Three dimensional lung nodule segmentation and estimation using thresholding on local thickness

A fast alternative user-selection procedure to semi-automatic Philips Extended Brilliance Workspace´s Lung Nodule Assessment™ is presented. By using modified Local Thickness algorithm which utilizes 3D Region Growing algorithm and IsoData auto-thresholding algorithm, the user can perform fewer steps to semi-automatically segment and estimate a respective nodule more accurately, especially on irregular shape nodules. Attached blood vessels and extruding thorn-like features of the selected nodule are also automatically excluded in segmentation with the proposed method. By using correlation comparison on the same lung nodule dataset between our proposed method and Philips Extended Brilliance Workspace´s Lung Nodule Assessment™´s semi-automatic method, we can see that the correlation is 0.9887.