Title :
Sizing tumors with TNM classifications and rough sets method
Author_Institution :
Dept. of Radiol., Johns Hopkins Univ., Baltimore, MD, USA
Abstract :
In this paper, medical decision support systems for TNM (tumor characteristics, lymph node involvement, and distant metastatic lesions) classification aiming to divide cancer patients to low and high risk patients are presented. In addition, the system also explained the decision in the form of IF-THEN rules and in this manner performed data mining and new knowledge discovery. The case studies show that the system is robust and not dependent on the database size and the noise. The accuracy was almost 80% which is comparable with the accuracy of physicians and much better then obtained with more conventional discriminant analysis (62% and 67%).
Keywords :
cancer; data mining; database management systems; decision support systems; logic design; medical diagnostic computing; pattern classification; rough set theory; tumours; IF-THEN rules; TNM classifications; cancer patient; data mining; distant metastatic lesions; knowledge discovery; lymph node involvement; medical decision support systems; rough sets method; tumor characteristics; tumors sizing; Cancer; Data mining; Databases; Decision support systems; Lesions; Lymph nodes; Metastasis; Neoplasms; Noise robustness; Rough sets;
Conference_Titel :
Computer-Based Medical Systems, 2004. CBMS 2004. Proceedings. 17th IEEE Symposium on
Print_ISBN :
0-7695-2104-5
DOI :
10.1109/CBMS.2004.1311718