US 12,254,624 B2
Artificial intelligence enabled reagent-free imaging hematology analyzer
Renjie Zhou, Hong Kong (CN); Xin Shu, Hong Kong (CN); and Rishikesh Pandey, Unionville, CT (US)
Assigned to The Chinese University of Hong Kong, Hong Kong (CN)
Filed by The Chinese University of Hong Kong, Hong Kong (CN)
Filed on Dec. 9, 2021, as Appl. No. 17/547,033.
Claims priority of provisional application 63/123,111, filed on Dec. 9, 2020.
Prior Publication US 2022/0180515 A1, Jun. 9, 2022
Int. Cl. G06T 7/00 (2017.01); G06N 3/045 (2023.01); G06V 10/778 (2022.01); G06V 10/94 (2022.01); G06V 30/18 (2022.01); G06V 30/19 (2022.01)
CPC G06T 7/0012 (2013.01) [G06N 3/045 (2023.01); G06V 10/778 (2022.01); G06V 10/94 (2022.01); G06V 30/18057 (2022.01); G06V 30/1912 (2022.01); G06V 30/19127 (2022.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/30024 (2013.01)] 28 Claims
OG exemplary drawing
 
1. A method of classifying information of leukocytes in a sample, the method comprising:
training a neural network according to parameters derived from digitally observed features of known leukocytes;
obtaining digitally observed features of known leukocytes using data cleaning, whereby outliers values are removed based on dry mass and area;
tuning the neural network as a function of at least one property of a precision-recall curve and F1 score representing leukocyte classifications of the known leukocytes generated by the neural network based on the digitally observed features;
configuring an imaging device with the trained and tuned neural network to generate data of observed leukocytes;
storing the data of observed leukocytes relating to observed leukocytes features; and
classifying the data of observed leukocytes into at least two classes by the trained and tuned neural network based on the observed leukocyte features.