IDEM: a Web application of case-based reasoning in histopathology

https://doi.org/10.1016/S0010-4825(98)00028-6Get rights and content

Abstract

Different software engineering and artificial intelligence methods can be used to design Internet retrieval of prototypical medical images. We used the case-based reasoning (CBR) approach to provide an `intelligent' access to a collection of illustrated medical cases through the Internet. This paper presents a Web interface for the CBR system IDEM (image and diagnosis from examples in medicine) in the domain of breast pathology. Thanks to the definition of a similarity measure between the descriptions of cases we propose a flexible querying of the case-base and a quantitative browsing among cases through similarity links. The resemblance rates provided by the system argue for the quality and the relevancy of the retrieved data. The flexibility of the querying process is robust to missing information and could be adapted to a daily practice. The CBR approach is a promising method for a clinical relevant and an efficient retrieval of reference images and diagnosis clues through Internet.

Introduction

Healthcare professionals use more and more Internet to retrieve knowledge reference sources in terms of decision making and education. Many researches have particularly focused on the design of Internet browsers and query systems for prototypical medical image retrieval1, 2, 3. Important issues are the correctness and relevance of the accessed medical images, as well as the interactive exploitation of these data for decision-making support.

One usual way to access image databases on Internet is by filling hyper text markup language (HTML) forms that generate, through special common gateway interface, a search process4, 5, 6, 7. Some institutions give to users a direct access to the underlying databases, using the native tools of the database management system, while others have created specialized query languages and user interfaces intended to better match the needs of healthcare users. Some institutions associate a browsing process to the database querying to enhance the efficiency of the server. So, for instance, the querying paradigm can be used reiteratively to create a browsing path through a report database[8].

The advantages of adopting Web tools for answering clinical questions are limited by the difficulty in learning, for each information resource, how to issue queries and for what results9, 10. Particularly, in image databases querying, the retrieval process depends on the representation of the image in the base and can sometimes provide a content based access to medical image archives or even to retrieve images similar to a query image[11]. Different software engineering and artificial intelligence methods can be used to define the best strategy of querying and/or browsing medical images archives.

The case-based reasoning (CBR) approach is one of these methods. It uses the specific knowledge of previously experienced concrete situations (cases). A new problem is solved by finding a similar case and re-using it in the new problem situation. From a theoretical point of view, CBR derives from analogical research[12]and can be considered as a form of intradomain analogy[13]. Applicability of CBR methodology is interesting when the knowledge can be represented by cases in the domain. It is particularly well suited to the medical domain and even more in weak-theory domains of medicine[14], such as histopathology.

The objective of the present work is to develop an Internet service site, allowing the exploration of a collection of illustrated medical cases in the domain of histopathology. Our main contention is that the CBR approach enriches both the querying and browsing processes and so enhances the opportunities presented by Internet to make computer-based medical experience more accessible, more usable in daily practice.

In this article, we describe the IDEM (images and diagnosis from examples in medicine) Internet site. The browsing and querying processes involved in the information retrieval are both supported by the conceptual architecture of the IDEM CBR system.

One main concern of the on-going project is to guarantee the quality of both the retrieved data and of the retrieval process i.e. its efficiency in providing decision support functionalities. For this, the user should easily control the output of the CBR system. For example, he/she could reiteratively modify the query or browse into a set of cases through similarity links.

Section snippets

Applicability of CBR techniques to histopathology

Diagnosis and prognosis decision support systems in histopathology rely on a wide variety of methods. Some of them are based on a bayesian network like Pathfinder15, 16, 17, while in others, like CancerStage[18], the expert knowledge is embedded in if-then-else rules. As a matter of fact, histopathologic imagery can hardly be modelled in the format of logical knowledge representation. The lack of statistical data and the enormous effort, which would be necessary to assert such data, impede the

Acknowledgements

The authors are grateful for valuable discussions with Dr. J.M. Guinebretière and Dr. G. Contesso, Department of Histopathology, Institut Gustave Roussy, France.

Christel Le Bozec received a M.D. in histopathology at the Paris V University and is a doctoral student in medical informatics. Her thesis deals with applying the case-based reasoning approach in developing decision making systems for medical imagery and especially in the field of histopathology. lebozec@hbroussaisfr

References (29)

  • D.E. Heckerman et al.

    An evaluation of the diagnostic accuracy of Pathfinder: 1992

    Comput. Biomed. Res.

    (1992)
  • L. Costaridou et al.

    Pronet services for distance learning in mammographic image processing

    Med. Inf. Eur.

    (1997)
  • N. Silberzahn et al.

    A new object request broker architecture in implementing a medical picture retrieval system

    Med. Inf. Eur.

    (1997)
  • C. Courtin et al.

    WWW cooperative multimedia interface in medecine

    Med. Inf. Eur.

    (1997)
  • Fernandez et al.

    ARMEDA: accessing remote medical databases over the WWW

    Med. Inf. Eur.

    (1997)
  • D. Pollard, E. Kucharz, W.E. Hammond, Implications of the Java language on computer-based patient records, Proc. Amia....
  • H.J. Lowe et al.

    The World Wide Web: a review of an emerging internet-based technology for the distribution of biomedical information

    J. Am. Med. Inform. Assoc.

    (1996)
  • U. Tachinardi et al.

    Hypermedia patient data retrieval and presentation through

    Proc. SCAMC

    (1995)
  • F. Consorti et al.

    Browsing and querying multimedia report collections

    Med. Inf. Eur.

    (1997)
  • D.F. Sittig, G.J. Kuperman, J.M. Teich, WWW-based interfaces to clinical information systems: the state of the art....
  • W.M. Detmer et al.

    A model of clinical query management that supports integration of biomedical information over the World Wide Web

    Proc. SCAMC

    (1995)
  • Zabulis et al.

    The 12Cnet service architecture paradigm

    Med. Inf. Eur.

    (1997)
  • C. Riesbeck, R. Schank, Inside Case-Based Reasoning, Hillsdale, NJ,...
  • Aamodt

    Case-based reasoning: fundational issues, methodological variations and system approaches

    AICOM

    (1994)
  • Cited by (6)

    • Semantics and CBIR: A medical imaging perspective

      2008, CIVR 2008 - Proceedings of the International Conference on Content-based Image and Video Retrieval
    • Medical imagery in case-based reasoning

      2008, Studies in Computational Intelligence
    • User-adapted clinical decision support systems

      2005, AAAI Spring Symposium - Technical Report
    • MedlSeek: A web based diffusion system for medical visual information

      2003, Proceedings of the Interntational Workshop on Web Information and Data Management

    Christel Le Bozec received a M.D. in histopathology at the Paris V University and is a doctoral student in medical informatics. Her thesis deals with applying the case-based reasoning approach in developing decision making systems for medical imagery and especially in the field of histopathology. lebozec@hbroussaisfr

    Marie-Christine Jaulent received the Ph.D. in computer science from the I.N.P.T. in 1986. Until 1988, she was a postdoctor at the `Intelligent Systems Laboratory' at Boston University, MA. Since 1990, she has been with INSERM (Institut National de la Santé et de la Recherche Médicale) as a full-time researcher in medical images interpretation. Her current fields of interest are in high level image interpretation and in application of fuzzy models for knowledge representation and medical reasoning from images.

    Eric Zapletal received his degree of engineer in computer science at the Institut d'Informatique d'Entreprise (Evry, France). He worked at the Medical Informatics Department of the Broussais hospital (Paris) as a C++ developer in several European projects (AIM-HELIOS, ESSI-NOMAD, …). He was also involved in Broussais proprietary developments such as X/MOTIF interface designing for a hypertension-turned application (ARTEMIS) and middleware tools. He is now in charge of the Laboratory Information System of the new Georges Pompidou hospital (Paris).

    Didier Heudes is M.D. He is hospital practitioner in the pathology laboratory and assistant in medical informatics at the Broussais Hospital (Paris). He has particularly focused his interests on image computer-assisted analysis applied on histopathology.

    Patrice Degoulet received his M.D. from the Broussais-Hotel-Dieu Hospital in Paris, France in 1977 and his Ph.D. degree from the University of Paris, in 1984. He is actually Professor in Broussais-Hotel-Dieu school of Medicine and the head of the Medical Informatics Department of the Broussais Hospital in Paris. His current fields of interest are in conceptual modelling in medicine, data and knowledge base management, hospital information systems and medical record management.

    View full text