Systematic assessment of the involvement of regional lymph nodes in cancer patients in a standardized manner is important for disease staging and potentially for prognostication of patient outcome. For certain body regions like thorax, international standards such as the one proposed by the International Association for the Study f Lung Cancer (IASLC) for defining nodal stations are available. While this formulation is helpful in standardizing a means of interpreting and reporting thoracic lymph node disease sites, it still leaves the radiologist with the arduous task of following the detailed specifications and finding the nodal stations and zones on images subjectively. Our hypothesis is that if the lymph node stations can be identified automatically during image interpretation, then the acceptance of the standard and the consistency of its interpretation will be greatly facilitated, which may rapidly promote standardized reporting. In this project, we not only develop nodal station definitions for other body regions where definitions do not exist currently, but also employ the AAR technology for automatically localizing the lymph node stations in different body regions in CT, PET/CT, and MRI images and subsequently quantify disease burden station-wise.
An example from such an implementation for identifying the IASLC defined thoracic nodal stations is shown below where 3D renditions of the fuzzy models of some stations together with some anatomic objects are displayed as well as the AAR-localized zones overlaid on CT slice images.