US 12,266,445 B2
System and method for characterizing cellular phenotypic diversity from multi-parameter cellular, and sub-cellular imaging data
Srinivas C. Chennubhotla, Pittsburgh, PA (US); Filippo Pullara, Pittsburgh, PA (US); and Samantha A. Furman, Pittsburgh, PA (US)
Assigned to UNIVERSITY OF PITTSBURGH—OP THE COMMONIA/BALTH SYSTEM OF HIGHER EDUCATION, Pittsburgh, PA (US)
Filed by University of Pittsburgh—Of the Commonwealth System of Higher Education, Pittsburgh, PA (US)
Filed on Apr. 3, 2024, as Appl. No. 18/625,784.
Application 18/625,784 is a continuation of application No. 17/605,423, granted, now 11,972,858, previously published as PCT/US2020/032637, filed on May 13, 2020.
Claims priority of provisional application 62/847,622, filed on May 14, 2019.
Prior Publication US 2024/0266036 A1, Aug. 8, 2024
Int. Cl. G16H 30/40 (2018.01); G06F 18/23 (2023.01); G16H 10/40 (2018.01); G16H 30/20 (2018.01)
CPC G16H 30/40 (2018.01) [G06F 18/23 (2023.01); G16H 10/40 (2018.01); G16H 30/20 (2018.01)] 25 Claims
OG exemplary drawing
 
1. A method of characterizing cellular phenotypes from multi-parameter cellular and sub-cellular imaging data for a number of tissue samples from a number of patients or a number of multicellular in vitro models, comprising:
receiving segmented multi-parameter cellular and sub-cellular imaging data which identifies a plurality of cells, wherein the segmented multi-parameter cellular and sub-cellular imaging data is generated by performing cellular segmentation on the multi-parameter cellular and sub-cellular imaging data; and
recursively applying soft/probabilistic clustering to the segmented multi-parameter cellular and sub-cellular imaging data to identify a plurality of computational phenotypes and create a cellular phenotypic tree having a plurality of levels, wherein the cellular phenotypic tree includes a plurality of terminal nodes each signifying a respective one of the plurality of computational phenotypes, wherein each cell in the segmented multi-parameter cellular and subcellular imaging data has a plurality of ownership probabilities each indicating the probability that the cell belongs to a respective one of the computational phenotypes and wherein each cell is probabilistically assigned to one or more of the computational phenotypes based on the ownership probabilities of the cell, wherein a subset of the cells are non-specialized cells wherein the ownership probabilities thereof are below a predetermined threshold, and wherein at each of the levels the ownership probabilities of the non-specialized cells have been optimized based on an ownership confidence term and a spatial coherence term to filter false positive non-specialized cells and promote ownership confidence of the non-specialized cells in the one or more computational phenotypes to which the non-specialized cells have been assigned and to promote spatial coherence among the non-specialized cells.