Machine learning using entropy-based texture features from MRI to differentiate histological subtypes of non-small cell lung cancer identified as metabolically active on PET/MRI

Zenodo (CERN European Organization for Nuclear Research)

The dataset contains magnetic resonance (MR) images of non-small lung cancer, categorized into two groups: adenocarcinoma (ADC, 24 patients, with 75 images) and squamous cell carcinoma (SCC, 20 patients, with 80 images). Measured features, including sample entropy, fuzzy entropy, permutation entropy, dispersion entropy, and distribution entropy, are examined across five scales. Directory structure:* data/ - numpy arrays corresponding to MRI images (128x128 pixels) +arrays are named: o_<number_image>_<slice_of_image>.png * oryg/ - numpy arrays corresponding to original MRI images +arrays are named: o_<number_image>_<slice_of_image>.png * MRI_slice_load.py code for data loading and calculating features* database.xlsx - a .xlsx file with two columns ['FileName', 'Group'] corresponding to MRI images (128x128 pixels) and group annotations (adenocarcinoma (ADC) and squamous cell carcinoma (SCC))* database_features.xlsx - a .xlsx file with three sheets of filtration ('Normalize', 'Median') and measured features in 5 scales (range 0-4) of every image: + No. + FileName + Group + Sample Entropy (SampEn) + Fuzzy Entropy (FuzzEn) + Dispersion Entropy (DispEn) + Distribution Entropy (DistEn) + Permutation Entropy (PermEn) Licence:The dataset is licensed under the Creative Commons Attribution 4.0 International LicenseTo view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Authors: Borowska, Marta, Mojsak, Malgorzata, Bębas, Ewelina, Pauk, Jolanta, Hładuński, Marcin, Domino, Malgorzata

DOI: https://doi.org/10.5281/zenodo.15845727

Publish Year: 2025