Recently, with greater focus on ethical issues related to animal testing, interest in artificial skin platforms has increased in both cosmetic and medical industries. Artificial skin comprises dermal and epidermal layers. In particular, proper differentiation and proliferation of keratinocytes in the epidermal layer has a considerable influence on the role of the skin as a barrier. However, during 3D culture, real-time monitoring and evaluation of the tissue being cultured are difficult. In this study, Convolutional Neural Network and Convolutional Long Short-Term Memory were utilized for prediction of the artificial skin image. To evaluate the designed models, the similarity between the predicted artificial skin image was compared with the real image. We verified the possibility and practicability of artificial skin image analysis and prediction using neural network models. In the future, this approach could be applied to image prediction under certain conditions, such as inflammatory or skin diseases.

最近，随着对与动物试验相关的伦理问题的更多关注，化妆品和医疗行业对人造皮肤平台的兴趣有所增加。人造皮肤包括真皮层和表皮层。特别是表皮层角质形成细胞的适当分化和增殖对皮肤作为屏障的作用具有相当大的影响。然而，在 3D 培养过程中，对正在培养的组织进行实时监测和评估是很困难的。在这项研究中，卷积神经网络和卷积长短期记忆被用于预测人造皮肤图像。为了评估设计的模型，将预测的人造皮肤图像与真实图像之间的相似性进行比较。我们使用神经网络模型验证了人工皮肤图像分析和预测的可能性和实用性。将来，这种方法可以应用于某些条件下的图像预测，例如炎症或皮肤病。