Segment and Support: a dual-purpose Deep Learning solution for Limited Angle Holographic Tomography - dataset

This dataset constains training data of two types:

1. Synthetic volumes:
   Containing randomly generated and placed spheres of various sizes. The dataset was augmented by adding noise, rotation, holes, flipping, blurring and shifting. Data generated this way is treated as the input data. An output pair for each datapoint has been generated by inducing missing cone artifacts, by masking the 3D Fourier spectrum of each volume (to mimic an experimental scenario). 
2. Experimental data:
   This part of the dataset contains erroneous Direct Inversion reconstructions of cells [1], organoids [2], phantoms [3-4]. The mask has been generated through a thresholding step of a GTVIC [5] algorithm.

Each volume has the shape (128,128,128), and is of float32 precision. They are saved as .tiff files, which can be easily read via the tifffile Python library.

Additionally, the dataset contains:

• SnSNet_best_loss_model.pth - PyTorch file containing the weights of the model, whose loss was the lowest during training. 
• SnSNet_best_metric_model.pth - PyTorch file containing the weights of the model, whose metrics were the highest during training. 
  
• SnSNet_config.yml - YAML file containing all config information about the model, its training and dataset. It is necessary for inference and model recreation.
• Inference.py - Short Python script for quick inference. The input should be a floating point, erroneous DI reconstruction. It produces a binary mask of the object with corrected geometry and saves it into the same directory as the input file, with the prefix PRED_.
• reqs.txt - text file, containing the library requirements necessary for running Inference.py. It is compiled, such that the script performs the inference on GPU, using CUDA 11.8 on Windows 10. If the configuration does not enable running on GPU, it will run on CPU and will be indicated in the logs.
• training.xlsx - Excel file, containing data about the training process.

[1] M. Baczewska, W. Krauze, A. Kuś, P. Stępień, K. Tokarska, K. Zukowski, E. Malinowska, Z. Brzózka, and M. Kujawińska, “On-chip holographic tomography for quantifying refractive index changes of cells’ dynamics,” in Quantitative Phase Imaging VIII, vol. 11970 Y. Liu, G. Popescu, and Y. Park, eds., International Society for Optics and Photonics (SPIE, 2022), p. 1197008.
[2] P. Stępień, M. Ziemczonok, M. Kujawińska, M. Baczewska, L. Valenti, A. Cherubini, E. Casirati, and W. Krauze, “Numerical refractive index correction for the stitching procedure in tomographic quantitative phase imaging,” Biomed. Opt. Express 13, 5709–5720 (2022).
[3] M. Ziemczonok, A. Kuś, P. Wasylczyk, and M. Kujawińska, “3d-printed biological cell phantom for testing 3d quantitative phase imaging systems,” Sci. Reports 9, 1–9 (2019).
[4] M. Ziemczonok, A. Kuś, and M. Kujawińska, “Optical diffraction tomography meets metrology — measurement accuracy on cellular and subcellular level,” Measurement 195, 111106 (2022).
[5] W. Krauze, P. Makowski, M. Kujawińska, and A. Kuś, “Generalized total variation iterative constraint strategy in limited angle optical diffraction tomography,” Opt. Express 24, 4924–4936 (2016).