Research On Web-based Medical Image Acquisition,Registration And Segmentation System Design

In recent years,with the development of computer technology,new techniques such as cloud computing,mobile Internet,GPU acceleration,and intelligent terminals have been continuously developing and penetrating into medical field.A new era of medical image Internet has been arriving.Therefore,the medical image data processing technology based on internet has gradually become a hot spot.To solve evolving medical limitations such as regional medical care,teleconsultation and telemedicine medical teaching,Internet based diagnostic methods is urgently needed allowing doctors to conduct disease diagnosis on any Internet browser.However,the current departments of the hospital are independent to each other and won't share important techniques and information,so that many business can't not be deeply integrated.This situation generally leads to numerous data isolation,cannot rationally allocate and use resources.Obviously,it increases the cost of medical care and affects the clinical needs of medical imaging information.Aiming at these problems,this paper carried out a medical image processing research that can be directly run in the Internet.That is,a Web-based medical image acquisition,registration and segmentation system is designed.The system was used in the cooperative hospital,and the image data of 20 patients in the hospital were collected for registration and segmentation processing.At the same time,the test results of the system were evaluated and analyzed by the professional knowledge of the imaging physician,and the accuracy of the system was found to be more than 85%by the system evaluation results.Specifically,following several modules are included in our Web-based platform:(1)Web-based DICOM(Digital Imaging and Communication of Medicine)medical image file collection and synthesis of 3D image data.This paper can analyzed the data structure and coding mode of DICOM image,realized the format conversion and display of DICOM medical images on the Web platform,and composited three-dimensional medical images in three phases(i.e.,plain phase,arterial phase and portal phase).The experimental results showed that the system platform can efficiently read medical image data and realize resource sharing.(2)Web-based medical image registration research.Feature points based TPS(Thin plate splines)registration algorithm was used for image registration.Representative and unique features of liver was manually chosen by professional experts and taken as the registration performance of ground truth.The effect of the registration was evaluated by using observation and quantitative methods.The experimental results show that the registration effect is good,which provides a reliable computer aided tool for clinicians to analyze the images of hepatocellular carcinoma patients.(3)Web-based medical image segmentation and 3D modeling.Our platform uses techniques including grey level transformation,edge detection,expansion corrosion,labeling and threshold differential segmentation to segment CT images of the plain and portal phases,so that the liver region can be extracted and the 3D reconstruction eventually can be performed by the software of 3D Doctor.Since the abdominal organs have complex structure and are closely related to each other.It is intuitive and clear to observe the patients' pathology by using the Internet and computer techniques to automatically extract the liver of the human abdominal organs and lay the foundation for future pathological analysis.