COVID-19 Isl Dataset

Description

In the dataset, lung CT images of 495 patients who were selected in accordance with the study criteria, had RT-PCR test results and had CT scans were accessed. In total, 361 of these patients were COVID-19 patients and 134 were non-COVID-19 patients. A total of 15584 lung CT slices were obtained, 8395 slices from 361 COVID-19 cases and 7189 slices from 134 non-COVID-19 cases. While the images were originally in DICOM format, they were converted to png format.

Steps to reproduce

In the study, patients who applied to Ondokuz Mayıs University Faculty of Medicine Health Practice and Research Hospital with suspicion of COVID-19 between 01.01.2020 - 31.12.2021, who were 18 years of age/ over 18 years of age, who had RT-PCR test, and who had at least one CT scan in the Radiology Department were included. The confirmation of whether the applicants were COVID-19 patients or not was done by RT-PCR test result. CT images with artefacts, inconclusive diagnosis and poor quality were excluded from the study. All CT examinations were performed on an 80-slice CT scanner (Aquilion PRIME, Canon Medical Systems, Otawara, Japan). Unenhanced lung CT scans were obtained with a low dose technique (mA; Automated exposure control, kV: 100, slice thickness: 1 mm) in the CT unit, which was specifically located in an isolated area of the emergency department. After all images were obtained, it was decided which slices to include or exclude in the presence of an expert radiologist (10 years of experience).

Categories

Funding

Licence