Deep learning radiomics model related with genomics phenotypes for lymph node metastasis prediction in colorectal cancer

Description

This repository is the implementation of the paper " Deep learning radiomics model related with genomics phenotypes for lymph node metastasis prediction in colorectal cancer".

Contents

This repository contains the code about developing the model, assessing the performance of model, transcript analysis (GO and KEGG analysis), and the correlation analysis.

• model developed and validated.R: developed the model through LASSO and 5 fold cross validation.

• Assessment of model.R: including the calibrate cure and decision curve analysis.

• Correlation analysis.R: including the correlation analysis of DPL features and genes, the correlation of DPL score and clinical characteristic.

• Transcript analysis.R: including the differential expressed genes, GO analysis, and KEGG analysis.

• Univarible and multivariable analysis.R: including the univarible and multivariable analysis of DPL features and top differential expressed genes for predicting the LNM.

• AI Model: including the python codes of extracting DPL features by using AutoEncoder.

Dataset

Due to privacy restrictions, we don’t upload our data used in the paper. If you are interested in our paper, please contact us.

Cite

If you use our code, please cite: