Lung cancer is one of the major diseases which causes death in developed as well as developing countries. It is still a difficult issue to find the effective medical treatment for curing lung cancer. In the earlier time, the treatment for lung cancer had made efforts to improve the 5-year-survival rate. The X-ray films on the chest and computed tomography (CT) have been used to diagnose the primary stage of lung nodule. Recently, the combination of the images from CT and computer-aided detection has become one the crucial solutions to discover lung cancer. This technology has not only greatly improved the traditional way of X-ray photography, which was used by doctors to diagnose lung cancer from their experiences, but also exposed the precise area of lung nodules and their sizes as well as the lung organism. In addition, western research has indicated that the geometrical features of lung nodules are the key parameters to differentiate between benign and malignant nodules. However, there are still differences between countries, for example, the areas of tumors in the lungs and their types. For this reason, this research aims at finding out the appropriate way to distinguish between benign and malignant nodules through case studies in Taiwan. Methods used in this research will be proceeded as follows: collecting data form CT, image processing, and then analyzing the data by statistics. Moreover, I will evaluate the effects of using the western criteria on patients’ data in Taiwan. Support Vector Machine (SVM) is used to predict tumor types, and we find the statistical feature values from different nodule shapes with the relationship with benign or malignant nodules. Hence, the result shows that our work can assist doctors to improve inaccuracies in using computer-aided detection.