Skip to main content
European Commission logo print header

Advanced Environment for Medical Image Interpretation

Objective


3 main research themes in the field of computer aided medical image interpretation have been investigated:
analysis of its role in medical practice, leading to a definition of clinical requirements;
development of a formalism for human computer interaction, leading to the definition of a user model;
exploration of technical options for implementation, leading to the definition of a system architecture.

Both radiological and microscopical image interpretation were considered. The types of investigation which constitute the bulk of the workload in a hospital department were studied, in order to identify potential mainstream applications of computer based methods. In both cases a representative set of images were collected and annotated with a description of the clinically significant features; there formed a basis for considering, critically, the role of image interpretation in current medical practice and for identifying the types of assistance which could be provided by a computer based system.
Those types identified were:
quality control;
prescreening or prompting. In prescreening the system presents to the user only those images which are suspicious. In prompting all images are presented to the user and the locations of potential abnormalities indicated;
intelligent display (eg 3-dimensional structure);
quantitative description;
communication and group working;
staff training and clinical research.

The underlying technology required in each case is very similar and it was therefore decided that the development of a flexible system could be attempted which would cope with any of these forms of support. This required a capability for automated image interpretation and therefore a knowledge based approach was adopted. The main issues of knowledge representation, reasoning and control were addressed by a number of activities, leading to a proposal for a system architecture. The functional requirements were translated into a set of technical requirements, developing a conceptual framework within which system design issues could be addressed.
Many patient management decisions are based on the interpretation of images. The most important examples are macroscopic images, generated in the radiology department, and microscopic images generated in the hospital laboratory. The last few decades have seen the development of new methods of visualization which allow the intricate relationship between structure and function to be explored in ever increasing detail. These provide the basis for better informed clinical decisions but have created a problem in dealing with the quantity and complexity of data. A computer-based system to help the radiologist or microscopist manage the collection and interpretation of image data could improve the clinical effectiveness of image-based decisions and maximize the use of scarce human resources.
An intelligent computer system could, in principle, provide routine support for the clinician by providing sophisticated display facilities for multi-dimensional or multi-modal data ; objective measurements of structure and function ; prescreebing to select cases or individual features for far more detailed assessment ; comparison with reference images ; advice on protocols for further investigation. Previous attempts to address this problem have succeeded in providing machine assistance for particular image interpretation tasks but generally the systems which have been developed either operate with a great deal of autonomy within a very narrow domain of application or they offer flexibility at the expense of very low-level user interface. This has resulted in a great deal of non-reusable technology which is of limited value. The need is for a general purpose system which is sufficiently intelligent that it can work co-operatively with the clinician without the need for detailed, low-level interaction.
The project we propose will build on recent advances in knowledge-based image interpretation and IKBS in general. The project team will comprise medical practitioners, academics and industrial researchers who will form a strong multi-disciplinary team. The objective is not at this stage to develop a working system but rather to concentrate on defining functional objectives for such a system, identifying the key medical and technical issues and assessing technical options. The work will be divided into three main phases :
I. Analysis of requirements
II. Development of a Design Proposal
III. Development of a plan for further action.
The first phase involves an analysis of current practice in radiology and microscopy. A careful consideration of the organization and clinical issues will lead to a proposal for the role of an image-based decision support system and hence to a definition of functional requirements. The design will identify the main structures and processes of the system based, as far as possible, on the integration of existing technology. The final phase will involve defining realistic goals for a pilot system and formulating an action plan.
Main Deliverables :
To define functional requirements for image interpretation in radiology and microscropy, knowledge representation requirements. Assessment of reasoning and control methodologies. Prototype of system architecture for medical image interpretation.

Topic(s)

Data not available

Call for proposal

Data not available

Funding Scheme

Data not available

Coordinator

University of Manchester
EU contribution
No data
Address
Oxford Road
M13 9PL Manchester
United Kingdom

See on map

Total cost
No data

Participants (5)