panoptes-he 0.0.6

A multi-resolution CNN to predict endometrial cancer features

Panoptes

panoptes is a InceptionResnet-based multi-resolution CNN architecture for cancer H&E histopathological image features prediction. It is initially created to predict visualize features of endometrial carcinoma (UCEC), hoping to automate and assist gynecologic pathologists making quicker and accurate decisions and diagnosis without sequencing analyses. Details can be found in the paper: It can also be applied to other cancer types.

Features included

Currently, it includes training/validating/testing of following features of endometrial cancer:

• 18 mutations (ARID1A, ATM, BRCA2, CTCF, CTNNB1, FAT1, FBXW7, FGFR2, JAK1, KRAS, MTOR, PIK3CA, PIK3R1, PPP2R1A, PTEN, RPL22, TP53, ZFHX3)
• 4 molecular subtypes (CNV.H, CNV.L, MSI, POLE); please type "subtype" in bash script. if you want to predict only 1 subtype (eg. CNV.H), then type "CNV.H".
• Histological subtypes (Endometrioid, Serous); please type "histology" in bash script

Modes

• train: training a new model from scratch.
• validate: load a trained model and validate it on a set of prepared samples.
• test: load a trained model and apply to a intact H&E slide.

Variants

PC are architectures with the branch integrating BMI and age; P are original Panoptes

• Panoptes1 (InceptionResnetV1-based; P1/PC1)
• Panoptes2 (InceptionResnetV2-based; P2/PC2)
• Panoptes3 (InceptionResnetV1-based; P3/PC3)
• Panoptes2 (InceptionResnetV2-based; P4/PC4)

User Interface

A simple user interface is included. If GUI is available, pop-up prompts will ask for inputs. Otherwise, read a prepared bash file (see example: Scripts/sample_bash_script.sh). If bash file not available or inputs are invalid, interactive command line prompts will ask for inputs.

Usage

• Create folders Results, images, tiles under Panoptes folder if they are not there.
• Please install through pip pip install panoptes (pending)
• Requirements are listed in requirements.txt
• Scanned H&E slide files should be put in images folder under Panoptes folder
• For train and validate mode, label file must also be provided. Example can be found in sample_lable.csv
• For train and validate mode, random split is default for data separation. If you prefer a customized split, please provide a split file (example can be found in sample_sep_file.csv)
• Output will be in a folder with name of your choice under Results folder.
• In validate and test, to load pre-trained model, please enter the full path to it.
• In test, please enter the full path to the image to test.
• For bash input, please refer to example at Scripts/sample_bash_script.sh