Wavelet snake for classification of nodules and false positives in digital chest radiographs

The authors have developed a technique using a wavelet-based snake model for classification of nodules and false positives reported by their computer-aided diagnosis (CAD) scheme for detection of pulmonary nodules in digital chest radiographs. In this technique, the boundary of a nodule is first approximated by the multiscale edge representation. These multiscale edges are used to “guide” the snake to estimate the true boundary of the nodule. The deformation of the snake is performed by changing the wavelet coefficients using a gradient descent algorithm. The degree of overlap between the fitted snake and the multiscale edges was calculated as a measure for classification of nodules and false detections. A set of ROIs consisting of 90 nodules and 152 false positives were used for evaluation of the authors´ method by means of the receiver operating characteristic (ROC) analysis. Their method yielded an area under the ROC curve of 0.74 in classification performance, which reduced 15% of the false positives with sacrifice of only one module