Disease recurrence STATUS DIAGNOSIS OF BREAST CANCER USING FUZZY NON-STATIONARY

Disease recurrence STATUS DIAGNOSIS OF BREAST CANCER USING FUZZY NON-STATIONARY

ABSTRACT: Breast cancer is never spontaneously completely cured. 20-30% breast cancer cases apparently recurred. Risk status of breast cancer recurrence diagnosis plays important role in recommending optimal treatment. Beside uncertainty, clinicians exhibit inter-expert and intra-expert variability in making a difficult decision. Variation may occur among the decisions of a panel of human experts (inter-expert variability), as well as in the decisions of an individual expert over time (intra-expert variability) based on new evidences provided by research efforts, latest fashion, or emotional state. Fuzzy model covers uncertainty. Fuzzy model with non-stationary fuzzy sets (FNIS) replicates human variability covers experts variability both inter and intra. This research aims to diagnose risk status of breast cancer recurrence based on assessment of six clinical variables comprises her2/neu, hormone receptor, age, tumor grade, tumor size, and lymph node. Fuzzy non-stationary inference system in this research using singleton fuzzification, product inference, and center average defuzzifier. Two perturbation functions to perturb underlying membership functions with 5 iterations are normally distributed random function and sinusoidal function. Normally distributed random function for input variables, Ï� for each input in each iteration are random numbers in range [1,5]. Sinusoidal function for output variable with α = 33, number constants k for each variables in the system are determined. The system has been tested on 20 new test data sets and 19 of 20 are found correct.