Computer Aided Detection of Lung Nodules in Multislice Computed Tomography

Early detection may be of critical importance in lung cancer prognosis. Multi-detector Computed Tomography (CT) increases sensitivity in early lung cancer detection by potentially identifying nodules of smaller size. A Computer Aided Detection (CAD) system for automatic identification of lung nodules is proposed. The system is multistage, including segmentation of lung boundaries, initial nodule pick up and False Positive (FP) reduction. The system is targeted to improve lung boundary identification by an automatic thresholding a approach, capable of dealing with juxta pleura nodules. A selective enhancement filter was implemented in combination with minimum error thresholding for initial nodule pick up to deal with vessels abutting small nodules. FP regions were subsequently removed using a Support Vector Machine (SVM) classifier employing morphological features extracted from corresponding nodule candidate regions of the enhanced and the original images. The proposed automated scheme was tested on a slice data set containing 279 nodules from 21 cases available by the Lung Image Database Consortium. System performance on a slice basis provided sensitivity of 81%, with an average of 5 FPs per slice. Further analysis of the slice dataset with respect to size, contrast and location of nodules provided sensitivities of 81%, 83% and 85% for nodules of small size, low contrast and near pleura. The proposed CAD scheme may be a useful tool in assisting radiologists in lung nodule detection.