This repository provides the official implementation of AI-based_GN_Diagnostic_Assistance_Tool。
Improving Diagnostic Efficiency in Glomerular Nephritis through an Integrated AI-based Pathological Image Analysis Approach (Under Review)
An AI-based GN diagnostic assistance tool is developed and the diagnostic pipeline comprises three sequential steps: glomerulus segmentation, glomerulus lesion feature extraction and patient-level diagnosis.
The tool consists of three core components: (1) a glomerular localization module for precise glomerulus segmentation; (2) two multi-classification module for identifying glomerular lesions; (3) a patient-level classification module for diagnosing four GN subtypes.
-
Clone the repository.
git clone https://github.com/Git-HB-CHEN/AI-based_GN_Diagnostic_Assistance_Tool.git cd AI-based_GN_Diagnostic_Assistance_Tool -
Create a virtual environment for AI-based_GN_Diagnostic_Assistance_Tool and activate the environment.
conda create -n GNDAT python=3.9 conda activate GNDAT -
Install Pytorch and torchvision. (You can follow the instructions here)
-
Install other dependencies.
pip install -r requirements.txt
- Inference for Light Microscopy Images
python running_GNDA_tool_LM_Image.py - Inference for Light Microscopy and Immunofluorescence Images
python running_GNDA_tool_LM_IF_Image.py
- Training the glomerulus segmentation model
python running_training_glomerulus_segmentation.py - Training the glomerular lesion classification model
python running_training_glomerular_lesion_classification.py - Training the patient-level classification model
python running_training_patient_classification.py
Details of the model architecture and its associated weights are being curated and will be released following the acceptance of this manuscript.