13 , wherein the method further comprises:

executing the scan plan for the patient.