Chest X-ray dataset and ground truth for lung segmentation

Abstract

Chest X-ray images are a valuable tool for accurately and efficiently diagnosing Covid-19 with the assistance of computer technology. These images enable the detection of diseases in internal organs, particularly the lungs, by providing crucial information about the pathological state of the lungs and other internal organs and tissues. Segmentation plays an essential role in the earliest stages of disease detection through computer-assisted analysis of medical images. This method enables the extraction of significant elements from the image, facilitating the identification of relevant areas. In the subsequent stage, healthcare professionals might acquire more precise diagnosis outcomes. Deep learning plays a significant role in developing models to achieve exact and efficient diagnostic results in picture segmentation and image classification procedures. However, using deep learning models in the image segmentation process necessitates the availability of image datasets and ground truth that radiologists have validated to facilitate the training process. The dataset provided in this article comprises 292 chest X-ray images obtained from Airlangga University Hospital in Indonesia. These images are accompanied with ground truth data that has been meticulously verified by radiologists. The offered X-ray images encompass those of patients diagnosed with Covid-19, pneumonia and those representing normal conditions. The provided dataset exhibits potential utility in advancing artificial intelligence techniques for segmentation and classification procedures.

Keywords

• Chest X-ray
• Covid-19
• Medical image segmentation
• Deep learning
• Diagnosis