Capstone code related to automated classifications of breast cancer lesions
- images: contains all of the original images in the data set and all of the marked images that the radiologist created
- spreadsheets: contains the file of annotations from the radiologist (BUSL_Notes.csv) as well as any other spreadsheets created or consumed by the code along the way
- Common.py
This file contains the primary function for opening image files.
- Radiomics.ipynb
Contains code for iterating through the image files and generating radiomics statistics
- Data_Prep.ipynb
Contains code for combining the radiomics data with the annotations provided by the radiologist, and then adding additional columns for abruptness and orientation metrics
- Boxplots.py
Contains code for comparing how the numerical algorithms perform relative to the annotations provided by the radiologist.
- Pycaret_Radiomics_Auot_Split.ipynb
As the name suggests, this file uses Pycaret to generate a list of models based on the cleanest version of the data, once unhelpful columns have been removed. The top candidates are then evaluated for performance and feature importance.