《电子技术应用》
您所在的位置:首页 > 人工智能 > 设计应用 > 基于区块链的医疗影像数据人工智能检测模型
基于区块链的医疗影像数据人工智能检测模型
网络安全与数据治理 4期
陈思源1,2,谭艾迪3,魏双剑3,盖珂珂2,4
(1.北京理工大学 计算机学院,北京100081;2.北京理工大学长三角研究院(嘉兴),浙江 嘉兴314019; 3.中国船舶工业综合技术经济研究院,北京100081;4.北京理工大学 网络空间安全学院,北京100081)
摘要: 基于深度学习的目标检测技术被广泛应用于医疗检测领域,该技术依赖大量医疗影像训练分类模型,从而为医生决策提供有力的辅助医疗手段。因涉及患者隐私并直接关系到医生诊断,所以医疗影像数据的共享必须保护患者隐私并确保数据准确不被篡改,而现有中心化的医疗数据存储方案面临隐私泄露等诸多安全问题。提出了一种基于区块链的医疗影像数据人工智能检测模型。该模型针对目标检测技术辅助医生诊断的问题,采用区块链技术实现去中心化、不可篡改的训练参数聚合,通过加密和签名技术保护数据隐私,利用智能合约评估服务器诊断准确率,有助于解决医疗数据壁垒和医疗隐私泄露问题。
中图分类号: TP311
文献标识码: A
DOI: 10.19358/j.issn.2097-1788.2022.04.003
引用格式: 陈思源,谭艾迪,魏双剑,等. 基于区块链的医疗影像数据人工智能检测模型[J].网络安全与数据治理,2022,41(4):21-25.
Blockchain-based artificial intelligence detection model for medical data
Chen Siyuan1,2,Tan Aidi3,Wei Shuangjian3,Gai Keke2,4
(1.School of Computer Science,Beijing Institute of Technology,Beijing 100081,China; 2.Yangtze Delta Region Academy of Beijing Institute of Technology,Jiaxing 314019,China; 3.China Institute of Marine Technology and Economy,Beijing 100081,China; 4.School of Cyberspace Science and Technology,Beijing Institute of Technology,Beijing 100081,China)
Abstract: Deep learning-based target detection technology is being widely used in the field of medical detections. For training a large number of medical images, we can construct an effective classification model to effectively predict the disease situation of patients and provide a powerful auxiliary medical means of decision-making. In order to improve the prediction accuracy, massive training data are the premise to construct an effective learning model. However, medical data involve patients′ privacy and are directly related to diagnoses. Sharing medical data needs to guarantee privacy, accuracy and tamper-proof. Existing centralized medical storage schemes face many security issues, e.g., privacy disclosure. This paper proposes a blockchain-based artificial intelligence detection model for medical data that uses a target detection technology to assist physicians during the diagnosis process. In our model, blockchain technology supports realizing the decentralized and un-tampered aggregation of training parameters. Encryption and signature technology are used to protect privacy and smart Contract is implemented to evaluate the accuracy of server diagnosis. The proposed model will contribute to solving the issues in medical data barriers and privacy disclosure.
Key words : deep learning;blockchain;secure data sharing;artificial intelligence detection

0 引言

医院每天产生和诊断大量的医疗影像,据统计在医疗数据中,影像数据占数据总量的90%以上。随着医疗检测设备的更新换代和不断增加,影像数据以每年超过30%的增长速度急剧增加。与此形成鲜明对比的是,医生数量缓慢增长,这使得影像诊断如阅读分析CT(计算机断层扫描)等工作对医生造成的负担日益加剧,经验缺乏与工作量增大容易造成误诊。随着大数据和人工智能技术的发展,利用计算机辅助诊断,使用基于人工智能的目标检测技术帮助医生做出快速判断,对减轻医生负担、增加诊断准确率、提高就诊效率而言就显得十分必要且具有现实意义。

目标检测技术因其广泛的现实应运用场景备受学术界和工业界关注。随着计算机算力的不断提升,目标检测技术蓬勃发展,衍化出双阶段和单阶段两大类。




本文详细内容请下载:https://www.chinaaet.com/resource/share/2000004985





作者信息:

陈思源1,2,谭艾迪3,魏双剑3,盖珂珂2,4

(1.北京理工大学 计算机学院,北京100081;2.北京理工大学长三角研究院(嘉兴),浙江 嘉兴314019;

3.中国船舶工业综合技术经济研究院,北京100081;4.北京理工大学 网络空间安全学院,北京100081)



微信图片_20210517164139.jpg

此内容为AET网站原创,未经授权禁止转载。