A study on image feature selection to improve the detection performance of lung nodules on chest X-ray images

摘要

Lung cancer is one of the most serious cancers. The performance of a computer-aided diagnosis system depends on the feature set used in the CAD system. This paper shows the results of image feature selection experiments. We took 210 features to look for the optimum feature set. For the purpose, a forward stepwise selection approach was employed. We used the area under the receiver operating characteristic (ROC) curve to evaluate the performance of each feature set. Analysis of the optimally selected feature set is given and the experiments using 247 chest x-ray images are also shown. The effectiveness of the selected features is demonstrated by comparing with our previous work.