OpenMed Medical Imaging Models

This repository contains trained ResNet50 models for medical imaging tasks from the OpenMed platform.

Models Included

Brain Tumor Classification

Model: resnet50_brain_tumor_full
Classes: brain_glioma, brain_menin, brain_tumor
Dataset: Brain Cancer MRI Dataset
Description: ResNet50 model fine-tuned for brain tumor classification in MRI scans

Tuberculosis Detection

Model: resnet50_tb_full
Classes: Normal, Tuberculosis
Dataset: TB Chest Radiography Database
Description: ResNet50 model fine-tuned for tuberculosis detection in chest X-rays

Pneumonia Detection

Model: resnet50_pneumonia_full
Classes: Normal, Pneumonia
Dataset: Chest X-ray Pneumonia Dataset
Description: ResNet50 model fine-tuned for pneumonia detection in chest X-rays

Quick Start

from huggingface_hub import hf_hub_download
import torch

# Download a specific model (example: brain tumor classification)
model_path = hf_hub_download(
    repo_id="hitmanonholiday/openmed-medical-imaging-models",
    filename="resnet50_brain_tumor_full/pytorch_model.bin"
)

# Load the model
checkpoint = torch.load(model_path, map_location="cpu")

Model Architecture

All models are based on ResNet50 with ImageNet pre-trained weights:

Input Size: 224x224x3 RGB images
Normalization: ImageNet statistics
Fine-tuning: Full network training (all layers trainable)

Usage Guidelines

Important: These models are for research and educational purposes only. They should NOT be used for actual medical diagnosis without proper validation and oversight by qualified healthcare professionals.

Performance Metrics

Performance metrics for each model can be found in their respective model cards.

Citation

@misc{openmed_models_2024,
  title={OpenMed: AI-Powered Medical Imaging Analysis Platform},
  author={OpenMed Team},
  year={2024},
  howpublished={\url{https://huggingface.co/hitmanonholiday/openmed-medical-imaging-models}}
}

License

MIT License - See individual model cards for details.

Disclaimer: These models are for research and educational purposes only. They should not be used for medical diagnosis or treatment decisions without proper validation and oversight by qualified healthcare professionals.