Research On Transformation And Segmentation Of Liver CT Image Based On OpenEHR

With the development of medical image and computer technology, the amount of medical image data is more and more big, and it plays an increasingly important role in disease diagnosis and treatment. Thus, standardization of medical image data and the intelligent medical service based on medical images become a hot research spot of medical informatization. Recently, most of the hospital medical imaging system (PACS) adopt the DICOM standard, but due to the closure and heterogeneity between different hospital systems, data gap and information isolated problem of medical image data is still serious, which not only hindered the sharing of medical image data and establishment of electronic health records on personal medical images, but also increased the difficulty for the construction of medical images based intelligent medical services.To solve these problems, this paper studied the DICOM and openEHR standard, taking advantage of openEHR’s feature of two layers modeling and using liver CT image data as sample, proposed and designed a liver CT image transformation and segmentation system based on openEHR standard. The system realized sharing of liver CT image data, established electronic health records and supported intelligent analysis service based on liver CT image. This paper also implement the two key module in the system respectively:Transformation and segmentation, and solved the two key problems:liver CT image conversion from DICOM to openEHR and liver CT image segmentation algorithm. For conversion module, this paper builds archetypes and template of liver CT images of openEHR specification, put forward a transformation method by the way of intermediate file mapping, and uses the corresponding technical solution to realize the conversion and relevant child modules. For the segmentation module, an improved region growing method based on adaptive threshold and semi-automatic seed point selection is proposed in this paper. Gray’s statistical characteristics of liver CT images was used to optimize threshold segmentation and improve segmentation results and correlation between the layers of liver CT image sequence was used for semi-automatic seed point selection, which can improve the efficiency of segmentation. The effectiveness of the algorithm is verified by simulation, as well as the segmentation module. The realization of the transformation and segmentation module provides support for the construction of electronic health records and intelligent analysis services of liver CT.