Fuzzy Bayesian Network-Based Inference in Predicting Astrocytoma Malignant Degree

This study proposes an improved fuzzy Bayesian network (FBN), which integrates fuzzy theory into Bayesian networks (BN) by introducing conditional Gaussian models to make a fuzzy procedure. This particular procedure will transform continuous variables into discrete ones when dealing with continuous inputs with probabilistic and uncertain nature. Moreover, it describes fuzzy features better than other methods. To validate our method, this paper applied the fuzzy Bayesian network to classification of astrocytoma malignant degree. We present a probabilistic model that employs FBN in fusing both continuous low-level features and discrete high-level semantics from MRI (magnetic resonance imaging). It realizes quantificational analysis in predicting astrocytoma malignant level and provides a novel assistant way for young doctors. An accuracy of 81.67% was achieved out of 60 test samples, which satisfies the basic requirement of neuroradiologists.