Segmenting anatomy in chest x-rays for tuberculosis screening

In this paper we describe the development of a screening system for pulmonary pathologies (i.e. pneumonia, tuberculosis) application in global healthcare settings. As a first step toward this goal, the paper presents a novel approach for detecting lungs and ribs in chest radiographs. The approach is a unified method combining two detection schemes resulting in reduced cost. The novelty of our approach lies on the fact that instead of using pixel-wise techniques exclusively we used region-based features computed as wavelet features that take into consideration the orientation of anatomic structures. Initial results are described. Next steps include classification of non-rib lung regions for radiographic patterns suggesting tuberculosis infection.