Web based AI Grading for Breast Cancer IHC Markers
To find: (1.) Mitotic cell detection on H&E images (2.) Cell Detection on IHC Endometrium images (3.) Breast Cancer cell classification on patches obtained from a WSI
breast-cancer_-endoNuke_-mitosis-_demo-video_.mp4
Datasets Used : EndoNuke(for Endometrium Cell Detection), Miccai 2015(for Mitotic Cell Detection) and IHC Breast WSI Model : YOLOv5
TO RUN:
Create .env file containing the following credentails: KEY= enter a random key unique to you DATABASE= enter your MongoDB Atlas URL TWILIO_ACCOUNT_SID= enter Twilio Account ID TWILIO_AUTH_TOKEN= enter Twilio Authentication Token TWILIO_PHONE_NUMBER= enter Twilio Phone Number
Create a static folder containing Ground truth values(for Mitosis, EndoNuke images), obtained from the labels.
Train YOLOv5 models on Mitosis and EndoNuke datasets after required pre-processing(split, augment, etc) and add the corresponding best.pt files to the root directory of the project.
Run using the command python app.py and view results in localhost:5000
Can deploy using AWS.