| Due to the increasingly serious environmental pollution,particulate matter,sulfur oxides,and heavy metals in the air are all the culprits that cause human lung diseases.Lung cancer is now one of the most common malignant tumors in the world.Early diagnosis and treatment of lung cancer is the key to improving the survival rate of patients and reducing the medical burden.With the development of artificial intelligence,the use of machine learning algorithms to detect lung nodules is a hot direction in the development of artificial intelligence medical devices,which can significantly improve the quality and efficiency of doctors in analyzing medical images.The data sets used for algorithm training and product testing are very important for the quality management and risk management of the entire life cycle.The training set from a single hospital will undoubtedly lead to over-adaptation of the model,and even model diagnosis bias.Therefore,it is necessary to aggregate data from multiple hospitals to form a lung data set with a large number of cases and rich types of cases to ensure that the built model is reasonable.In this context,Shanghai Institute of Computing Technology,Shanghai Changzheng Hospital,Huashan Hospital affiliated to Fudan University and other 9hospitals,under the guidance of the consensus of relevant experts,built a lung disease data collection platform and realized the transformation of results.In order for the hospital’s patient data to safely come out of the hospital and reach the data collection platform,ethical review must be passed to protect the personal privacy of patients.The problems addressed in this article are as follows:First,we need to clarify what patient data we want to protect;second,in actual operation,due to the different intranet restrictions of each hospital,some hospitals can directly upload patient data to the data collection platform However,some hospitals can only download the data to a U disk to bring it out.A privacy protection scheme is designed for these two situations;finally,how to implement a patient privacy protection scheme.This article mainly focuses on the privacy protection application of the CT image of the DICOM standard.Aiming at the privacy problem in the process of gathering hospital lung disease data to the public data platform,this article mainly proposes a privacy protection scheme and designs a privacy protection tool based on this scheme.The specific work and innovations of this paper are as follows:(1)Based on the research and analysis of the DICOM standard and the structure of the DICOM file,combined with the interpretation of related laws and regulations,we determined that the privacy protection object is the important tag group involving sensitive information in the DICOM file,and this sensitive information is mainly from the United States.18 protected health information stipulated in the HIPPAAct;(2)This article designs a privacy protection scheme.For hospitals that can send data directly to the dataset platform,the main purpose is to desensitize sensitive data in DICOM files using anonymization methods,so that the released data does not contain sensitive information.For hospitals that need to download data to physical media,the RSA encryption algorithm is mainly used to encrypt DICOM files,which not only ensures the security of the data in the transmission process,but also has its own identity recognition.Even if it is an encrypted file,you can know Which hospital the data comes from.(3)According to actual production needs,since the users of these privacy protection schemes are hospitals,considering the hospital’s intranet restrictions,a stand-alone version of the privacy protection tool was chosen,using the Py Charm integrated development environment and Python 3.6.5 interpreter to achieve privacy Protect the tools,and conduct experimental tests.The entire privacy protection tool is encapsulated and the UI page design is carried out through the pyqt5 and Qt Designer that comes with python. |
PDF Full Text Request