US 12,236,606 B2
System and method of evaluating neural networks to segment medical images
Joseph Ross Mitchell, Tampa, FL (US)
Assigned to H. LEE MOFFITT CANCER CENTER AND RESEARCH INSTITUTE, INC., Tampa, FL (US)
Appl. No. 17/613,642
Filed by H. LEE MOFFITT CANCER CENTER AND RESEARCH INSTITUTE, INC., Tampa, FL (US)
PCT Filed May 26, 2020, PCT No. PCT/US2020/034537
§ 371(c)(1), (2) Date Nov. 23, 2021,
PCT Pub. No. WO2020/237242, PCT Pub. Date Nov. 26, 2020.
Claims priority of provisional application 62/851,975, filed on May 23, 2019.
Prior Publication US 2022/0237785 A1, Jul. 28, 2022
Int. Cl. G06T 7/11 (2017.01); G06T 7/00 (2017.01); G06V 10/26 (2022.01); G06V 10/82 (2022.01); G16H 30/20 (2018.01)
CPC G06T 7/11 (2017.01) [G06T 7/0012 (2013.01); G06V 10/26 (2022.01); G06V 10/82 (2022.01); G16H 30/20 (2018.01); G06T 2207/10088 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/30016 (2013.01); G06T 2207/30096 (2013.01); G06V 2201/03 (2022.01)] 19 Claims
OG exemplary drawing
 
1. A computer implemented method of determining accuracy of a neural network in producing computerized segmentations within magnetic resonant (MR) images, comprising:
saving computer segmented images on a first computer connected to a network;
communicating with test computers connected to the network, wherein the test computers display the computer segmented images alongside manually segmented test images for scoring; and
receiving, at the first computer, scores for the accuracy of the manually segmented test images and the computer segmented images from the test computer; and
displaying at least one manually segmented test image and at least one computer segmented image on at least one of the test computers in a series, in random order, and in a blind identification process.