Fuzzy inference system for endocardial edge detection

An approach for left ventricular endocardial edge detection based on fuzzy IF-THEN inference rules is proposed in this paper. The echocardiographic image is preprocessed to enhance the contrast and smoothness. Four inputs are given to fuzzy inference system from 3 × 3 spatial window; each input has five linguistic variables named VLOW, LOW, MED, HIGH and VHIGH. The fuzzy set and rules are derived heuristically to detect the edges even in low contrast region. The approach is successfully applied to echocardiographic images to detect the LV endocardial edges. The IF-THEN rules can be easily changed to adjust the edge thickness and it has a flexible structure to be adjusted. Some experimental results are given and compared to conventional edge detectors.