Emphysema_Quantification

Accompanies "Emphysema Quantification and Severity Classification with 3-Dimensional Averaging Kernel and Airways Removal" on medRxiv; DOI: https://doi.org/10.1101/2022.10.31.22281562

This repository provides a pipeline for accurately quantifying emphysema score from chest CT scans.

See requirements.txt for required packages.

• report_extraction.py provides basic code for extracting emphysema extent classified by radiologist from a patient report.
• data_processing_utils.py contains a wrapper_fn that takes in a folder directory that contains a series of DICOM file representing one particular CT scan and obtains the emphysema from that CT scan. This function can be incorporated with multiprocessing that is not included here.
• image_preprocessing_utils.py and emphysema_visualizer_utils.py have some nice utility functions for visualizing 3D lung CT scans and emphysema.
• train.py provides code for determining cutoffs and associated metrics for emphysema classification into "none," "mild to moderate," and "severe" according to a training dataset of (emphysemas score, radiologist classification) pairs.