Research On Optical Coherence Tomography Three-dimensional Imaging And Data Segmentation Technology

Optical coherence tomography(OCT)is a non-invasive,high-resolution,threedimensional imaging technique.OCT is based on the principle of low-coherence interference to reveal the internal structure of biomedical tissues or materials.OCT imaging technology first achieved success in ophthalmology,and research on applying OCT technology to other biomedical fields has increasingly received more attention.The placenta is an important organ for material exchange between maternal and fetus.Placental villus is the key structure of the placenta.The application of OCT imaging technology to reveal the three-dimensional structure of multi-complicated placental villi will be of great significance for studying the influence of different pregnancy diseases on the morphology of placental villi.At present,OCT technology has been applied in dermatology to reveal skin layering and internal structure details.Medical image segmentation based on deep learning is a research direction that has developed rapidly in recent years.Manual labeling is time-consuming and requires experienced doctors.Using deep learning methods to automatically segment and reconstruct the three-dimensional structure of the skin epidermis can improve efficiency and obtain more intuitive results.This dissertation uses the OCT imaging system built in the laboratory to perform threedimensional imaging of human placental villi and human fingertip skin and proposes an automatic skin segmentation network model based on the U-net network,which can automatically segment epidermis layer of human skin.The main work of this dissertation is as follows:(1)This dissertation uses the OCT imaging system built in the laboratory to perform three-dimensional imaging of healthy and placental villi with different pregnancy diseases,revealing the three-dimensional morphology of multi-complicated placental villi and the impact of different pregnancy diseases on the morphology of placental villi.The data of OCT placental villi were statistically analyzed,and the four morphological parameters of placental villi were quantitatively quantified,which demonstrated that OCT could reveal the three-dimensional morphology of placental villi and quantitative statistical analysis.(2)This dissertation uses the OCT imaging system built in the laboratory to perform three-dimensional imaging of the skin of the fingertip area,revealing the internal layered structure of the skin.In addition,an automatic skin segmentation network model based on the U-net structure is proposed,using the Efficient Net b2 backbone network to extract epidermis feature of human skin in the encoder module,and using residual block to restore image details in the U-net decoder module to improve the segmentation ability of the network.Experiments show that the proposed skin automatic segmentation network could segment epidermis layer in human skin OCT images more precisely and accurately than the traditional U-net network and U-net++ network,and is superior to other networks in various evaluation indicators.