bertrandchauveau/KidneyCell2Tissue_combined

Model Card for KidneyCell2Tissue H&E+molecular models

Requesting Access

Access is granted only to users whose Hugging Face primary email matches their institutional/academic email. Requests from personal domains (e.g., @gmail, @hotmail) will be declined.

Model Description

• Developed by: Bertrand Chauveau, Bordeaux University Hospital
• Model type: Ultralytics YOLO26 model, channels first
• Task: instance segmentation of glomeruli, tubules, arteries/arterioles and peritubular capillaries in kidney transplant biopsies (H&E + molecular)
• Dataset: Trained on multispectral images (5 μm-thick H&E + cell semantic masks), BGR tile 1024 x 1024 pixels at a resolution around 0.29 μm/pixel
• Input: shape (1024, 1024, 6), raw pixel values 0.-255. as uint8, BGR
• Repository: https://github.com/bertrandchauveau/KidneyCell2Tissue
• Paper: pending
• License: cc-by-nc-sa-4.0

Intended Use & Limitations

Intended Use

These models are intended for research purposes, for automated segmentation of tissue compartments in kidney transplant from Xenium datasets.

Limitations

• Research use only
• Stain specificity: The model was trained specifically on multispectral images (5 μm-thick H&E + cell semantic masks).

How To Use

Load model

from huggingface_hub import login
from huggingface_hub import hf_hub_download
from ultralytics import YOLO

login()  # login with your User Access Token, found at https://huggingface.co/settings/tokens

#load model (example for glomeruli)
REPO_ID = "bertrandchauveau/KidneyCell2Tissue_combined"
FILENAME = "yolo26_1024_glomerulus_xenium_multi_thr0480.pt"

# Download the .pt file from the Hub (cached locally for future use)
model_path = hf_hub_download(repo_id=REPO_ID, filename=FILENAME)

# Load the weights into Ultralytics YOLO
model = YOLO(model_path)

Inference

import cv2
import numpy as np

img_size = (1024, 1024)
threshold = float(FILENAME.split('thr')[1].split('.')[0]) / 10000 #recommended threshold

#get an example image
!wget https://raw.githubusercontent.com/bertrandchauveau/KidneyCell2Tissue/main/example_image/example_combined.tiff

#load image
success, frames_list = cv2.imreadmulti("example_combined.tiff")
img_array = np.stack(frames_list, axis=2)

#prediction
res = model.predict(img_array, imgsz=(1024, 1024), conf=threshold, iou=0.7, max_det=300, retina_masks=True)

Software Dependencies

Python Packages

• ultralytics >=8.4.0

Contact

For any additional questions or comments, contact Bertrand Chauveau (bertrand.chauveau@chu-bordeaux.fr)