Combination of radiologists´ diagnosis by a fuzzy number approach

The authors present a fuzzy set approach to combine experts´ imprecise opinions. The case study deals with medical radiologists´ diagnoses expressed as fuzzy numbers that reflect the confidence and uncertainty of their decision. Specifically, the independent, fuzzy diagnoses of four radiologists are combined by a fuzzy number approach, called a mixed linear extension. The combined result is a fuzzy number. Four essential properties of this combination are discussed, namely, agreement preservation, order independence, possibility conservation and possibility interval preservation. Defuzzification yields crisp results which coincide well with the actual observations representing 45 true values