This paper presents a methodology for extraction knowledge of Cardiac Single Proton Emission Computed Tomography (SPECT) diagnosis with the use of hybrid techniques represented by soft computing to classify patterns from SPECT database. In this paper the searching capability of a Genetic Algorithm (GA) has been exploited for automatically evolving the structure of neural network as well as proper parameters of neural network. This paper concerns with extraction knowledge of the dataset describes diagnosing of cardiac SPECT images, each of the patients is classified into two categories: normal and abnormal. The GA is considered to face contemporaneously the optimization of the design of neural network architecture and the choice of the best learning method. After that, supervised classification algorithm (Kohonen winner-take-all network) determines the class under which each feature vector belongs to was used. At the last stage, (IF-Then rule) to form several rules that govern each class attributes were used. The proposed methodology achieved 95% accuracy and provides fast and adaptive learning for extraction knowledge.