US 12,260,558 B2
Deep multi-magnification networks for multi-class image segmentation
Thomas Fuchs, New York, NY (US); and David Joon Ho, New York, NY (US)
Assigned to Memorial Sloan-Kettering Cancer Center, New York, NY (US)
Filed by MEMORIAL SLOAN KETTERING CANCER CENTER, New York, NY (US)
Filed on Oct. 3, 2022, as Appl. No. 17/959,249.
Application 17/959,249 is a continuation of application No. 17/062,340, filed on Oct. 2, 2020, granted, now 11,501,434.
Claims priority of provisional application 62/909,401, filed on Oct. 2, 2019.
Prior Publication US 2023/0100881 A1, Mar. 30, 2023
This patent is subject to a terminal disclaimer.
Int. Cl. G06T 7/00 (2017.01); G06F 18/214 (2023.01); G06K 9/62 (2022.01); G06N 3/04 (2023.01); G06N 3/045 (2023.01); G06T 7/11 (2017.01); G06T 19/00 (2011.01); G06V 10/25 (2022.01)
CPC G06T 7/11 (2017.01) [G06F 18/214 (2023.01); G06N 3/045 (2023.01); G06T 7/0012 (2013.01); G06T 19/00 (2013.01); G06V 10/25 (2022.01); G06T 2207/30068 (2013.01); G06T 2219/004 (2013.01)] 20 Claims
OG exemplary drawing
 
1. A method, comprising:
identifying, by a computing system, for a first tile of a biomedical image, a first patch generated from the first tile at a first magnification factor and a second patch generated from the first tile at a second magnification factor, the first tile including at least a portion of a region of interest (ROI) in the biomedical image;
applying, by the computing system, the first patch and the second patch to a machine learning (ML) model, the ML model comprising:
a first network to generate a first plurality of feature maps using the first patch,
a second network to generate a second feature map using (i) the second patch and (ii) the first plurality of feature maps transferred from the first network in accordance with a shift between the first network and the second network, and
a terminal block to generate a second tile using the second feature map, the second tile identifying the portion of the ROI in the first tile; and
storing, by the computing system, in one or more data structures, an association between the biomedical image and the second tile.