THE APPLICATION OF A FUZZY REASONING IN RADIOLOGICAL DIAGNOSIS

This study represents an attempt to clarify the vagueness brought about in X-ray dia-gnosis, by means of questionnaires. This study is also designed to explore appropriate countermeasures for coping with the phenomenon of Fuzzy reasoning used in the complicated X-ray diagnostic outcome. The case mat...

Full description

Saved in:
Bibliographic Details
Published inShika Hoshasen Vol. 21; no. 2; pp. 84 - 103
Main Author NISHIOKA, Toshio
Format Journal Article
LanguageJapanese
Published Japanese Society for Oral and Maxillofacial Radiology 1981
Online AccessGet full text

Cover

Loading…
More Information
Summary:This study represents an attempt to clarify the vagueness brought about in X-ray dia-gnosis, by means of questionnaires. This study is also designed to explore appropriate countermeasures for coping with the phenomenon of Fuzzy reasoning used in the complicated X-ray diagnostic outcome. The case material comprised patients examined at Nihon University Dental Hospital during the year of 1979. Thirty dentists on the teaching staff of Nihon University School of Dentistry and 100 dental students in clinical training participated for collaboration in the study. Each subject was shown a radiogram or a set of radiograms and asked to read a specified region so that his X-ray findings and diagnosis were entered by himself in a questionnaire sheet. Each film or set of films were accompanied by pertinent clinical data from the assigned case which included the past history, present history and present illness. Calculation of the membership function set up for proper evaluation of the answered diagnostic X-ray findings was achieved by regressive principal component analysis, on the grounds of previously summarized clinical and X-ray finding data from 101 cases at this hospital. Namely, principal components representing overall characteristic values for establishing the diagnosis were derived from the groups of clinical findings, whereupon a multiple regression model was estimated for each group of X-ray findings, assigned as objective variable, employing the extracted principal components as explanatory variables. The reasoning was carried out as modus ponens linking the answered diagnostic findings to the membership function, with a connective: a) An X-ray finding (yj) is true and that the corresponding clinical finding (xi) is also true. b) If the clinical finding is true, then the X-ray diagnosis of the case is (dk). c) Therefore, the X-ray diagnosis of the case is (dk) as far as the X-ray finding (yj) is true. Although plurality of diagnosis was obtained for each according to the data collected in questionnaires, the diagnosis inferred on the basis of Fuzzy's theory coincided with the established diagnosis. The variations seen in the results of judgement to a stimulus may indicate that there were inter-subject differences in criteria for judgement. While it is true that insufficiency in the amount of information and ambiguity in the wording of the serial questions on the questionnaire has to be taken into account, the results of the present study seem to indicate that a possible involvement of institution and subjectivity peculiar to humans mainly gave rise to uncertainty in the judgement. Introduction of the Fuzzy's theory would be most advisable in dealing with such human subjective values that cannot be treated with the theory of probability.
ISSN:0389-9705
2185-6311
DOI:10.11242/dentalradiology1960.21.84