Voice-activated retrieval of mammography reference images.

We undertook this project to integrate context sensitive computer-based educational and decision making aids into the film interpretation and reporting process, and to determine the clinical utility of this method as a guide for further system development. An image database of 347 digital mammography images was assembled and image features were coded. An interface was developed to a computerized speech recognition radiology reporting system which was modified to translate reported findings into database search terms. These observations were used to formulate database search strategies which not only retrieved similar cases from the image database, but also other cases that were related to the index case in different ways. The search results were organized into image sets intended to address common questions that arise during image interpretation. An evaluation of the clinical utility of this method was performed as a guide for further system development. We found that voice dictation of prototypical mammographic cases resulted in automatic retrieval of reference images. The retrieved images were organized into sets matching findings, diagnostic hypotheses, diagnosis, spectrum of findings or diagnoses, closest match to dictated case, or user specified parameters. Two mammographers graded the clinical utility of each form of system output. We concluded that case specific and problem specific image sets may be automatically generated from spoken case dictation. A potentially large number of retrieved images may be divided into subsets which anticipate common clinical problems. This automatic method of context sensitive image retrieval may provide a "continuous" form of education integrated into routine case interpretation.

Telemedicine: delivering medical expertise across the state and around the world.

Many aspects of medical care can now be delivered at a distance using telemedicine technology. Rapid video and computer-based communication of medical information makes it possible for a physician to "examine" a patient located in another city, to view highly detailed medical images, to consult with distant subspecialists, or to supervise complex medical procedures. This same technology can bring scattered health-care workers together for joint teaching conferences. The Yale Telemedicine Center has initiated a number of such programs ranging from providing consultations in real time to physicians in Saudi Arabia, to interpreting medical images across town or across the state. Telemedicine will become a powerful tool for managed health-care organizations which are responsible for the medical needs of widely distributed patients in a vertically integrated health-care delivery system. This paper reviews the evolution of telemedicine, its technical fundamentals, specific medical applications, and the activities of the Yale Telemedicine Center. Evolving uses for telemedicine in Connecticut are described.

The prospect of expert system-based cognitive support as a by-product of image acquisition and reporting.

In order for computer-based decision-support tools to find routine use in the everyday practice of clinical radiology, further development of user interface and knowledge content are required. In an ideal interface, the interaction between the radiologist and the computer would be minimized and painlessly integrated into existing work patterns. In this article, we explore some of the ways that pre-existing computer interactions in the processes of image acquisition and reporting can be used to feed case information into an expert system and thereby allow users to acquire advice from it in an automatic fashion. We describe interface models that we have developed in the domains of mammography and obstetric ultrasound, and discuss interface and content-related questions that have arisen from informal evaluations of these systems. In particular, the need for clinical outcome-relevant decision support and training level-appropriate decision support are discussed in detail.

Knowledge-based radiologic image retrieval using axes of clinical relevance.

This paper describes an approach to computer-based intelligent retrieval of feature-coded radiographic images relevant to a specific case being evaluated. The approach involves partitioning the search space along clinically natural groups of attributes which we call "axes of clinical relevance." By embedding knowledge about the domain to help direct the search process, a clinician's needs may be met more comprehensively. Domain knowledge, supplied to the system as "axis heuristics," may make search more robust. These heuristics provide a graded, progressive relaxation of the search constraints. This approach helps show the user groups of images in order of probable relevance to a current case. AXON is a prototype knowledge-based system constructed to illustrate this approach in the domain of chest imaging. This paper describes the AXON system, demonstrates some searches which illustrate the potential utility of this approach, and discusses preliminary tests of the search strategies used by AXON.

Exploring subjective vs. objective issues in the validation of computer-based critiquing advice.

Evaluation is an important part of the development of computer-based medical expert systems. Such evaluation may be particularly difficult when judging a critiquing system which responds to a proposed management strategy with a discussion of the advisability of that approach. DxCON is an expert system which produces a prose critique discussing the radiologic workup of obstructive jaundice. This paper describes DxCON, and its experimental validation by three independent judges. A central component of the validation involved allowing the judges to react to the system's advice in a quite flexible, unstructured fashion. This project provides a case study of how subjective issues impact both the design and implementation of a validation of a medical expert system whose output is explanatory prose.

Expert system-controlled image display.

Conventional computer-based medical expert systems deliver advice to physicians as written text. While such advice is useful, it has distinct limitations in a visually oriented discipline such as diagnostic radiology, in which decisions often depend on pattern recognition and appreciation of subtle morphologic features. The authors developed a prototype expert computer system, IMAGE/ICON, which displays groups of images sorted into a series of axes based on different ways in which they may be similar. They may share a common feature, group of features, causes, or clinical setting. IMAGE/ICON may display examples of morphologic variations of a dominant finding or a spectrum of abnormalities seen in an specific disease or group of diseases. The system also assembles a written analysis of key features of a case. Such a tool may be useful as a diagnostic aid or for continuing medical education. It is likely to have particular impact in the form of an intelligent radiologic workstation, as picture archiving and communication systems become available.

Optimizing radiologic workup: an artificial intelligence approach.

The increasing complexity of diagnostic imaging is presenting an ever expanding variety of radiologic test options to clinicians. As a result, it is becoming more difficult for referring physicians to select an appropriate sequence of tests. The current economic pressures on medicine make it particularly important that resources be used judiciously. Radiologic workup often involves a sequence of tests that lead from presenting signs and symptoms to a definitive diagnosis or intervention. This sequence ideally begins with simple, inexpensive, safe, non-invasive tests and progresses to more complex, expensive, and hazardous tests only if the simpler tests are insufficient to establish a diagnosis. DxCON is a developmental artificial intelligence-based computer system that gives advice to physicians about the optimum sequencing of radiologic tests. DxCON evaluates basic clinical information and a physician's proposed workup plan. The system then creates an analysis of the strengths and weaknesses of his plan. The domain chosen to explore computer-based workup advice is the radiologic workup of obstructive jaundice.

Hypermedia in radiology: computer-assisted education.

Hypertext is a new computer-based method of presenting information that provides greater flexibility than conventional methods of continuing education. With a hypertext system, an individual using the computer can acquire more information on a word or concept that needs to be pursued in depth. Hypermedia is an expanded concept which uses the computer's ability to incorporate images, sounds, and video images in addition to text. This interactive, multimedia approach customizes information for more effective learning. The authors devised a prototype hypermedia textbook of nuclear medicine using a personal computer with hypermedia software that contains text, graphs, tables, figures, literature citations, and an easily persuable image database. All the information is organized with multiple cross-references, allowing instant branching to relevant facts, in different levels of detail. The system's applications and the ease of expansion or modification by the user are described.

ICON: a computer-based approach to differential diagnosis in radiology.

ICON is a computer-based expert system being developed to help radiologists with the process of differential diagnosis. ICON focuses on the domain of lung disease as seen on a chest radiograph in patients with lymphoproliferative disorders. Rather than attempt to tell the radiologist what the diagnosis is, the system asks him or her to propose a diagnosis and then discusses the evidence supporting that diagnosis or competing diagnoses. The system's output is in the form of an English prose critique. This approach combines the computer's ability to recall detailed information with human reasoning skills

HYDRA: a knowledge acquisition tool for expert systems that critique medical workup.

HYDRA is a computer-based knowledge acquisition tool under development to assist in the creation of expert systems which critique medical workup. To use HYDRA, a domain expert first outlines the recommended approaches to the workup of a chosen medical problem, using the Augmented Transition Network formalism. From this model, HYDRA produces a list of the various conditions for which critiquing comments may be required to react to all possible approaches that might be proposed by the user of the critiquing system. Domain-specific constraints can be used to restrict the number of conditions suggested. In this way, HYDRA assists the domain expert by providing a model for structuring the problem, and by breaking down the domain expert's work into a set of small, easily understood tasks.

Critiquing the process of radiologic differential diagnosis.

ICON is a developmental expert system designed to critique the process of radiologic differential diagnosis. To use ICON, a physician outlines (1) findings observed in a chest radiograph, (2) a small amount of clinical information describing the patient, and (3) a proposed diagnosis. ICON critiques the appropriateness of that diagnosis in detail, analyzing why and how well the findings serve to confirm it, or to rule it out. ICON may also suggest further information to look for. ICON explores the design issues involved in critiquing the process of differential diagnosis, and is currently implemented in a limited domain: the radiographic diagnosis of a lung mass in a patient with Hodgkin's disease.

Acute bilobar collapse secondary to endobronchial metastatic seminoma.

We report here a patient with tracheal and endobronchial metastatic seminoma who presented with acute respiratory failure with right upper and middle lobe collapse. Diagnosis of the etiology of the atelectasis was made by fiberoptic bronchoscopy. Radiation therapy caused a marked reversal of the atelectasis.

Thoracic alterations after cardiac surgery.

A series of 92 adult patients undergoing elective cardiac surgery was reviewed to distinguish routine postoperative radiographic alterations from signs of clinically significant complications. Two postoperative complications required decisive clinical intervention: mediastinal hemorrhage (7% of cases) and sternal wound infections (3% of cases). Mediastinal hemorrhage was most often diagnosed by excessive bloody mediastinal tube drainage alone, although progressive mediastinal widening and pleural or apical extrapleural hematomas provided corroborating or, rarely, the sole evidence of mediastinal hemorrhage. Sternal wound infections were most often diagnosed clinically, but increasing pre- and retrosternal gas collections provided radiographic confirmation. Atelectasis was the most common postoperative finding. There were many abnormal gas and soft-tissue collections posteroperatively that were notable for their lack of clinical importance. Serial postoperative films were necessary to demonstrate the progression of radiographic findings which indicate the two important postoperative complications.

Thoracic manifestations of liver disease.

The close proximity of the liver to the right lung may be the basis for thoracic changes that occur in the presence of lung disease, or such changes may occur as part of the natural course of the disease. Whatever the etiology, the selective use of plain film radiography, fluoroscopy, radionuclide scanning, gallium scanning, and computed tomography will facilitate diagnosis and management.

Popliteal cysts: presentation as thrombophlebitis.

Synovial-lined cysts in the popliteal space associated with rheumatoid arthritis or meniscal tears may compress the popliteal vein and produce signs and symptoms suggestive of thrombophlebitis. Because of a presumptive diagnosis of thrombophlebitis, venography may be performed initially and demonstrate compression of the popliteal vein by an extrinsic mass. The correct diagnosis of a popliteal cyst can easily be made with B-mode ultrasonography, and, if desired, confirmed by arthrography. Four cases are presented which emphasize these features.