Proposed and evaluated a novel CAD system for pleural effusion detection on 638 chest radiographs.
A robust way to localize the costophrenic region using chest wall contour as a landmark structure, in addition to the lung segmentation is introduced.
Additional region descriptors are proposed based on intensity and morphology information in the region around the costophrenic recess.
Superior results are achieved compared to prior work with AUC values of 0.84 and 0.90 for the left and right pleural effusion detection, respectively.