Abstarct: In recent decades, genetics particularly molecular genetics has been widely developed. The advent of microarray technology as a powerful tool for simultaneous studying and analyzing of the thousands of genes behavior is one of the developments. Images of microarray technology play an important role in the detection, diagnosis and treatment of diseases and analysis of the images has a direct impact on results. But systems to analyze these images are only able to quantify the images (to extract raw data from the image) of microarray and are generally operated manually. This project aims to provide an automated system to extract and analyze microarray images data in order to detect cancer diseases. The proposed system consists of three main phases of image processing, data mining, and disease diagnosis. The image processing includes operations such as noise cancelling, rotation fixing, the gene locating, background deleting and raw data extracting from the image. Data mining involves normalization of the extracted data and selection of the more effective genes. In the third phase, according to the extracted data, identification and diagnosis of the disease are carried out. In this study, microarray images of breast cancer, leukemia and lymphatic cancer developed by Stanford University are used to evaluate and test the proposed system. The test results show that the system accuracy at genes locating phase on the microarray images is more than 98%. In addition, the system is capable to detect the cancer type in the breast cancer data baxse, lymphatic cancer and leukemia with an accuracy of 45.95%, 100% and 11.94%, respectively.