Current PACS provides for the acquisition, storage, display, and communication of radiologic image data. However, the contemporary practice of radiology is highly information rich, increasingly distributed, and imposes formidable requirements for the extraction of physiologic information from image data. Medical information exists within heterogeneous databases that use disparate medical lexicons and for which there are no robust mechanisms to unite data based on concepts. Next generation PACS systems must facilitate cost-efficient management of patients and image-related data over distributed sites, including the management of very high volume image data sets such as magnetic resonance angiography, electron beam and helical CT, and digital subtraction angiography. PACS must also include advanced information processing functions such as concept-based data retrieval, complex image analyses and knowledge-based decision support. In order to address contemporary health care needs efficiently and economically, new PACS architectures should be designed systematically from process models that describe the functions required by PACS users, and from data models that characterize the types, attributes , and relationships of the data components to be managed. Moreover, these systems should be transportable to different health care settings.
In this program project, we propose four highly interrelated projects that address problems in several aspects of radiologic and medical practice. Specific aims are as follows:
  1. To promote efficient consultative health care in loco-regional, national, and global environments through teleradiology systems that provide intelligent access to, and monitoring of, remote PACS resources
  2. To facilitate communication through PACS between radiology subspecialists and other health care providers.
  3. To develop innovative methods for image navigation and the presentation of large multi-sequence image data sets using sophisticated data modeling and image indexing schemes.
  4. To facilitate rapid clinical access to advanced image analyses and related physiologic information through PACS.
  5. To provide transparent access to the heterogeneous databases comprising the electronic medical record from multiple sites, through the formation of new client-mediator-server architectures.
  6. To provide concept-based retrieval of multimedia data in a PACS network through the development of automated natural language processing agents for various medical documents.

The objective of this project is to develop practical and expandable teleradiology systems that will contribute to the distributed practice of subspecialty radiology. Integral to this project are:
  1. significant past accomplishments and current capabilities at UCLA in developing techniques for remote viewing of radiologic images;
  2. a contracted service site (clinical and technical "laboratory") to evaluate the feasibility of teleradiology for subspecialty consultation nationally; and
  3. an academically based multi-specialty radiology practice to test the value of such consultation, particularly with respect to quality of care and cost-effectiveness.

Systemic data and process models have been developed in collaboration with other projects in the Program Project application, from which practical teleradiology systems can be developed. Technical developments for teleradiology emphasize teleconsultation, display paradigms, network management and innovative storage methodologies. When fully developed, these technical developments contribute to expanding the PACS concept. The clinical component of this project addresses cost calculations and cost-effectiveness analyses associated with transporting subspecialty consultation to a non-metropolitan area via teleradiology in daily clinical practice.

Contemporary diagnostic Neuroradiology has evolved in two critical respects, imposing challenging requirements on PACS architectures. First, neuroradiology is associated with high volume image datasets. This requires that intelligent image sorting and presentation algorithms be developed that are patterned after the radiologists' mental paradigms (e.g. sagittal T1 with contrast). Second, interventional neuroradiology has made quantitative image analysis of angiographic data highly desirable. To provide meaningful decision support in the angiographic suite, image data and related computations such as blood flow must be linked and accessible during the procedure. In addition, the ability to interrogate the entire inventory of accumulated image and alphanumeric data on neurointerventional patients would offer both decision support and a basis for outcomes analyses.
In this project, we propose to:
  1. implement logical sorting and display strategies for rapidly viewing large image data sets and
  2. demonstrate on-line acquisition, computation, and integration of quantitative information within a neurointerventional setting.

Based on process models that define the functions of the radiologist and data models that characterize the types of data to be managed, the data attributes and data relationships, we will develop image indexing and presentation strategies to improve the efficiency of soft-copy diagnosis. The analytical capabilities of the workstation will be extended through deployment of densitometric tools to acquire blood flow measurements from digital angiographic data. Following validation in sequential phantom and animal models, the utility of these computational techniques will be tested on-line in the analysis of inflow and outflow vessels in angiographic data on patients with vascular malformations. Finally, through PACS, an alphanumeric and image inclusive database will be developed which allows transparent access to all information pertinent to the treatment of neurointerventional patients. The success of these sophisticated PACS capabilities will help to overcome the remaining practical and psychological barriers to full PACS acceptance.

A number of factors have been instrumental in shaping contemporary thoracic radiologic practice. First, thoracic imaging data is increasingly digital in nature, which has made quantitative image analysis a practical goal. Secondly, thoracic radiology consultation within the multidisciplinary setting of oncology demands that increasing amounts of data be abstracted, indexed, linked, and efficiently communicated within a distributed environment. Although some of the analytical and information management tools required to support these needs exist, it is within the electronic infrastructure of PACS that they become practical and clinically accessible. We propose to expand the capabilities of PACS beyond its historical role to enhance the research and clinical needs of thoracic imaging by providing:
  1. image analysis tools to extract quantitative information from image data pertaining to respiratory function as well as tumor volumes; and
  2. PACS-based information processing tools to integrate all of the multimedia data relevant to the management of cancer patients.

This will be accomplished by developing PACS-based analytical software to address specific physiologic questions in patients with airflow obstruction as well as knowledge-based methods for the measure of tumor and lung volumes. In addition, an Oncology Imaging Time Line will be developed for lung cancer patients that indexes diagnostically relevant image data with textual and numerical data from other databases in the electronic medical record. Using advanced data integration software and automated free text analysis routines to be developed with Project 4, a multimedia summary will be generated by which the progress of cancer patients can be efficiently tracked. Addressing these PACS applications will stimulate advances in the very fabric of contemporary PACS technology, ranging from network strategies to workstation design to sophisticated data management.

The objective of this proposal is to develop a medical information processing system that can perform the following:
  1. provide transparent and easy access to the complete electronic patient record, and
  2. retrieve medical documents and images based on information content (e.g., pathology, radiology, finding) rather than artificial keys such as patient hospital identification number.

A flexible six-level information processing architecture is proposed to perform high-level query answering functions. Central to the project is a three-level object-oriented data model to capture the semantic contents of medical documents and images. The data model will incorporate the National Library of Medicine's Unified Medical Language System (UMLS) Medical Semantic Network and Metathesaurus. Powerful computer-aided software engineering (CASE) tools are used to perform graphical data modeling and rapid code development for the integration of legacy medical information systems. Testbed systems for thoracic and neuroradiology will be developed to demonstrate the clinical utility of the system. A highly graphical multimedia query language is introduced as the system's user interface.

The objectives of the Technology Core are to develop, maintain, and enhance the common PACS infrastructure and computing services shared by the sub-projects. It provides a focus for development and deployment of the following common PACS services
  1. image acquisition;
  2. image archival and retrieval;
  3. image distribution and communication;
  4. database services;
  5. workstation interface development;
  6. deployment of tested software;
  7. consultation for experimental design, data collection, and statistical services; and
  8. PACS user training and education.

The design of our infrastructure should be a model demonstration site for other future PACS installations and will address the following general issues
  1. improvement of system connectivity through adaptation of standards such as the ACR-NEMA DICOM 3.0 standard;
  2. improvement of overall system intelligence and process coordination through a common information processing model;
  3. increase of system workload capacity by implementation of high-speed networks, mass storage devices, hierarchical storage algorithms, data compression methods, and faster computer CPU's;
  4. facilitation of PACS maintenance by use of graphical CASE (Computer-Aided Software Engineering) tools for process modeling and system design;
  5. improvement of PACS reliability by development of centralized system monitoring and recovery software; and
  6. improvement of clinical operation through implementation of clinical quality control protocols.

The PPG Project is funded by the Nat ional Institute of Health (NIH): The NIH gives grants to research institutions for a variety of medical research projects, and serves as the foundation for several other large government organi zed efforts. Current medical research can be found here.