Research On The Key Technology Of The Regional Medical Imaging Collaboration Services Platform Based On Hadoop

With the development of medical imaging examination technology,it is an important basis for modern clinical diagnosis and treatment,and the accuracy of medical imaging diagnosis will directly affect the therapeutic effect of clinical diseases.The main factors that affect the accuracy of medical imaging diagnosis include two aspects,one is the accuracy of medical imaging equipment,the other is a medical imaging diagnosis'level of physicians experience.Because of the development of medical equipment manufacturing technology and the support of national policy,medical imaging equipment has been widely used in the township level,and the accuracy of the inspection has increased every year.However,there are still a lot of problems in the medical imaging diagnosis,the use of imaging equipment and the management of image in primary medical institutions:1)Serious lack of high level of image diagnosis physician.2)because of the cost problem,the primary hospital can not put a large amount of cost to improve the hardware and software performance of PACS/RIS.The instability of system is easy to cause loss of patient data,privacy leaks and so on.3)The information is relatively occluded and it is difficult to exchange the medical Imaging professional information with the doctors in the high level hospital.so it is hard to improve the diagnostic level of the image.4)A small number of patients that wasting the basic hospital imaging equipment resources.How to improve the present situation of the primary medical institutions is a major challenge to the development of our medical undertakings.Thanks to the advent of cloud computing technology,many service industries has brought new service patterns.Especially in the health care industry,A large number of health care institutions and health care providers have shown great interest in cloud medicine.The hospitals,medical institutions and health care providers can ease the burden that is the need for computerized technology,management and implementation by hosting health care in the cloud.Besides,hospitals need not spend a lot of manpower and resources to design,implement and maintain data center(server),database and application tasks that entrusted to IT industry service providers.It can help healthcare providers focus more on increasing quality of delivered healthcare instead of managing their IT.This is especially important for smaller hospitals,community care and physician practices.In view of the problems existing in the grassroots hospitals,this paper uses cloud computing technology to realize the construction of remote cooperative cloud service platform.It carried out long-distance consultation of medical image,inquiry of typical cases,referral of medical image,distance teaching and other services.It realized the mutual sharing of medical imaging equipment in different regions and the mutual communication and collaboration between imaging diagnosis physicians,which is balanced medical resources.It is of great significance to relieve the pressure in large hospitals,to improve the image diagnosis level of primary medical institutions,to improve the utilization rate of medical imaging equipment in primary hospitals,to improve medical service quality and to reduce medical expenses[1].This paper mainly studies the Cloud service platform of medical image remote collaboration that construct the medical image data storage system and the Integrated Management Service platform through the cloud computing technology.The main work of this article includes:1)The paper analyzes the business requirements of the medical image remote collaboration cloud service platform,the technical progress involved and the main problems that currently facing.The test designed an overall architecture of medical image remote collaboration Cloud service platform is designed,including logic architecture,storage architecture of system and the design of Hospital Gateway.2)Storing a large number of medical image small files directly in Hadoop can cause the HDFS main node to consume too much namenode memory,thereby reducing the performance of the entire cluster.Based on the information hierarchy model of DICOM medical image,we give a merging model of medical images on the basis of Hadoop sequencefile and Mapfile,aiming at the performance problems brought by mass medical image small file storage.The image merging model which is suitable for mass medical image storage in HDFS is obtained.3)However,the simple HDFs Distributed File system is not very suitable for the random reading of a single record,in order to meet the needs of the data retrieval based on medical image,we put forward a set of medical image retrieval scheme based on multilevel index on the basis of the HBase database and the merging model of mass medical image.HBase database has the characteristics of large scale and many dimensions,and can support large scale tables,so it accords with the management of massive medical image metadata.At present,we take the Foshan hospital as the medical Imaging Diagnosis Center and establish interconnection with other primary hospital that can upload medical image data.The project has built 9 medical institutions on medical image cloud service platform.All the patients' medical information can be standardized and collected,so that the diagnosis and treatment information resources of the hospital on the platform could be share each other and cross-hospital transfer.