US 12,260,607 B2
Augmented digital microscopy for lesion analysis
George L. Mutter, Boston, MA (US); Peter Hufnagl, Berlin (DE); Sebastian Lohmann, Berlin (DE); and David James Papke, Jr., Boston, MA (US)
Assigned to THE BRIGHAM AND WOMEN'S HOSPITAL, INC., Boston, MA (US); and CHARITE UNIVERSITATSMEDIZIN BERLIN, Berlin (DE)
Appl. No. 17/414,010
Filed by THE BRIGHAM AND WOMEN'S HOSPITAL, INC., Boston, MA (US); and CHARITE UNIVERSITATSMEDIZIN BERLIN, Berlin (DE)
PCT Filed Dec. 16, 2019, PCT No. PCT/US2019/066579
§ 371(c)(1), (2) Date Jun. 15, 2021,
PCT Pub. No. WO2020/124084, PCT Pub. Date Jun. 18, 2020.
Claims priority of provisional application 62/818,446, filed on Mar. 14, 2019.
Claims priority of provisional application 62/780,224, filed on Dec. 15, 2018.
Prior Publication US 2022/0050996 A1, Feb. 17, 2022
Int. Cl. G06K 9/00 (2022.01); G06T 7/00 (2017.01); G06V 10/426 (2022.01); G06V 20/69 (2022.01); A61B 90/20 (2016.01)
CPC G06V 10/426 (2022.01) [G06T 7/0014 (2013.01); G06V 20/69 (2022.01); G06V 20/698 (2022.01); A61B 90/20 (2016.02); G06T 2207/10056 (2013.01); G06T 2207/10081 (2013.01); G06T 2207/30024 (2013.01)] 9 Claims
OG exemplary drawing
 
1. A system comprising:
an digital microscope that provides an image of tissue having a glandular epithelial component as a whole slide image of a stained histological slide, the image representing a plurality of medium-scale epithelial components;
a processor; and
a non-transitory computer readable medium storing instructions executable by the processor, the executable instructions comprising:
a cell identification component that identifies, for each of a plurality of cells within the image, a representative point to provide a plurality of representative points for each of the plurality of medium-scale epithelial components;
a graph constructor that constructs, for each of a subset of the plurality of medium-scale epithelial components, a graph connecting the plurality of representative points;
a feature extractor that determines, for each of the subset of medium- scale epithelial components, a plurality of classification features from the graph constructed for the medium-scale epithelial component, the feature extractor extracting the plurality of classification features only for medium-scale epithelial components for which a number of representative points in the plurality of representative points for the medium-scale epithelial component exceeds a threshold value, such that the subset of the plurality of medium-scale epithelial components is a proper subset; and
a machine learning model that assigns a clinical parameter to each medium-scale epithelial component according to the extracted plurality of classification features.