Lung Nodule Detection from PA Chest Radiographs

Research description

The developed system achieves effective results by exploiting three consecutive steps:

  1. identification of the lung area by means of a multiscale edge tracking technique driven by shape constraints;
  2. identification of candidate nodules through a multiscale approach based on a modified difference of Gaussian filter;
  3. identification of nodules by cost-sensitive SVMs designed to deal with highly unbalanced datasets.

The work has been developed with Niguarda Hospital (Milan) and Policlinico Hospital (Milan)

lung nodules

Publications

P. Campadelli, E. Casiraghi and G. Valentini. Support vector machines for candidate nodules classification. Neurocomputing, Elsevier 68, 2005.

P. Campadelli, E. Casiraghi and D. Artioli. A fully automated method for lung nodule detection from postero-anterior chest radiographs. IEEE transactions on medical imaging, IEEE 25(12), 2006.