US 12,223,704 B2
Label-free cell classification and screening system based on hybrid transfer learning
Xuantao Su, Jinan (CN); Zhuo Wang, Jinan (CN); Chao Liu, Jinan (CN); Junkun Jia, Jinan (CN); Kun Song, Jinan (CN); and Hong Liu, Jinan (CN)
Assigned to SHANDONG UNIVERSITY, Jinan (CN)
Filed by SHANDONG UNIVERSITY, Jinan (CN)
Filed on May 25, 2022, as Appl. No. 17/804,073.
Claims priority of application No. 202110592184.5 (CN), filed on May 28, 2021.
Prior Publication US 2022/0383629 A1, Dec. 1, 2022
Int. Cl. G06V 10/82 (2022.01); G06N 3/088 (2023.01); G06T 7/00 (2017.01); G06V 10/26 (2022.01); G06V 10/764 (2022.01)
CPC G06V 10/82 (2022.01) [G06N 3/088 (2013.01); G06T 7/0012 (2013.01); G06V 10/273 (2022.01); G06V 10/764 (2022.01); G06T 2207/20081 (2013.01); G06T 2207/20084 (2013.01); G06T 2207/30024 (2013.01)] 8 Claims
OG exemplary drawing
 
1. A label-free cell classification and screening system based on hybrid transfer learning, comprising:
i) a data preprocessing circuit, which is configured to: acquire 2D light scattering video data, preprocess the 2D light scattering video data to obtain images after removing disturbance;
ii) an image archiving circuit, which is configured to: sort and label the images to get labeled images according to a ground truth;
iii) a feature extraction circuit, which is configured to: get feature vector of the labeled images using a first convolutional neural network with pre-trained parameters; and
iv) a cell classification and screening circuit, which is configured to: input the feature vector into a trained support vector machine model to get cell classification results;
wherein,
the support vector machine model is trained with feature vectors of clinical samples and transferred feature vectors of cell-lines;
preprocessing the 2D light scattering video data, including a digital cell filtering technique, including:
videos of the 2D light scattering video data are divided into the images data frame by frame, and then the images are filtered;
each image of the images is processed by morphological granularity analysis algorithm to obtain an image morphological granularity characteristic value; determine whether a feature value meets a pre-set standard, if so, keep the image, otherwise remove the image; and
a trained machine learning model is used to further filter the images.