repository·main·Indexed 41 days ago
https://github.com/eidoslab/torchstainGPU-accelerated stain normalization and augmentation tools for histopathological images, compatible with PyTorch, TensorFlow, and Numpy. Supports algorithms including Macenko, Reinhard, Modified Reinhard, Multi-target Macenko, and Macenko-Aug.
Install the base package using pip. You can also specify a backend for GPU acceleration using extras: torchstain[torch] for PyTorch or torchstain[tf] for TensorFlow. The numpy backend is included by default in both.
pip install torchstain
# Or for specific backends:
pip install torchstain[torch]
pip install torchstain[tf]To normalize histopathological images, use a normalizer from torchstain.normalizers. The workflow involves:
backend (e.g., 'torch')..fit(target_image) on a target image to learn the stain characteristics..normalize(I=source_image, stains=True) on the source image to perform the transformation.Note: The input images should be compatible with the chosen backend (e.g., PyTorch tensors).
import torch
from torchvision import transforms
import torchstain
import cv2
target = cv2.cvtColor(cv2.imread("./data/target.png"), cv2.COLOR_BGR2RGB)
to_transform = cv2.cvtColor(cv2.imread("./data/source.png"), cv2.COLOR_BGR2RGB)
T = transforms.Compose([
transforms.ToTensor(),
transforms.Lambda(lambda x: x*255)
])
normalizer = torchstain.normalizers.MacenkoNormalizer(backend='torch')
normalizer.fit(T(target))
t_to_transform = T(to_transform)
norm, H, E = normalizer.normalize(I=t_to_transform, stains=True)torchstain supports several algorithms across numpy, torch, and tensorflow backends:
Normalization Algorithms:
Augmentation Algorithms: