A System for Automated Screening for Tuberculosis using Digital Chest X-rays for Resource-Constrained Regions

Friday, November 08, 2013 — Poster Session IV
2:00 p.m. – 4:00 p.m.	FAES Academic Center (Upper-Level Terrace)	NLM	COMPBIO-22

Authors

Z. Xue
S. Jaeger
A. Karargyris
S. Candemir
S. Antani
R. Long
G.R. Thoma
C. McDonald

Abstract

Tuberculosis (TB) is widespread and affects a third of the world. The National Library of Medicine is collaborating with AMPATH (the Academic Model Providing Access to Healthcare), an organization that runs the largest AIDS treatment program in the world, to develop a system to screen HIV-positive patients in resource-constrained rural Kenya for evidence of pulmonary tuberculosis in chest x-rays. NLM has provided AMPATH with lightweight digital x-ray units for use by their staff screen the population for the presence of disease. These units are already on site in Eldoret, in western Kenya, and are being readied for deployment in a mobile x-ray truck. The computer-aided screening system is being designed to automatically analyze the patient chest CXR after it is acquired. It comprises a mobile CXR unit and reader mounted on a truck, a laptop computer, and a software system that contains four main components: image reader, lung segmentation, feature extraction, and classification. After images are acquired and received by the image reader, lung regions are automatically identified and an SVM-based classifier categorizes extracted features to determine whether the input chest x-ray is normal or abnormal. System performance has been evaluated on two TB datasets with promising results.