Heart Dataset

This dataset contains X-ray imaging of the chest. X-rays were prescribed mostly for routine check-ups or due to various cardiovascular or pulmonary disorders. Chest X-rays were taken in the posterior-anterior projection in the upright position with the chest in the mid-inspiratory condition using Siemens Siemens Ysio Max or Axiom Luminos dRF. The X-ray images were resized to a unified size of 512 by 512, and then three radiologists manually segmented the lungs and the heart. Radiologists separately annotated a single X-ray image and then analyzed it together to unify the process of segmentation of lungs and heart. The dataset's final version consists of 114 2D images (342 segmentation)

The dataset contains two folders: The images folder contains the X-ray images, and The segmentations folder contains three segmentations by three experts in the following format: {i}_{1,2,3} where i is the image number and {1,2,3} are labels. The dataset (images/segmentations folder) is further divided into 65 images in the training folder and 49 images in the testing folder.