Results of Computer-Aided Diagnosis of Peripheral Bronchial Carcinoma

Publisher Website
Google Scholar
Abstract
Radiographic images can be converted into numerical sequences that can be evaluated by computer to determine the significance of certain radiographic findings in human peripheral pulmonary nodules. Of 3 methods (discriminant analysis, Bayes'' theorem, and the CALM system), discriminant analysis was the best with an accuracy rate of 85%. By taking into account serial radiographs and statements on the possible growth of the neoplasm, the accuracy of the method can be increased to 90%. A simplified, inexpensive and effective method for application of this diagnostic tool is described.

This publication has 3 references indexed in Scilit: