electricsheepafrica/Multimodal-Chest-X-ray-dataset-for-Normal-and-Bacterial-Pneumonia-in-Africans

African Diseases Multimodal Dataset

Dataset Description

This is a synthetic multimodal dataset of chest X-ray images for Normal and Bacterial Pneumonia. It includes synthetic Electronic Health Records (EHR) to provide additional modalities for multimodal learning.

Dataset Summary

• Total Samples: 4353
• Labels: Normal (1576), Pnemonia (2777)
• Modality: Multimodal (Images, Text, Tabular)
• Source: Based on Coronahack Chest X-Ray Dataset, filtered to Normal and Bacterial Pneumonia, with synthetic EHR added.
• Focus: Western Africa

Data Structure

Images

• Location: dataset/Coronahack-Chest-XRay-Dataset/
• Format: JPEG
• Description: Chest X-ray images

Metadata

• File: dataset/Comprehensive_Metadata.csv

• Columns:

  • Unnamed: 0 (index)
  • X_ray_image_name
  • Label
  • Dataset_type (TRAIN/TEST)
  • Label_2_Virus_category
  • Label_1_Virus_category
  • Age
  • Gender
  • Location (Western African countries)
  • Symptom_1, Symptom_2, Symptom_3, Symptom_4
  • Diagnosis
  • Notes (narrative EHR)
  • WBC_Count
  • CRP_Level

Synthetic Components

• Demographics: Age (1-100), Gender, Location
• Symptoms: Up to 4 symptoms per patient
• Lab Results: WBC_Count (4-20), CRP_Level (0-200)
• Notes: Generated narrative based on symptoms

Usage

This dataset can be used for:

• Multimodal disease diagnosis models
• Chest X-ray classification
• Synthetic EHR generation for medical AI

Ethical Considerations

• This dataset contains synthetic data for research purposes.
• Ensure compliance with medical data privacy regulations.
• Not intended for clinical use.

License

MIT License

Citation

If you use this dataset, please cite Electric Sheep Africa.