SetkaFluo Demo Dataset and Training Inputs for Noise2Noise Denoising of Multi-Element XRF Maps

SetkaFluo is a self-supervised deep-learning workflow for denoising multi-element X-ray fluorescence (XRF) detector maps using Noise2Noise training. This Zenodo deposit provides the example datasets and training inputs required to run all Jupyter notebooks distributed in the SetkaFluo GitHub repository. Reference:Shishkov R, Laugros A, Vigano N, Bohic S, Karpov D, Cloetens P.Self-Supervised Deep-Learning Denoising for X-Ray Fluorescence Microscopy with Multi-Element Detectors.ChemRxiv (2025). DOI: 10.26434/chemrxiv-2025-lsxpc Dataset Contents input_data.zip – example fitted XRF maps and detector-element images for data exploration and inference. training.zip – detector-element stacks used to construct Noise2Noise training pairs in the notebooks. These files are intended for use with the Jupyter notebooks provided in the GitHub repository:https://github.com/redrodion/setkafluo The notebooks demonstrate the complete workflow for: exploring multi-element detector data, constructing Noise2Noise training pairs from detector-element splits, training the U-Net denoiser, benchmarking hyperparameter choices, and comparing denoised results with conventional approaches. How to Use This Dataset (for Google Colab) Step 1: Download the two archives provided here: input_data.zip training.zip Step 2: Create the following folder on your Google Drive (top level): MyDrive/setkafluo_demo/ Step 3: Unzip both archives inside this folder. After extraction, you should have: MyDrive/setkafluo_demo/input_data/ MyDrive/setkafluo_demo/training/ Step 4: From the SetkaFluo GitHub repository, copy the following into Google Drive: the five Jupyter notebooks, and the libs/ directory Place them in this exact location: MyDrive/setkafluo_demo/notebooks_and_library/ 01_data_exploration.ipynb 02_denoising_prep.ipynb 03_denoising_params.ipynb 04_denoising_main.ipynb 05_denoising_compare.ipynb libs/ Step 5: Open any of the notebooks in Google Colab. They expect the dataset to be located under MyDrive/setkafluo_demo/ and will run without modification. This structure reproduces the environment used in the companion publication and ensures seamless execution on free-tier Google Colab GPU machines.