SSMCH-ECG: 849 Clinically Validated 12-Lead Paper ECG Images from Bangladesh

A dataset from Shaheed Suhrawardy Medical College and Hospital in Dhaka, Bangladesh, containing 849 digitized 12-lead paper-based electrocardiogram (ECG) images. The collection is clinically validated and includes three classes: 434 Normal, 344 Abnormal, and 71 Myocardial Infarction (MI) cases. It was prepared for public dissemination through PhysioNet and last updated in April 2026.

Use Cases

• Train deep learning models for multi-class cardiac abnormality detection based on the Normal, Abnormal, and MI labels.
• Develop algorithms to digitize paper-based ECG reports into analyzable signals based on the high-resolution image scans.
• Benchmark model generalizability on real-world clinical samples from a developing nation setting, as mentioned in the background.
• Create automated screening tools to reduce clinician workload in primary care, as suggested by the dataset's purpose.

Strengths

• 849 clinically validated ECG images with ground-truth labels established in collaboration with a clinical specialist.
• Clear class distribution: 434 Normal, 344 Abnormal, and 71 Myocardial Infarction cases.
• Images were digitized under standardized conditions for uniform illumination, minimal distortion, and full 12-lead visibility.
• Stored in lossless PNG format to preserve signal fidelity.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data reflects a single geographic source (Bangladesh), which may limit generalizability to other populations.

Provenance

Source

Shaheed Suhrawardy Medical College and Hospital (SSMCH), Dhaka, Bangladesh.

Collection Method

Paper-based ECG reports were digitized via high-resolution scanning or controlled photography under standardized conditions.

Time Range

null

Freshness

Last updated 2026-04-15 06:57:52; freshness should be verified.

Geography

Dhaka, Bangladesh.