This is the implementation of the submitted paper “Development and Validation of a Prediction Model for Treatment-related Malignant Fistulas in Advanced Esophageal Squamous Cell Carcinoma”.
Here is the main framework of the model
To facilitate model training, the data used should be cropped out of the oesophageal cancer region after preprocessing with a size of (10,32,32) for model training and inference.
This code has been tested On Ubuntu 20.04.
The required python package versions are shown below:
Python =3.8.6
numpy =1.23.1
tensorflow =2.12.0
six =1.16.0
sklearn =1.1.1
matplotlib =3.7.3
pandas =1.4.3
tqdm =4.64.0
SimpleITK =2.2.0
First, install Miniconda on your machine (download the distribution that comes with python3).
wget https://mirrors.tuna.tsinghua.edu.cn/anaconda/miniconda/Miniconda3-py38_4.8.3-Linux-x86_64.sh --no-check-certificate
bash Miniconda3-py38_4.8.3-Linux-x86_64.sh
conda -V
After setting up Miniconda, instal SimpleITK
conda install simpleitk
Then, create a conda environment
conda create -n ESCC_MEF python=3.8.6
Activate the environment
conda activate ESCC_MEF
Finally, install the python package as described above.
- Cropping of oesophageal cancer tumour region image using Rertangle.py file
- Run train.py to train the model
- Run test.py to test the model
If it runs properly, you will get the following results
