Research on Technologies of Computer Aided Diagnosis for Solitary Pulmonary Nodule Based on CT Images

摘要

The paper proposed an algorithm for detecting solitary pulmonary nodules. It can achieve a good distinction between pulmonary nodules and non-nodules. And the algorithm can improve the accuracy of the subsequent classification of pulmonary nodules. We use the DICOM standard chest CT sequential images in the LIDC-IDRI image library as objects investigated. The results show that the method of this study achieves a good preliminary classification between real pulmonary nodules and non-nodules. The sensitivity of the result is 96.0%, the specificity is 96.55%, the false-positive rate is 2.30%, the false-negative rate is 6.0%. Among these indexes, the most important index is specificity. In contrast the existing method, specificity can improve 3% to 4%.