Fuzzy system for analysis of left ventricular ejection fraction in SPECT imaging

Background and Objectives: This study employed fuzzy logic to extract information and analyze nuclear medicine imaging. Data was extracted for accurate and fast calculation of left ventricular ejection fraction (LVEF), that is so competent in diagnosis of left ventricular disease. A novel system for predicting the intended parameter is generated by creating a desired relationship between input and output.Materials and Methods: The proposed method benefits from ۱۶ frame gated SPECT images. Input and output values from acquired data were extracted by an expert and related membership functions were defined. Fuzzy inference system(FIS) is a main tool in this procedure. Inputs were divided in ۵ types and a unique output was observed at the end. Inputs include age, BMI, heart rate, End diastolic volume, end systolic volume and LVEF is the output. In order to determine the form of membership functions of each variable which leads to generate the behavior of the system, a membership function editor was employed. According to the defined data, we used MIN as AND method, MAX as OR method, and CENTROID as a defuzzifier method in MAMDANI type.Findings: Items are dynamically linked to each other and varying in FIS leads to changes in other details of the system and it was observed that there is no limitation in the number of inputs.A rule matrix that was designed by an expert described the structure of this network and demonstrated the most effective item in output is BMI. Conclusions: In fuzzy structures all parameters are considered relative and not absolute. Following this network in SPECT, it was concluded that each of the parameters make a distinct contribution in output specially in this study, the highest one which is more important for cardiovascular assessment is BMI and should be noted in nuclear medicine imaging.

کلیدواژه ها:

Fuzzy logic ، Nuclear medicine ، SPECT ، cardiac imaging