D

data preprocessing

The CBIS-DDSM dataset is popular for training computer vision models for breast cancer detection from mammograms. However, the description files (CSV) have errors where file paths point to incorrect images, among other issues. To make it more accessible, I have openly shared the corrected dataset along with the processing tool to enable reproducibility.

This script contains a collection of functions designed to put EthoVision tracks into a more usable format for further data analysis. This includes interpolating, smoothing, extracting data and metadata and writing them out as separate csv files.

website: https://simula-srl.gitlab.io/damast/README.html