Creation and validation of a chest X-ray dataset with eye-tracking and report dictation for AI development

Abstract We developed a rich dataset of Chest X-Ray (CXR) images to assist investigators in artificial intelligence. The data were collected using an eye-tracking system while radiologist reviewed and reported on 1,083 CXR images. contains the following aligned data: image, transcribed radiology report text, radiologist’s dictation audio eye gaze coordinates data. hope this can contribute various areas research particularly towards explainable multimodal deep learning/machine learning methods. Furthermore, disease classification localization, automated generation, human-machine interaction benefit from these experiments that utilize attention maps produced by show potential utility dataset.