Automatic Detection of Dermatological Diseases Using Adaptive Neuro-Fuzzy Inference Systems Zikrija Avdagić, Lejla Begić-Fazlić, Huse Fatkić Faculty of Electrical Engineering of University of Sarajevo zikrija.avdagic@etf.unsa.ba lejla.begic@fds.ba huse.fatkic@etf.unsa.ba Abstract A new approach based on adaptive neuro-fuzzy inference system (ANFIS) was presented for the detection and recognition of dermatological diseases. The domain contained records of patients with known diagnosis. Given a training set of such records, the ANFIS classiffiers learned how to differentiate a new case in the domain. The five ANFIS classiffiers were used to detect and recognize dermatological diseases when 10 basic features defining skin clases indications were used as inputs. The proposed ANFIS model combined the neural network adaptive capabilities and the fuzzy logic qualitative approach. Some conclusions concerning the impacts of features on the detection of dermatological diseases were obtained through analysis of the ANFIS. The results confirmed that the proposed ANFIS model has some potential in detecting dermatological diseases. The ANFIS model achieved accuracy rates which were higher than that of the stand- alone neural network model. 1. Introduction The differential diagnosis of dermatological diseases is a difficult problem. The diseases observed in this group are Pityriasis Lichenoides, Pityriasis Alba, Lichen Striatus, Lichen Planus and Psoriasis. They all share the clinical features of erythema and scaling with very few differences. This is where fuzzy set theory plays an important role in dealing with uncertainty when making decisions. Fuzzy sets have attracted the growing attention and interest in modern information technology, production technique, decision making, pattern recognition, diagnostics, etc. [1- 4]. Neuro-fuzzy systems are fuzzy systems which use artificial neural networks (ANNs) theory in order to determine their properties (fuzzy sets and fuzzy rules) by processing data samples. A specific approach in neuro-fuzzy development is the adaptive neuro-fuzzy inference system (ANFIS), which has shown significant results in modelling nonlinear functions. In ANFIS, the membership function parameters are extracted from a data set that describes the system behavior. The ANFIS learns features in the data set and adjusts the system parameters according to a given error criterion [2,3]. A new approach based on ANFIS was presented for the detection dermatological erythema diseases. Five ANFIS classifiers were used to detect class of dermatological diseases. Each of the ANFIS classifiers was trained so that they are likely to be more accurate for one type of diseases classes than for the other classes. The predictions of the five ANFIS classifiers were combined in sixth classifiers. Data base investigated in this study consisted of 260 analyzed data cases. The proposed ANFIS model was then evaluated and performances of the ANFIS model, were reported. We were able to achieve significant improvement in accuracy by applying ANFIS model compared to the stand-alone neural networks. Finally, some conclusions were drawn concerning the impacts of features on the detection of tar content in smoke condensate. 2. Sugeno fuzzy model A typical fuzzy rule in a two-input-single- output Sugeno fuzzy mode has the format: If x is 1 A and y is 1 B then ) , ( 1 y x f z  , where 1 A and 1 B are fuzzy sets in the antecedent; ) , ( 1 y x f z  is a crisp function in the consequent. Usually, ) , ( 1 y x f z  is a polynomial in the input variables x and y Figure 1.