摘要：Developments in deep learning techniques have led to significant advances in automatedabnormality detection in radiological images and paved the way for their potential use in computer-aided diagnosis (CAD) systems. However, the development of CAD systems for pulmonary tubercu-losis (TB) diagnosis is hampered by the lack of training data that is of good visual and diagnosticquality, of sufficient size, variety, and, where relevant, containing fine-region annotations. This studypresents a collection of annotations/segmentations of pulmonary radiological manifestations that areconsistent with TB in the publicly available and widely used Shenzhen chest X-ray (CXR) datasetmade available by the U.S. National Library of Medicine and obtained via a research collaborationwith No. 3. People’s Hospital Shenzhen, China. The goal of releasing these annotations is to ad-vance the state of the art for image segmentation methods toward improving the performance of thefine-grained segmentation of TB-consistent findings in digital chest X-ray images. The annotationcollection comprises the following: (1) annotation files in JavaScript Object Notation (JSON) formatthat indicate locations and shapes of 19 lung pattern abnormalities for 336 TB patients; (2) maskfiles saved in PNG format for each abnormality per TB patient; and (3) a comma-separated values(CSV) file that summarizes lung abnormality types and numbers per TB patient. To the best of ourknowledge, this is the first collection of pixel-level annotations of TB-consistent findings in CXRs.
