H Optimus 0

🚀 H-optimus-0 Model Card

H-optimus-0 is an open - source foundation model for histology, developed by Bioptimus. It is a vision transformer with 1.1B parameters, trained on a proprietary collection of over 500,000 H&E stained whole slide histology images. This model can extract powerful features from histology images for various downstream applications, such as mutation prediction, survival analysis, or tissue classification. For more details, visit our GitHub repository.

🚀 Quick Start

✨ Features

• H-optimus-0 is designed for histology image feature extraction.
• It can be applied to multiple downstream tasks in the medical field.
• Trained on a large - scale proprietary dataset of H&E stained whole slide histology images.

📦 Installation

The installation mainly involves setting up the necessary Python libraries. Ensure you have huggingface_hub, torch, timm, and torchvision installed.

💻 Usage Examples

Basic Usage

The following code can be used to run inference with H-optimus-0. Note that the model expects images of size 224x224 extracted at 0.5 microns per pixel.

from huggingface_hub import login
import torch
import timm 
from torchvision import transforms

# Login to the Hugging Face hub, using your user access token that can be found here:
# https://huggingface.co/settings/tokens.
login()

model = timm.create_model(
    "hf-hub:bioptimus/H-optimus-0", pretrained=True, init_values=1e-5, dynamic_img_size=False
)
model.to("cuda")
model.eval()

transform = transforms.Compose([
    transforms.ToTensor(),
    transforms.Normalize(
        mean=(0.707223, 0.578729, 0.703617), 
        std=(0.211883, 0.230117, 0.177517)
    ),
])

input = torch.rand(3, 224, 224)
input = transforms.ToPILImage()(input)

# We recommend using mixed precision for faster inference.
with torch.autocast(device_type="cuda", dtype=torch.float16):
    with torch.inference_mode():
        features = model(transform(input).unsqueeze(0).to("cuda"))

assert features.shape == (1, 1536)

📄 License

This project is licensed under the Apache 2.0 license.

BibTeX Citation

If you find this repository useful, please consider citing our work:

@software{hoptimus0,
  author = {Saillard, Charlie and Jenatton, Rodolphe and Llinares-López, Felipe and Mariet, Zelda and Cahané, David and Durand, Eric and Vert, Jean-Philippe},
  title = {H-optimus-0},
  url = {https://github.com/bioptimus/releases/tree/main/models/h-optimus/v0},
  year = {2024},
}