Anonymized thoracic radiographs (X-rays) and computed tomography (CT) scans of patients with suspected COVID-19. Images are accompanied by a positive or negative diagnosis for SARS-CoV-2 infection via RT-PCR. The dataset was provided by Hospital das Clínicas da Universidade de São Paulo, Hospital Sirio-Libanes, and Laboratory Fleury.
Use Cases
- Train image classification models to detect COVID-19 based on thoracic radiographs.
- Develop multimodal diagnostic tools using both X-ray and CT scan images.
- Benchmark model performance using RT-PCR confirmed labels.
- Study visual patterns of SARS-CoV-2 infection across different imaging modalities.
Strengths
- Images are accompanied by RT-PCR confirmed diagnostic labels.
- Data is sourced from multiple Brazilian medical institutions.
- Includes two distinct imaging modalities: X-rays and CT scans.
Limitations
- Column-level documentation is absent; field semantics must be inferred after download.
- Row count is unknown, which may limit suitability assessment.
- Data may reflect geographic bias inherent to the Brazilian sources.
Provenance
- Source
- Faculdade de Medicina da Universidade de São Paulo Institute of Radiology (InRad)
- Collection Method
- Provided by Hospital das Clínicas da Universidade de São Paulo, Hospital Sirio-Libanes, and Laboratory Fleury.
- Geography
- Brazil (likely)