Image content extraction: application to MR images of the brain.

A system for automatically extracting image content features was developed that combines registration to a labeled atlas with natural language processing of free-text radiology reports. The system was then tested with T1-weighted, spoiled gradient-echo magnetic resonance (MR) imaging studies of the brain performed in nine patients. The locations of 599 structures were visually assessed by an experienced radiologist and compared with the locations indicated by automated output. The in-plane accuracy of the contours was subjectively evaluated as either good, moderate, or poor. The criterion for classifying a structure as correctly located was that 90% or more of all the images containing the structure had to be correctly identified. For 98% of the structures, the images identified by the automated algorithm agreed with those identified by the radiologist, and in 83% of cases, image contours showed a good in-plane overlap. The results of this validation study demonstrate that this combination of registration and natural language processing is accurate in identifying relevant images from brain MR imaging studies. However, the range of applicability of this technique has yet to be determined by applying the technique to a large number of studies.

Automatic structuring of radiology free-text reports.

A natural language processor was developed that automatically structures the important medical information (eg, the existence, properties, location, and diagnostic interpretation of findings) contained in a radiology free-text document as a formal information model that can be interpreted by a computer program. The input to the system is a free-text report from a radiologic study. The system requires no reporting style changes on the part of the radiologist. Statistical and machine learning methods are used extensively throughout the system. A graphical user interface has been developed that allows the creation of hand-tagged training examples. Various aspects of the difficult problem of implementing an automated structured reporting system have been addressed, and the relevant technology is progressing well. Extensible Markup Language is emerging as the preferred syntactic standard for representing and distributing these structured reports within a clinical environment. Early successes hold out hope that similar statistically based models of language will allow deep understanding of textual reports. The success of these statistical methods will depend on the availability of large numbers of high-quality training examples for each radiologic subdomain. The acceptability of automated structured reporting systems will ultimately depend on the results of comprehensive evaluations.

Requirements of a Web-based experiment management system.

Recent advances in tools for scientific data acquisition, visualization, and analysis have lead to growing information management problems for medical research laboratories. An exponential increase in the volume of data, combined with a proliferation of heterogeneous formats and autonomous systems, has driven the need for flexible and powerful Experiment Management Systems (EMS). This paper provides a detailed analysis of the informatics requirements of an EMS, and proposes a new type of middleware called an EMS-Building Environment (EMSBE), which enables the rapid development of web-based systems for managing laboratory data and workflow. We describe the Web-Interfacing Respository Manager (WIRM), an open-source application server for building customizable experiment management systems. WIRM is being used to manage several ongoing experiments, including a natural language processor of radiological findings, and an interdisciplinary project for studying brain function.

Evaluation of SNOMED3.5 in representing concepts in chest radiology reports: integration of a SNOMED mapper with a radiology reporting workstation.

Standardized medical terminologies are gaining importance in the representation of medical data. In this paper, we present the evaluation of the SNOMED3.5 medical terminology to code concepts routinely used in chest radiology reports. Integration of this terminology mapper into a radiology reporting workstation that incorporates a speech recognition system and a natural language processor is also discussed. A total of 700 anatomical location terms (including synonyms) were tested and 72% of the terms had corresponding SNOMED terms. Of the 28% that did not result in a match, 16% were either morphological variants of SNOMED terms or could be found from a combination of terms from two or more SNOMED axes. Only 12% of the terms (primarily specialized radiology terms) were concepts not actually included in the SNOMED terminology.

Teleradiology as a foundation for an enterprise-wide health care delivery system.

An effective, integrated telemedicine system has been developed that allows (a) teleconsultation between local primary health care providers (primary care physicians and general radiologists) and remote imaging subspecialists and (b) active patient participation related to his or her medical condition and patient education. The initial stage of system development was a traditional teleradiology consultation service between general radiologists and specialists; this established system was expanded to include primary care physicians and patients. The system was developed by using a well-defined process model, resulting in three integrated modules: a patient module, a primary health care provider module, and a specialist module. A middle agent layer enables tailoring and customization of the modules for each specific user type. Implementation by using Java and the Common Object Request Broker Architecture standard facilitates platform independence and interoperability. The system supports (a) teleconsultation between a local primary health care provider and an imaging subspecialist regardless of geographic location and (b) patient education and online scheduling. The developed system can potentially form a foundation for an enterprise-wide health care delivery system. In such a system, the role of radiologist specialists is enhanced from that of a diagnostician to the management of a patient's process of care.

Initial experiences with building a health care infrastructure based on Java and object-oriented database technology.

A multi-tiered telemedicine system based on Java and object-oriented database technology has yielded a number of practical insights and experiences on their effectiveness and suitability as implementation bases for a health care infrastructure. The advantages and drawbacks to their use, as seen within the context of the telemedicine system's development, are discussed. Overall, these technologies deliver on their early promise, with a few remaining issues that are due primarily to their relative newness.

Process models for telehealth: an industrial approach to quality management of distant medical practice.

Process modeling is explored as an approach for prospectively managing the quality of a telemedicine/telehealth service. This kind of prospective quality management is more appropriate for dynamic health care environments compared to traditional quality assurance programs. A vector model approach has also been developed to match a process model to the needs of a particular site.

Creating and indexing teaching files from free-text patient reports.

Teaching files based on real patient data can enhance the education of students, staff and other colleagues. Although information retrieval system can index free-text documents using keywords, these systems do not work well where content bearing terms (e.g., anatomy descriptions) frequently appears. This paper describes a system that uses multi-word indexing terms to provide access to free-text patient reports. The utilization of multi-word indexing allows better modeling of the content of medical reports, thus improving retrieval performance. The method used to select indexing terms as well as early evaluation of retrieval performance is discussed.

A statistical natural language processor for medical reports.

Statistical natural language processors have been the focus of much research during the past decade. The main advantage of such an approach over grammatical rule-based approaches is its scalability to new domains. We present a statistical NLP for the domain of radiology and report on methods of knowledge acquisition, parsing, semantic interpretation, and evaluation. Preliminary performance data are given. A discussion of the perceived benefit, limitations and future work is presented.

Hyperad: augmenting and visualizing free text radiology reports.

Hyperad is an automated computer system designed to extract key concepts from thoracic radiology reports and give physicians access to a large database containing the reports and key concepts. The concepts are extracted from textual documents with natural language processing techniques, then stored with the original documents in the database, which can be queried in terms of findings or associated attributes from an intuitive and easily accessible interface. The extracted concepts are represented both textually in a coded hypertext format and graphically on a coronal cross-sectional anatomy atlas, an idealized graphical model of human anatomy. To facilitate implementation, the communication protocols and standards of the World Wide Web (Web) were adopted. The reports and associated forms are encoded in standard hypertext markup language, which makes it possible to use hypermedia links to navigate the Hyperad database with any graphical Web browser. In the future, Hyperad may prove useful for other applications.

A multimedia database system for thermal ablation therapy of brain tumors.

A prototype multimedia medical database is described for supporting thermal ablation therapy of brain tumors. Its design is motivated by the major need to manage and access multimedia information on the progress and reaction of tumors to various therapy protocols. The database links images to patient data in a way that permits the use to view and query medical information using alphanumeric, temporal, and feature-based predicates. Visualization programs permit the user to view or annotate the query results in various ways. These results support the wide variety of data types and presentation methods required by neuroradiologists to manage thermal ablation therapy data. The database satisfactorily meets the requirements defined by thermal ablation therapy. A similar approach is being undertaken for supporting different therapies of other types of tumors, thus showing the generality of our approach.

A unified timeline model and user interface for multimedia medical databases.

A multimedia medical database model and prototype is described for supporting a timeline-based presentation of information. The database links image and text data in a way that permits users to look at medical information in a single unified view. Various visualization programs permit the user to view data in various ways, including full image views, graphs, and tables. Our technology is applied for proof-of-concept to two areas: thoracic oncology and thermal tumor ablation therapy of the brain. This effort is part of the multidisciplinary KMeD project in collaboration with medical research and clinical treatment projects at UCLA.

Integrated multimedia timeline of medical images and data for thoracic oncology patients.

A prototype multimedia medical database has been developed to provide image and textual data for thoracic oncology patients undergoing treatment of advanced malignancies. The database integrates image data from the hospital picture archiving and communication system with textual reports from the radiology information system, alphanumeric data contained in the hospital information system, and other electronic medical data. The database presents information in a timeline format and also contains visualization programs that permit the user to view and annotate radiographic measurements in tabular or graphic form. The database provides an efficient and intuitive display of the changing status of oncology patients. The ability to integrate, manage, and access relevant multimedia information may substantially enhance communication among distributed multidisciplinary health care providers and may ensure greater consistency and completeness of patient-related data.

Architectural design and tools to support the transparent access to hospital information systems, radiology information systems, and picture archiving and communication systems.

The fragmentation of the electronic patient record among hospital information systems (HIS), radiology information systems (RIS), and picture archiving and communication systems (PACS) makes the viewing of the complete medical patient record inconvenient. The purpose of this report is to describe the system architecture, development tools, and implementation issues related to providing transparent access to HIS, RIS, and PACS information. A client-mediator-server architecture was implemented to facilitate the gathering and visualization of electronic medical records from these independent heterogeneous information systems. The architecture features intelligent data access agents, run-time determination of data access strategies, and an active patient cache. The development and management of the agents were facilitated by data integration CASE (computer-assisted software engineering) tools. HIS, RIS, and PACS data access and translation agents were successfully developed. All pathology, radiology, medical, laboratory, admissions, and radiology reports for a patient are available for review from a single integrated workstation interface. A data caching system provides fast access to active patient data. New network architectures are evolving that support the integration of heterogeneous software subsystems. Commercial tools are available to assist in the integration procedure.

A concept-based retrieval system for thoracic radiology.

Current digital information systems in radiology are insufficient to accommodate the retrieval needs of academicians. Significant efforts are required in retrieving clinical cases for teaching and research. We describe a prototype system that supports intelligent case retrieval based on a combined specification of patient demographics, radiologic findings, and pathologic diagnoses. The documents for these cases can be distributed among multiple heterogeneous data bases. The system features automatic indexing of radiology and pathology reports, a comprehensive lexicon for thoracic radiology, an interface to a hospital information system, radiology information system, and picture archiving and communication systems, and a graphical user interface for query formulation and results visualization. The prototype system was developed within the domain of thoracic radiology involving patients with lung cancer.

Technical considerations in planning a distributed teleradiology system.

A large-scale teleradiology project is under way to link a Florida imaging center to the UCLA Department of Radiology. The initial goal is to provide Florida patients in a routine clinical practice environment with subspecialty consultation by academic radiologists. The plan then calls for the addition of other domestic and international sites. Technical issues in planning to establish the necessary teleradiology infrastructure include wide area network design, image compression, distributed archiving, and special viewing station features. Special emphasis is placed on archive design that makes intelligent use of information, such as triggering events from the radiological information system (RIS) for image prefetching and visual cues from photo-icons for full-size image retrieval. Concepts such as teleconsultation and remote procedure monitoring are aimed at providing the same level of services at distant sites that would be available in-house. This article highlights the system design parameters that must be considered to engineer a scalable distributed teleradiology system.

Interfacing aspects between the picture archiving communications systems, radiology information systems, and hospital information systems.

Information relevant to radiological applications is commonly managed by several autonomous medical information systems including hospital information systems (HIS), radiological information systems (RIS), and picture archiving and communications systems (PACS). In this report, we explain the need to coordinate these systems and to provide some framework in which they can exchange information. In the first half of this report, we describe the integration of a PACS system into a hospital operation. Next, we present the interfacing methods between the HIS and the RIS, and between the RIS and the PACS. Two methods are further detailed for the communication between the RIS and the PACS (1) the triggered database to database transfer, and (2) the query protocol. The implementation of the first method successfully allows RIS reports, procedure and patient demographic information to be displayed at the request of the user along with the associated images at a PACS workstation. The query protocol allows a PACS to dynamically query RIS information. It will be eventually integrated into the design of a scientific multimedia distributed medical database system built on top of the HIS, the RIS, and the PACS.

An automatic method for enhancing the display of different tissue densities in digital chest radiographs.

Digital chest radiographs are often too bright and/or lack contrast when viewed on a video display. This often occurs in radiographs taken of patients with dense lungs, or when incorrect x-ray exposure techniques or inappropriate image preprocessing operations are performed (eg, by the computed radiography system or laser scanner). This article describes a method to automatically provide brightness and contrast adjustments to selectively enhance either soft or dense tissues. This method reduces viewer interaction and improves displayed image quality. The algorithm analyzes the gray-level histogram of a chest radiograph and determines the breakpoints that separate the region outside the patient (background), the radiographically soft tissues, and the radiographically dense tissues. From these breakpoints, a series of piecewise linear look-up tables (LUTs) is generated to selectively enhance either the soft tissues or the dense tissues. This is performed by: (1) varying the contrast in the patient background to achieve the desired overall brightness, (2) selectively increasing the contrast of the tissue region of interest, and (3) reducing or maintaining the contrast of the remaining region. The resulting LUTs are applied to the original image via video display.

Implementation of a large-scale picture archiving and communication system.

This paper describes the implementation of a large-scale picture archiving and communication system (PACS) in a clinical environment. The system consists of a PACS infrastructure, composed of a PACS controller, a database management system, communication networks, and optical disk archive. It connects to three MR units, four CT scanners, three computed radiography systems, and two laser film digitizers. Seven display stations are on line 24 h/day, 7 days/wk in genitourinary radiology (2K), pediatric radiology in-patient (1K and 2K) and outpatient (2K), neuroradiology (2K), pediatric ICU (1K), coronary care unit (1K), and one laser film printing station. The PACS is integrated with the hospital information system and the radiology information system. The system has been in operation since February 1992. We have integrated this PACS as a clinical component in daily radiology practice. It archives an average of 2.0-gigabyte image data per workday. A 3-mo system performance of various components are tabulated. The deployment of this large-scale PACS signifies a milestone in our PACS research and development effort. Radiologists, fellows, residents, and clinicians use it for case review, conferences, and occasionally for primary diagnosis. With this large-scale PACS in place, it will allow us to investigate the two critical issues raised when PACS research first started 10 yrs ago: system performance and cost effectiveness between a digital-based and a film-based system.