COVID-19 Lung CT Image Database

This repository contains a computed tomography (CT) image database, ground truth annotations, image processing results, and classification data developed for automatic lung parenchyma segmentation and COVID-19 severity classification research.

Repository Contents

The repository contains the following folders:

aOriginal_image_GGO-PI_PP

Contains:

• original computed tomography (CT) images,
• lung parenchyma (PP) images,
• ground truth masks,
• ground-glass opacity and pulmonary infiltrate (GGO-PI) annotations.

All images are provided in JPG format.

cPROCESSING_images_RESULTS

Contains:

• processed CT image results,
• saliency map outputs,
• DCT energy map results,
• fused feature images (FF),
• final quantitative classification outputs.

All processed images are provided in JPG format.

Dataset Characteristics

• Imaging modality: Computed Tomography (CT)
• Disease: COVID-19
• Number of patients: 44
• Image format: JPG
• Classification task: COVID-19 severity assessment
• Segmentation targets:
• Lung parenchyma (PP)
• Ground-glass opacity (GGO)
• Pulmonary infiltrates (PI)

The images provided in this dataset were generated in:

 TELLO-MIJARES, Santiago; WOO, Fomuy. Novel COVID-19 Diagnosis Delivery App Using Computed Tomography Images Analyzed with Saliency-Preprocessing and Deep Learning. Tomography, 2022, vol. 8, no 3, p. 1618-1630.