This repository contains the core scripts required to reproduce the image-processing pipeline used in the study:
“Personalized Machine Learning-Guided Radiation Dose Escalation in Newly Diagnosed Glioblastoma: A Prospective Early-Stage Clinical Trial.”
The repository includes a set of components necessary for the reported preprocessing, segmentation, infiltration map generation, and statistical analysis.
Clinical imaging data are not included in this repository.
preprocessing/ DICOM sorting, conversion, and preprocessing wrappers
segmentation/ DeepMedic segmentation wrapper
infiltration_maps/ Infiltration map generation scripts and trained model
Statistics_analysis/ Final statistical analysis scripts
DISCLAIMER.txt
For each subject (s) and timepoint (t = PatientID_timepoint), the pipeline performs the following steps:
- DICOM sorting and renaming
- DICOM to NIfTI conversion
- Image preprocessing
- Tumor segmentation
- Infiltration map generation
- Statistical analysis
- Paths in the original pipeline were environment-specific and have been simplified for this release.
- External imaging software dependencies may be required depending on the environment.
If you use this code, please cite:
Personalized Machine Learning-Guided Radiation Dose Escalation in Newly Diagnosed Glioblastoma: A Prospective Early-Stage Clinical Trial.
A Zenodo DOI will be generated for this repository to provide a permanent and citable version of the code.
See Disclaimer.docx.