http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
딥러닝을 이용한 각막세포로 분화하는 중간엽 줄기세포의 형태학적 예측
박건혁 대한기계학회 2021 대한기계학회 춘추학술대회 Vol.2021 No.4
Human neural crest-derived turbinate mesenchymal stem cells (hTMSCs) have demonstrated promising outcomes for clinical cell therapies. These hTMSCs differentiate into specific cells characterized by molecular techniques such as immunostaining or lineage tracing. However, the characterization of hTMSCs from their morphology is a challenging task. Deep learning can extract features through an image that are still abstruse for experts. We hypothesize that the appearance of the hTMSCs contains morphological information that can be analyzed by convolutional neural networks (CNN) to identify their lineage. Here, we used deep learning to establish an automated method to identify cell types without molecular characterizing techniques. CNNs were trained to predict whether morphological images stained with the surface marker CD105 contain pluripotent cells, and the accuracy of classification in single cells was estimated as 94%. Immunofluorescence staining for SSEA3 was used to validate the predictions. Additionally, an object detection algorithm has been trained to uncover differences in morphology from immunofluorescent images. Hence, the proposed assay could be used to assess the pluripotency of mesenchymal stem cells in a non-invasive manner.