US 12,112,483 B2
Systems and methods for anatomic structure segmentation in image analysis
Leo Grady, Darien, CT (US); Peter Kersten Petersen, Palo Alto, CA (US); Michiel Schaap, Leiden (NL); and David Lesage, Redwood City, CA (US)
Assigned to HeartFlow, Inc., Mountain View, CA (US)
Filed by HeartFlow, Inc., Redwood City, CA (US)
Filed on Feb. 21, 2023, as Appl. No. 18/171,915.
Application 18/171,915 is a continuation of application No. 17/203,964, filed on Mar. 17, 2021, granted, now 11,610,318.
Application 17/203,964 is a continuation of application No. 17/013,922, filed on Sep. 8, 2020, granted, now 10,984,535, issued on Apr. 20, 2021.
Application 17/013,922 is a continuation of application No. 15/975,197, filed on May 9, 2018, granted, now 10,803,592, issued on Oct. 13, 2020.
Claims priority of provisional application 62/503,838, filed on May 9, 2017.
Prior Publication US 2023/0196582 A1, Jun. 22, 2023
This patent is subject to a terminal disclaimer.
Int. Cl. G06T 7/174 (2017.01); G06T 7/00 (2017.01); G06T 7/12 (2017.01); G06T 7/149 (2017.01)
CPC G06T 7/174 (2017.01) [G06T 7/0012 (2013.01); G06T 7/12 (2017.01); G06T 7/149 (2017.01); G06T 2200/04 (2013.01); G06T 2207/10072 (2013.01); G06T 2207/10081 (2013.01); G06T 2207/10088 (2013.01); G06T 2207/10104 (2013.01); G06T 2207/10108 (2013.01); G06T 2207/10132 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/20112 (2013.01); G06T 2207/30004 (2013.01); G06T 2207/30101 (2013.01); G06V 2201/03 (2022.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method of machine-learning based anatomic structure segmentation in image analysis, the method comprising:
receiving image data of an anatomic structure of a patient;
obtaining an annotation of the anatomic structure;
determining, based on the annotation, one or more keypoints;
determining a respective distance between each keypoint and a boundary of the anatomic structure;
determining respective intensities along a respective ray associated with each keypoint;
training a patient-specific Convolutional Neural Network (CNN), based on the respective distance and the respective intensities of each keypoint, to predict sub-pixel or sub-voxel locations of the boundary of the anatomic structure; and
generating, using the trained patient-specific CNN a sub-pixel or sub-voxel boundary of the anatomic structure.