US 12,131,481 B2
Automated Tumour-Stroma interface zone detection for anti-tumour response assessment by immunogradient indicators
Arvydas Laurinavicius, Verbiskiu village (LT); Allan Rasmusson, Vilnius (LT); Ausrine Nestarenkaite, Vilnius (LT); Dovile Zilenaite, Vilnius (LT); and Renaldas Augulis, Vilnius (LT)
Assigned to Vilnius University, Vilnius (LT)
Appl. No. 17/600,614
Filed by Vilnius University, Vilnius (LT)
PCT Filed Apr. 9, 2020, PCT No. PCT/IB2020/053396
§ 371(c)(1), (2) Date Oct. 1, 2021,
PCT Pub. No. WO2020/208571, PCT Pub. Date Oct. 15, 2020.
Claims priority of application No. 2019 509 (LT), filed on Apr. 9, 2019.
Prior Publication US 2022/0138955 A1, May 5, 2022
Int. Cl. G06T 7/00 (2017.01); G06T 7/13 (2017.01); G06T 7/136 (2017.01)
CPC G06T 7/13 (2017.01) [G06T 7/136 (2017.01); G06T 2207/10056 (2013.01); G06T 2207/30024 (2013.01); G06T 2207/30096 (2013.01)] 11 Claims
OG exemplary drawing
 
1. An automated method for assessment of cell density and biomarker expressions in tumour/host interface zone, wherein the method comprises the following steps:
a) grid-based subsampling of digital image analysis data (DIA) of one or more microscopy images of tissue sections stained by antibodies binding specifically to antigens expressed by immune cells and other inflammatory cells;
b) detecting a tumour edge by combining tissue content and neighbourhood properties of subsampled image analysis data to sample the tumour edge independent of tumour growth pattern;
c) spatial ranking of sub-sampled tissue sections according to tissue content and distance to the sampled tumour edge;
d) extracting an interface zone of variable width, wherein the interface zone contains all ranked tissue sections within the variable width from the tumour edge; and
e) quantifying cell densities, biomarker expressions, and their variances across the interface zone to generate gradient-type indicators expressing a toward tumour-gradient, wherein said gradient-type indicators are selected from one or more of: centre of mass (CM) and immunodrop (ID).