US 12,444,491 B2
Machine learning methods for classifying cells
Ivie Aifuwa, Seattle, WA (US); Jon Charles Gunther, Seattle, WA (US); Keith Rawson, Seattle, WA (US); Vanessa Hasle, Summit, NJ (US); and David Reiss, Summit, NJ (US)
Assigned to Juno Therapeutics, Inc., Seattle, WA (US)
Appl. No. 17/638,789
Filed by Juno Therapeutics, Inc., Seattle, WA (US)
PCT Filed Aug. 28, 2020, PCT No. PCT/US2020/048642
§ 371(c)(1), (2) Date Feb. 25, 2022,
PCT Pub. No. WO2021/041994, PCT Pub. Date Mar. 4, 2021.
Claims priority of provisional application 62/894,463, filed on Aug. 30, 2019.
Prior Publication US 2022/0392613 A1, Dec. 8, 2022
Int. Cl. G06N 3/045 (2023.01); G06F 18/214 (2023.01); G06F 18/2411 (2023.01); G06N 3/08 (2023.01); G06N 5/01 (2023.01); G06N 20/10 (2019.01); G06N 20/20 (2019.01); G16B 40/00 (2019.01); G16H 30/20 (2018.01); G16H 30/40 (2018.01)
CPC G16H 30/20 (2018.01) 16 Claims
OG exemplary drawing
 
1. A method of classifying T cells, the method comprising:
receiving image data associated with each cell of a plurality of T cells of a population of cells comprising T cells, wherein the image data comprises one or more of phase image data, intensity image data, and superposition image data from each cell of the plurality of T cells, and wherein the image data is obtained using differential digital holographic microscopy (DDHM) without using a marker or dye;
determining, from the image data, one or more input features for each cell of the plurality of T cells, wherein the one or more input features comprises a morphological feature of the image data, an optical feature of the image data, an intensity feature of the image data, a phase feature of the image data, a system feature of the image data, or any combination thereof; and
applying the image data from each cell of the plurality of T cells as input to a process, the process comprising a convolutional neural network trained on image data associated with T cells known to belong to a first group and T cells known to belong to a second group, wherein the image data to train the convolutional neural network comprises one or more of phase image data, intensity image data, and superposition image data from each cell of the T cells known to belong to the first group and the T cells known to belong to the second group, to classify each cell of the plurality of T cells as belonging to the first group or the second group, wherein the first group and the second group are selected from: a) live and dead cells, b) CD4+ and CD8+ T cells, or c) recombinant receptor positive and recombinant receptor negative cells.