COMPUTER-AIDED DETECTION (CAD) OF INTRACRANIAL ANEURYSMS

• US 20120201446A1
• Filed: 07/15/2010
• Published: 08/09/2012
• Est. Priority Date: 07/15/2009
• Status: Active Grant

• Current Assignee: Mayo Foundation For Medical Education And Research (Mayo Clinic)
• Original Assignee: Mayo Foundation For Medical Education And Research (Mayo Clinic)
• Inventors: Erickson, Bradley J., Kallmes, David F., Yang, Xiaojiang, Blezek, Daniel J.
• Granted Patent: US 9,349,176 B2
• Time in Patent Office: Days
• US Class Current: 382/134
• CPC Class Codes: G06T 2207/10072 Tomographic images
• CPC Class Codes: G06T 2207/20101 Interactive definition of point of interest, landmark or seed
• CPC Class Codes: G06T 2207/30101 Blood vessel; Artery; Vein; Vascular
• CPC Class Codes: G06T 7/0012 Biomedical image inspection

Abstract

A computer-aided system identifies aneurysm suspects in 3D image datasets. The system takes the raw image dataset as input and assigns one or more points of interest (POIs) in the image data. The system determines one or more features for each POI and identifies one or more aneurysm suspects from among the assigned POIs based on the determined features.

Claims

1. A method, comprising:

assigning points of interest (POIs) on blood vessels within an image dataset of a subject generated by a medical imaging system based on defined characteristics that may be associated with an aneurysm;

calculating one or more features of each POI; and

identifying, from among the assigned POIs, one or more aneurysm suspects based on the calculated features.

2. The method of claim 1, wherein assigning POIs comprises at least one of assigning a POI when a blood vessel includes a main centerline and a branch centerline and a distance between the POI and main centerline satisfies a minimum threshold distance, assigning a POI when a blood vessel includes a bifurcation, assigning a POI when a blood vessel includes a centerline having a radius or curvature that satisfies a threshold value, and assigning a POI when a blood vessel includes a main centerline and a branch centerline that completes a loop with the main centerline.

3. The method of claim 1, wherein calculating one or more features of each POI comprises at least one of calculating a distance from the POI to a center line, calculating a radius of the vessel at a nearest point in the centerline or at a surface of the vessel, calculating a planeness of the vessel at the POI, calculating a cylinder surfaceness of the vessel, calculating Gaussian and mean curvatures of the vessel at the POI, and calculating a shape index (SI) at the POI.

4. The method of claim 1, further comprising receiving at least one user selected setting that allows a user to define or customize user adjustable settings for viewing and interacting with a list of the one or more aneurysm suspects.

5. The method of claim 4, wherein receiving at least one user selected setting comprises at least one of receiving a user-selected probability score, receiving a user-selected rank, receiving a user-selected aneurysm suspect classification, and receiving a user-selected region within an intracranial space.

6. The method of claim 1, further comprising assigning a probability score to the one or more aneurysm suspects.

7. The method of claim 6, further comprising ranking the one or more aneurysm suspects according to the assigned probability scores.

8. The method of claim 1, further comprising applying a machine learning algorithm to a training data set to determine the features to be calculated.

9. The method of claim 8, further comprising applying the machine learning algorithm to a diversified population training data set.

10. The method of claim 8, further comprising applying the machine learning algorithm to a population-specific training data set.

11. The method of claim 10, further comprising applying the machine learning algorithm to a population-specific training data set taken from a population of subjects with polycystic kidney disease.

12. The method of claim 10, further comprising applying the machine learning algorithm to at least one of a population-specific training data set taken from a population of subjects grouped by at least one of age, race, nationality, gender, or genetic profile.

13. The method of claim 10, wherein identifying one or more aneurysm suspects comprises identifying one or more aneurysm suspects based on a population-specific feature analysis.

17. The method of claim 1 wherein the image dataset is generated using one of magnetic resonance angiography or computed tomography.

18. The method of claim 1 wherein identifying one or more aneurysm suspects comprises identifying one or more intracranial aneurysm suspects.

19. The method of claim 1, wherein identifying one or more aneurysm suspects comprises applying one or more sieving rules to the assigned POIs to remove false positives, wherein the one or more sieving rules compare one or more of the calculated features for each of the assigned POIs with a corresponding one or more sieving threshold values.

20. The method of claim 1 further comprising displaying a list of the one or more aneurysm suspects and a corresponding probability score for each of the one or more aneurysm suspects.

14. A method, comprising:

assigning points of interest (POIs) on blood vessels within an image dataset of a subject based on defined characteristics that may be associated with an aneurysm;

for each assigned POI, calculating one or more features that may be indicative of the presence of an aneurysm;

generating a list of aneurysm suspects from among the assigned POIs based on whether the one or more features for each assigned POI satisfy one or more threshold values;

assigning a probability score to each aneurysm suspect;

ranking the aneurysm suspects based on the probability score assigned to each aneurysm suspect; and

displaying the list of aneurysm suspects and the probability score assigned to each aneurysm suspect.

15. The method of claim 14, wherein displaying the list of aneurysm suspects comprises displaying the list of aneurysm suspects in descending order of probability score.

16. The method of claim 14, wherein, for each of the assigned POIs, calculating one or more features comprises at least one of calculating a distance from the POI to a center line, calculating a radius of the vessel at a nearest point in the centerline or at a surface of the vessel at the POI, calculating a planeness of the vessel at the POI, calculating a cylinder surfaceness of the vessel at the POI, calculating Gaussian and mean curvatures of the vessel at the POI, and calculating a shape index (SI) at the POI.