A Medical Image Archive Solution in the Cloud

Chia-Chi Teng, Jonathan Mitchell, Christopher Walker, Alex Swan, Cesar Davila, David Howard, Travis Needham
School of Technology Brigham Young University Provo, UT, USA ccteng@byu.edu

AbstractGrowing long-term cost of managing an onsite medical imaging archive has been a subject which the health care industry struggles with. Based on the current trend, it is estimated that over 1 billion diagnostic imaging procedures will be performed in the United States during year 2014, generating about 100 Petabytes of data. The high volume of medical images is leading to scalability and maintenance issues with healthcare providers onsite picture archiving and communication system (PACS) and network. Cloud computing promises lower cost, high scalability, availability and disaster recoverability which can be a natural solution some of the problems we faced for long-term medical image archive. A prototype system was implemented to study such as solution on one of the industry leading cloud computing platform, Microsoft Windows Azure. It includes a Digital Imaging and Communications in Medicine (DICOM) server which handles standard store/query/retrieve requests; a DICOM image indexer that parses the metadata and store them in a SQL Azure database; and a web UI for searching and viewing archived images based on patient and image attributes. The comprehensive tools and functionality of Windows Azure made it an ideal platform to develop and deploy this kind of service oriented applications. Keywords-medical imaging; cloud computing; DICOM; PACS

Fig. 1. System block diagram of the DICOM image archive service. Traditional healthcare IT stores and manages image archives onsite using the internal hospital network that is protected by a firewall from the outside. However, the rising capital and management cost of onsite systems as well as the lack of disaster recovery provision have motivated alternative offsite solutions. Recent advancements in cloud computing technology show promises of lower cost, higher scalability, accessibility, availability and disaster recoverability. Computing and storage in the cloud seem to be a natural solution to many problems we face today for long-term medical image archives [3]. While researchers and policy makers are still actively studying the security, privacy, and liability issues involving sensitive medical information in the cloud [4], various technology vendors such as IBM and Amazon have started to provide solutions for early adopters [5][6]. As cloud computing continues to gain momentum and attract research activities, a group of senior Brigham Young University Information Technology (IT) students are inspired to design and implement a medical image archive service using the Digital Imaging and Communications in Medicine (DICOM) standard as a cloud computing based solution based on Microsoft Windows Azure [12] platform and tools as their senior capstone project. The prototype

I.

INTRODUCTION

Growing long-term cost of managing an onsite medical imaging archive has been a subject which the health care industry struggles with. Based on the current trend, it is estimated that over one billion diagnostic imaging procedures will be performed in the United States during year 2014, which will generate about 100 Petabytes of data [1]. The high volume of medical images is leading to scalability and maintenance issues with healthcare providers onsite picture archiving and communication system (PACS) and network. For some healthcare providers among the twothirds of the worlds population who have no access to quality medical imaging services [2], secure on-site data center may not even be possible because of political or civil instability. Information technology for healthcare providers needs long-term image archive solutions that balance cost, image volume, storage capacity and access demand.

Fig. 2. Azure service configuration interface in Microsoft Visual Studio. system includes: 1) a DICOM server which handles standard DICOM store/query/retrieve requests; 2) a DICOM image indexer that parses the metadata (tags) and store the information a SQL Azure database; 3) a web user interface (UI) implemented in ASP.NET and Microsoft Silverlight [14] technologies that allows users to search and view archived DICOM images based on any and combination of DICOM tags. The DICOM server and indexer are implemented in C# programming language based on the open source project DICOM# [10] which implements the DICOM server communication protocol. The programming language support, rich functionality and tool set has shown that Windows Azure is an excellent platform for developing and deploying a service oriented application such as this. II. BACKGROUND RESEARCH Comprehensive development tool from Microsoft and third party, e.g. Visual Studio, Azure Storage Explorer. - Extensive programming language support including C#, ASP.NET, Python. - Availability of .NET services and application programming interface (API). - Local cloud simulacrum known as the development fabric with both computing and storage support. It allows engineers to develop, test and debug applications on local desktop more efficiently before deploying to the cloud. - Simple web based deployment and management interface. Data Grid architecture has been utilized for medical image archive services by Hewlett-Packard (HP) [15] and in other research oriented systems [7]. While Cloud Computing and Grid Computing share many commonalities in their technology [8], Cloud services have a few advantages as a service platform such as flexibility in dynamic provisioning through virtualization [9]. Windows Azure platform provides scalable virtualization of execution, storage, and database which makes in an ideal platform for this application. III. SYSTEM OVERVIEW -

Cloud computing is emerging as a new trend in computational and storage resource allocation and provisioning technology. Many market researchers had projected multi-billion dollar growth in the industry [16][17] as it promises lower cost and higher scalability to customers who need flexible and on demand access to computing resources. Buyya et al. [9] provided an extensive analysis on some of the commercially available industry leading cloud platforms, including Amazon Elastic Compute Cloud (EC2) [11], Google App Engine [12], and Microsoft Azure [13]. The Azure provides an integrated environment for development, hosting and management with more homogeneous hardware and software optimized for performance. It also offers other features such as,

On Windows Azure, a service application usually has multiple instances of various components running in virtual machines (VM). Each instance can be defined as either a Web role for front end UI or Worker role for UI-less back end. Preconfigured VMs optimized for Web or Worker role

Fig. 3. Microsoft Windows Azure SDK development fabric. is provisioned for each instance. As shown in Figure 1, there are three major components in our prototype system: - DICOM server: a UI-less Worker role which communicates with clients using standard DICOM protocol through external TCP connections. It responds to clients requests such as querying, storing or retrieving of DICOM images. This is implemented in C# programming language based on the open source project DICOM#. - DICOM file indexer: also a UI-less Worker role which parses the header of DICOM files as they are uploaded from the client through DICOM server or Web UI. Values of standard attributes (tags) are extracted from the header and stored in a SQL Azure database, where the image files are indexed by their attributes or metadata. This is also implemented in C# based on DICOM# project [10]. - Web UI: a Web role component which provides user interface to upload images on a web page or via HTTP. Users can also search and browse archived DICOM images through a search page implemented in Microsoft Silverlight and ASP.NET. This provides the convenient connectivity to non-DICOM client to upload, search and browse images. Each of these components can be scaled up or down independently on demand by increasing or decreasing the number of instances. The initial instances can be configured in either in the Visual Studio development environment (see Figure 2) or Web based Azure online management portal. The system design also includes a portable application program interface (API) layer which will enable this open source service to be ported to other cloud platforms or even standalone servers. The API can be categorized into the following modules, - Storage API: supports Azure Storage or local file systems. - Queue API: supports Azure message queue. - Database API: uses ADO.NET library to connect to SQL Azure or local SQL databases. A. Implementation Detail In order to take advantage of the built-in scalability of Windows Azure fabric, this system is designed based on asynchronous programming design pattern and stateless dataflow architecture. That is, each instance of execution, either Web Role or Worker Role, does not save any state information. The Web UI and DICOM server will receive a file upload request from the client and send an indexing message to the indexer through the Azure message queue. An instance of the indexer will wake up and pop the indexing message from the queue, parse the newly uploaded DICOM file and store the metadata in the SQL Azure database. The indexing message will only be removed if the indexing process is successful. In case the instance terminates abnormally before completing the indexing, the message will be made available for other instances and get another chance to be processed. The mechanism provides a level of transactionality and robustness. The UI-less DICOM server worker role instances communicate with standard DICOM clients through external TCP end points provided by the Azure .NET layer. Azure also provides an IP filtering mechanism to prevent connections from unintended clients.

computing platform for rapid development. Each healthcare provider can easily customize the system to serve their unique requirement. In addition to the security protocols built in to the Azure cloud services, we will need to implement additional security through IP filtering and DICOM Part 15 security profiles such as transport layer security and digital signature. Further study is needed to evaluate the full compliance of Health Insurance Portability and Accountability Act (HIPAA) [19]. ACKNOWLEDGMENT This project is supported by the Microsoft Windows Azure academic pilot program. Fig. 4. Partial schema in the SQL Azure database.
[1]

REFERENCES
Prepare for disaster & tackle terabytes when evaluating medical image archiving, Frost & Sullivan, http://www.frost.com, 2008. WHO report, Essential diagnostic imaging, World Health Organization, http://www.who.int/eht/en/DiagnosticImaging.pdf. B.J. Liu, F. Cao, M.Z. Zhou, G. Mogel, L. Documet, Trends in PACS image storage and archive, Computerized Medical Imaging and Graphics P.T. Jaeger, T. Lin, and J.M. Grimes, Cloud computing and information policy computing in a policy cloud, Journal of Information Technology & Politics, vol 5(3), pp. 269-283, 2008. http://insiteone.com/news-pressreleases.php?newsID=63 http://aws.amazon.com/solutions/case-studies/harvard/ H.K. Huang, A. Zhang, B. Liu, Z. Zhou, J. Document, N. King, and L.W.C. CHan, Data grid for large-scale medical image archive and analysis, in Proceedings of the 13th annual ACM international conference on Multimedia, Nov. 2005, pp. 100513. http://doi.acm.org/10.1145/1101149.1101357 I. Foster , Y. Zhao , I. Raicu , and S. Lu , Cloud Computing and Grid Computing 360-Degree Compared, Proc. IEEE Grid Computing Environments Workshop (GCE08), pp. 1-10, 2008. http://dx.doi.org/10.1109/GCE.2008.4738445. R. Buyya, C.S. Yeo, S. Venugopal, J. Broberg, and I. Brandic, Cloud comping and emerging IT plaforms: vision, hype, and reality for delivering computing as the 5th utility, Future Generation Computing Systems, vol. 25(6), pp. 599-616, June, 2009. http://dicom-cs.sourceforge.net/ Amazon elastic compute cloud (EC2). http://www.amazon.com/ec2/ Google app engine. http://appengine.google.com Microsoft windows azure. http://www.microsoft.com/azure/ Microsoft silverlight. http://www.microsoft.com/silverlight/ HP medical archive solution. http://h71028.www7.hp.com/enterprise/us/en/software/informationmanagement-governance-ediscovery-medical-archive-solution.html S. Hamm, How cloud computing will change business, Business Week, June 2009. http://www.businessweek.com/magazine/content/09_24/b413504294 2270.htm Morgan Stanley, Technology trends, June 2008. http://www.morganstanley.com/institutional/techresearch/pdfs/TechT rends062008.pdf Image Information Systems. http://www.image-systems.biz/iq-view. Health Insurance Portability and Accountability Act. http://www.hhs.gov/ocr/privacy/.

B. Database Schema Diagnostic imaging procedures usually produce DICOM images as studies and series. Clinicians also usually query and view images as studies and series. Metadata of the images are stored and indexed for fast retrieval in common scenarios and minimize the size of the data store. The partial logical schema in Figure 4 shows that images are linked to series; and series are linked to studies. IV. RESULTS

[2] [3]

[4]

Windows Azure development tools provide an integrated environment for various stages in the product life cycle. First of all, the online portal provides a simple way to create and manage services. Once the service is created, Visual Studio in conjunction with Azure SDK and development fabric allow developers to code, run and debug in local fabrics that simulate cloud VMs. Being able to debug the system locally before deploying to the cloud can dramatically reduce the development time and increase productivity. Finally, deploying the project to the cloud is a simple package-andupload process. This prototype system was tested with a variety of public domain DICOM test images in various modalities. The DICOM client-server operation was tested with commercially available DICOM tools such as iQDICOMTEST and iQ-VIEW [18]. Image series are successfully sent from clients, received and indexed by server in the cloud, returned in the queries, and retrieved as requested. V. CONCLUSIONS AND FUTURE WORK

[5] [6] [7]

[8]

[9]

[10] [11] [12] [13] [14] [15]

[16]

The design and implementation of this prototype system demonstrated the feasibility of using the cloud computing platform to provide a long term offsite medical image archive solution. It has the potential to lower the cost of storage and management and to increase disaster recoverability. The rich feature set, advanced tools and extensive library support made Azure an ideal cloud

[17]

[18] [19]