Abstarct: The development of Takayasu, one of the most dangerous vessels inflammation disease, may causes risky effects such as death. The well-timed treatment and fast diagnosis help us to control it well. The physicians can also avoid the whole destruction of vessels if it is detected perfectly at earlier stages. This thesis proposes an approach that uses Artificial Neural Network (ANN) with Levenberg-Marquardt (LM) training method along with Genetic Algorithm. First, the dataset is analysed to choose the most effective parameters and they are then normalized using z-core technique. Second, we propose a multilxayer perceptron (MLP) structure and a GA-learned one to classify persons into different categories: “patient”, “normal”, and “not detected”. Finally, their performances are computed using different measures. The results show that the proposed structures can detect Takayasu effectively. In addition, the the GA-trained MLP returns higher accuracy than the LM-trained one.