| 초록 |
Prediction of cellular responsiveness to drugs and biologically applied materials is ultimately required in clinical and laboratory studies. In this study, A549, HEK293, and NCI-H1975 cells were used, which have different molecular shape and drug responsiveness to doxorubicin. The microscopic images of these cells following the exposure to various concentrations of doxorubicin were trained with the exposed molar concentrations. The pretrained model of MobileNetV2 to predict the optical density produced by the colorimetric CCK cell viability assay showed the correlation coefficient r2 0.8723~0.9350, between measured and predicted values. The IC50 values of the above cells for doxorubicin were automatically calculated based on the cellular images according to the exposure concentration of doxorubicin. Based on this study, for customization of the cell type-specific pretraining, user-interface to construct the prediction model was developed in web-based accession protocol. |
