---
license: unknown
---
# iSight: AI-assisted Automatic Immunohistochemistry Staining Assessment

A deep learning-based multi-task prediction system for automated analysis of immunohistochemistry (IHC) pathology images and protein staining patterns.

## 🔗 Links

- **Code Repository**: [https://github.com/zhihuanglab/iSight](https://github.com/zhihuanglab/iSight)
- **Training Dataset**: [nirschl-lab/hpa10m](https://huggingface.co/datasets/nirschl-lab/hpa10m)


## 🎯 Prediction Tasks

The model simultaneously predicts 5 key attributes of IHC images:

| Task | Classes | Labels |
|------|---------|--------|
| **Staining Intensity** | 4 | negative, weak, moderate, strong |
| **Staining Location** | 4 | "none", "cytoplasmic/membranous", "nuclear", "cytoplasmic/membranous,nuclear" |
| **Staining Quantity** | 4 | none, <25%, 25%-75%, >75% |
| **Tissue Type** | 58 | Various human tissue types |
| **Malignancy** | 2 | normal, cancer |

## 📦 Model Files

```
checkpoints/
└── iSight_model_checkpoint.pth    # Model weights
```

## 🚀 Usage



### Run inference

Please refer to the [GitHub repository](https://github.com/zhihuanglab/iSight) for detailed inference instructions:

```bash
# Clone the repository
git clone https://github.com/zhihuanglab/iSight.git
cd iSight

# Run inference
bash inference_script.sh
```

## 📧 Contact

For questions or suggestions, please contact: [zhi.huang@pennmedicine.upenn.edu](mailto:zhi.huang@pennmedicine.upenn.edu) or [jjnirschl@wisc.edu](mailto:jjnirschl@wisc.edu)