Non-invasive Trisomy 21 Diagnosis Using Fuzzy Cognitive Maps

Abstract

A fuzzy Diagnostic Decision Support System (DDSS) has been developed, exploiting the possibilities of a specific algorithm which transforms a crisp dataset into fuzzy. The purpose of the developed fuzzy DDSS is the diagnosis of fetuses with Trisomy 21 (T21). The main reason behind the selection of this diagnosis problem is the need of sonographers to escape the preconditions of having specialized screening systems and being highly precise during prenatal screening tests. With the help of an intelligent fuzzy system, a sonographer can input the states of the prenatal screening tests in linguistic terms (which are based on his experience and knowledge) and get a diagnosis result. Fuzzy Cognitive Map (FCM) has been chosen to be the model of the fuzzy DDSS model for this diagnosis problem. Essentially, FCM is a network connecting fuzzy concepts with edges resembling their cause-effect interrelations. The proposed model was built based on a dataset of 50900 pregnancy cases given by the Fetal Medicine Foundation. Since most of the dataset’s parameters were in crisp form a certain algorithm was used for the fuzzification for a part of the dataset. The resulted fuzzified dataset was used to train and test the proposed model. The trained model was tested on an unknown testing dataset, achieving scores of 74% and 86% for sensitivity and specificity correspondingly. The results of this work are presented and discussed at the end of this paper.