This repository contains the final submission of team "달려달려" in the Medical Image AI Challenge 2023: Pathology data hosted by Seoul National University Hospital.
- Competition: Medical Image AI Challenge 2023: Pathology data
- Host: Seoul National University Hospital
- Objective: Prediction of melanoma recurrence using whole slide image and tabular information
- Result: Reached top 5 private score out of 62 teams, advanced to the final presentation stage but failed to win awards
- Reproduction: Clone this repository and execute
main.py - Presentation: Here
cf. The datasets used cannot be publicly available due to the regulations.
Vault
├── main.py
├── preprocessing
│ ├── patch_conversion.py
│ ├── patch_extraction.py
│ └── removing_background.py
├── resnet
│ ├── predictions
│ ├── weights
│ ├── dataset.py
│ ├── model.py
│ ├── predictor.py
│ └── trainer.py
├── unetvit
│ ├── predictions
│ ├── weights
│ ├── dataset.py
│ ├── model.py
│ ├── predictor.py
│ └── trainer.py
├── utils
│ ├── config.py
│ └── utils.py
└── dataset
├── train
│ ├── train
│ │ ├── train_0001.png
│ │ ├── ...
│ │ └── train_xxxx.png
│ ├── train_bg_rm
│ │ ├── train_0001.png
│ │ ├── ...
│ │ └── train_xxxx.png
│ ├── train_patch
│ │ ├── train_0001_1.png
│ │ ├── ...
│ │ ├── train_0001_5.png
│ │ ├── train_0002_1.png
│ │ ├── ...
│ │ └── train_xxxx_5.png
│ └── train_patch_rs
│ ├── train_0001_1.npy
│ ├── ...
│ ├── train_0001_5.npy
│ ├── train_0002_1.npy
│ ├── ...
│ └── train_xxxx_5.npy
├── test
│ ├── test
│ ├── test_bg_rm
│ ├── test_patch
│ └── test_patch_rs
├── train_dataset.csv
└── test_dataset.csv