Towards Accountable AI-Assisted Eye Disease Diagnosis: Workflow Design, External Validation, and Continual Learning
This repository provides related codes, data, and models for the paper titled 'Towards Accountable AI-Assisted Eye Disease Diagnosis: Workflow Design, External Validation, and Continual Learning'.
Have python3.8+ and Tensorflow 2.9.1 + installed.
git clone https://github.com/ncbi-nlp/deepseenet-plus.git cd deepseenet-plus
pip install -r requirements.txt
Please download the models from here.
Inference can be performed with the following command. Replace options with the correct paths. This will grade scores for each risk factor, as well as a final simplified severity score.
python model.py -i --model_folder=models/ --image_folder=image_set/ --input_file=input_files.csv --output_file=output_file.csv
To use the example data provided, run the following, replacing models/ with the model folder.
python model.py -i --model_folder=models/ --image_folder=examples/example_images/ --input_file=examples/example_input_file.csv --output_file=examples/example_output.csv
A saved risk factor model can be further trained. Specify the targeted risk factor with risk_factor (either "drusen", "pigment", or "amd").
python model.py -t --model_path=model.h5 --image_folder=image_set/ --input_file=input_files.csv --risk_factor=drusen/pigment/amd
This will load the model from model_path, read images and labels from input_file, train the model, and save the latest best model.
This tool shows the results of research conducted in the Computational Biology Branch, NCBI.
The information produced on this website is not intended for direct diagnostic use or medical decision-making without review and oversight by a clinical professional. Individuals should not change their health behavior solely on the basis of information produced on this website. NIH does not independently verify the validity or utility of the information produced by this tool. If you have questions about the information produced on this website, please see a health care professional.
More information about NCBI's disclaimer policy is available.
About text mining group.
The performance characteristics of this product have not been evaluated by the Food and Drug Administration and is not intended for commercial use or purposes beyond research use only.
This research was supported by R00LM014024 and the NIH Intramural Research Program of National Library of Medicine and National Eye Institute. Dr. Mehta would also like to acknowledge a departmental unrestricted grant by Research to Prevent Blindness. The views do not reflect the official policy or position of the U.S. Army Medical Department, the U.S. Army Office of the Surgeon General, the Department of the Army, Department of Defense, the Uniformed Services University of the Health Sciences, or any other agency of the U.S. Government.