A Method for Detecting Suspicious Border Regions of Extracted CT Lung Region

For re-including the omitted suspicious region along lung boundary, a method based on SIFT information and the adapted chain code is proposed. First of all, all SIFT points in the binary image are extracted in the binary lung segmentation image. After excluding noise points from SIFT points, the support edges are calculated by the Euclidean distance. Then the suspicious boundary region recognition step is conducted using both a combined code and a bond number. Experiments on three real CT images demonstrate the present scheme provides an efficient way to identify suspicious lung boundary.