Lu J. Huang

Design of a graphical user interface for an intelligent multimedia information system for radiology research

With the increase in the volume and distribution of images and text available in PACS and medical electronic health-care environments it becomes increasingly important to maintain indexes that summarize the content of these multi-media documents. Such indices are necessary to quickly locate relevant patient cases for research, patient management, and teaching. The goal of this project is to develop an intelligent document retrieval system that allows researchers to request for patient cases based on document content. Thus we wish to retrieve patient cases from electronic information archives that could include a combined specification of patient demographics, low level radiologic findings (size, shape, number), intermediate-level radiologic findings (e.g., atelectasis, infiltrates, etc.) and/or high-level pathology constraints (e.g., well-differentiated small cell carcinoma). The cases could be distributed among multiple heterogeneous databases such as PACS, RIS, and HIS. Content- based retrieval systems go beyond the capabilities of simple key-word or string-based retrieval matching systems. These systems require a knowledge base to comprehend the generality/specificity of a concept (thus knowing the subclasses or related concepts to a given concept) and knowledge of the various string representations for each concept (i.e., synonyms, lexical variants, etc.). We have previously reported on a data integration mediation layer that allows transparent access to multiple heterogeneous distributed medical databases (HIS, RIS, and PACS). The data access layer of our architecture currently has limited query processing capabilities. Given a patient hospital identification number, the access mediation layer collects all documents in RIS and HIS and returns this information to a specified workstation location. In this paper we report on our efforts to extend the query processing capabilities of the system by creation of custom query interfaces, an intelligent query processing engine, and a document-content index that can be generated automatically (i.e., no manual authoring or changes to the normal clinical protocols).

Implementation of system intelligence in a 3-tier telemedicine/PACS hierarchical storage management system

Our tele-medicine/PACS archive system is based on a three-tier distributed hierarchical architecture, including magnetic disk farms, optical jukebox, and tape jukebox sub-systems. The hierarchical storage management (HSM) architecture, built around a low cost high performance platform [personal computers (PC) and Microsoft Windows NT], presents a very scaleable and distributed solution ideal for meeting the needs of client/server environments such as tele-medicine, tele-radiology, and PACS. These image based systems typically require storage capacities mirroring those of film based technology (multi-terabyte with 10+ years storage) and patient data retrieval times at near on-line performance as demanded by radiologists. With the scaleable architecture, storage requirements can be easily configured to meet the needs of the small clinic (multi-gigabyte) to those of a major hospital (multi-terabyte). The patient data retrieval performance requirement was achieved by employing system intelligence to manage migration and caching of archived data. Relevant information from HIS/RIS triggers prefetching of data whenever possible based on simple rules. System intelligence embedded in the migration manger allows the clustering of patient data onto a single tape during data migration from optical to tape medium. Clustering of patient data on the same tape eliminates multiple tape loading and associated seek time during patient data retrieval. Optimal tape performance can then be achieved by utilizing the tape drives high performance data streaming capabilities thereby reducing typical data retrieval delays associated with streaming tape devices.

Design and implementation of an integrated information model for optimizing workflow in radiology practice

Efficiency is crucial to economic viability in the current climate of intense competition in the healthcare industry. The heterogeneity of the information systems, RIS, PACS, HIS, DDS, often found in radiology practices has made it difficult to fully leverage these costly investments in information technology into efficiency gains. An approach to integrating these heterogeneous systems that will make information technology more effective as an efficiency tool, while preserving the separate hardware and software environment of each system, is presented. Based on an integrated information model and automated message passing between the different systems, this approach can be used to support an automated, proactive sequencing of the radiology workflow.

Large-scale implementation of optimal image compression algorithms

Despite over a decade of research and development, medical image compression has not yet been widely implemented on clinical picture archiving and communication systems (PACS). We have developed a prototype interface which incorporates both lossless and lossy compression into a browsing system that enables the efficient use of network and storage resources. Such a system allows a user to quickly browse through a large set of image icons created from lossy compression and selectively retrieve the original images for diagnosis from the optical disk that contains losslessly compressed image data. For lossless compression, we implemented modality specific techniques which combines preprocessing, adaptive prediction and entropy coding, giving a compression improvement of 20% over JPEG predictors. The lossy compression algorithm consists of subsampling followed by wavelet transform coding and achieves compressed CR images of sufficient quality for browsing at a compression ratio of about 2000:1.

Performance evaluation of a high-speed switched network for PACS

We have replaced our shared-media Ethernet and FDDI network with a multi-tiered, switched network using OC-12 (622 Mbps) ATM for the network backbone, OC3 (155 Mbps) connections to high-end servers and display workstations, and switched 100/10 Mbps Ethernet for workstations and desktop computers. The purpose of this research was to help PACS designers and implementers understand key performance factors in a high- speed switched network by characterizing and evaluating its image delivery performance, specifically, the performance of socket-based TCP (Transmission Control Protocol) and DICOM 3.0 communications. A test network within the UCLA Clinical RIS/PACS was constructed using Sun UltraSPARC-II machines with ATM, Fast Ethernet, and Ethernet network interfaces. To identify performance bottlenecks, we evaluated network throughput for memory to memory, memory to disk, disk to memory, and disk to disk transfers. To evaluate the effect of file size, tests involving disks were further divided using sizes of small (514 KB), medium (8 MB), and large (16 MB) files. The observed maximum throughput for various network configurations using the TCP protocol was 117 Mbps for memory to memory and 88 MBPS for memory to disk. For disk to memory, the peak throughput was 98 Mbps using small files, 114 Mbps using medium files, and 116 Mbps using large files. The peak throughput for disk to disk became 64 Mbps using small files and 96 Mbps using medium and large files. The peak throughput using the DICOM 3.0 protocol was substantially lower in all categories. The measured throughput varied significantly among the tests when TCP socket buffer was raised above the default value. The optimal buffer size was approximately 16 KB or the TCP protocol and around 256 KB for the DICOM protocol. The application message size also displayed distinctive effects on network throughput when the TCP socket buffer size was varied. The throughput results for Fast Ethernet and Ethernet were expectedly lower but the patterns were interestingly different from those for ATM. To achieve the optimum throughput in a TCP-based high-speed switched medial imaging network, the size of the TCP socket buffer is the most important parameter to optimize. If the DICOM 3.0 protocol is used, however, the performance gain by tuning system parameters is minimal, particularly if small files are used. Compared to socket-based TCP, the decrease in throughput caused by DICOM 3.0 protocol overhead is significantly larger in a high-speed switched network. This suggests that the protocol itself is the bottleneck in high-speed networks and that the protocol should be fine-tuned to take advantage of the services provided by such networks and not to duplicate them. To design a successful high-speed PACS network, it is important that bandwidth-demanding workstations and servers be on the same subnet and use the same technology so that no routing and data conversions are required.

Optimization of the image cache throughput for a large-scale PACS archive

With the increasing number of PACS image display workstations deployed in clinical operations, a PACS image server needs to be tuned more effectively to achieve optimal system throughput. There are many factors that affect the performance of image transfer, including the application software program, the storage subsystem, and the network system. This paper primarily addresses the issues involving the image storage subsystem. Related research has been done to address the issues in network system and application software. One primary factor that affects the overall performance of a PACS image server is the throughput and reliability characteristics of the magnetic storage devices for image caching. A redundant array of inexpensive disks (RAID) is a mean to provide reliable and high performance storage. The UCLA PACS image server relies on a software-based RAID system to fulfill the bandwidth requirement and reliability demand. An extensive evaluation on various hardware and software parameters in the I/O subsystem has been conducted to derive an optimal configuration, which specifically attempts to address the following questions: How many disks should be in a stripe group? What is the optimal stripe chunk size? How big should the maxcontig be? What should be the IO request size? And how many concurrent I/O accesses should be allowed?

Scalable storage architecture for the archiving of medical images and information

Storage and archive systems are essential components of a Picture Archiving and Communication System (PACS) and become increasingly important in medical enterprise-wide information systems. Building a large-scale, long-term medical data storage system is complex, involving accurate identification of system requirements, judicious tradeoff between flexibility and complexity and effective execution of implementation plans. The proposed object-oriented software architecture fulfills the key system requirements, while maintains a high degree of flexibility for extending and customizing the framework in various medical data archive application areas. The complexity in developing such a framework has been proven manageable with various prototypes.

Fault tolerance techniques to assure data integrity in high-volume PACS image archives

Picture archiving and communication systems (PACS) perform the systematic acquisition, archiving, and presentation of large quantities of radiological image and text data. In the UCLA Radiology PACS, for example, the volume of image data archived currently exceeds 2500 gigabytes. Furthermore, the distributed heterogeneous PACS is expected to have near real-time response, be continuously available, and assure the integrity and privacy of patient data. The off-the-shelf subsystems that compose the current PACS cannot meet these expectations; therefore fault tolerance techniques had to be incorporated into the system. This paper is to report our first-step efforts towards the goal and is organized as follows: First we discuss data integrity and identify fault classes under the PACS operational environment, then we describe auditing and accounting schemes developed for error-detection and analyze operational data collected. Finally, we outline plans for future research.

System and software engineering for a large-scale PACS

PACS research and development in the past decade mainly emphasized technical issues such as networking, archiving, data communication standards and image display workstation development. In order to fully realize the benefits of these advanced digital technologies in current and future radiology practice, a discipline for development of next generation PACS must be established and practiced. We devised and applied a development process or life cycle that facilitated PACS system definition, development and operation. With the application of the development process, user requirements and expectations of a PACS can be effectively and accurately transformed into system models, and the implementation and the operation of a PACS can be carried out in a systematic way. The development process also provides a framework for repeatable systems and software engineering.

Effective migration to switched-based PACS networks

Asynchronous transfer mode (ATM) switch-based networks can provide high bandwidth for digital video and image data transmission as well as scalability and interoperability between local-area networks and metropolitan or wide area networks. However, migration to ATM is complicated by the limitations of existing radiology networks. Our goal was to develop an effective migration strategy for incorporating ATM into existing radiology networks. We characterized traffic flow for a large-scale, clinical picture archiving and communication system (PACS) with multiple imaging modalities, archives, and display workstations; we then developed a network delay model and, based upon the design criteria, the network delay model, and estimates of future traffic, we designed a switch-based network that uses both Ethernet and ATM switches. Our strategy allows an existing PACS to utilize ATM technology where appropriate, to gain interoperability with wide area networks, and to maximize the investment in an existing PACS infrastructure.