1.    Introduction
In-situ tomography experiments may impose limited angle data collection, because not all sample environments can be made X-ray transparent.  At PSICHE, the UToPEC (Ultra-fast Tomography Paris-Edinburgh Cell) for fast tomography at high pressure and temperature (<1s, >10GPa, >1000C) allows projections over 165° [1]. Resulting artefacts are shown in the figure (A). Various algebraic and/or iterative reconstruction schemes (the well-known ART and SART, but also for example Turpin [2]) have been developed to attempt to overcome these difficulties. For the growing UToPEC user community, we identify the need for a correction tool which is suited to this geometry, is easy to understand, and can be quickly adapted to different samples. It must be fast and efficient enough to be used during a synchrotron beamtime.  
2.    Materials and Methods
In the situation of limited angle tomography, a range of projection angles is missing in the sinogram. We propose here an iterative reconstruction method, inspired by [2]. This uses thresholding as simple and powerful regularization method, controlled via the histogram of reconstructed gray levels. This avoids computationally intensive morphological operations. The sinogram is completed using forward projection from the regularized volume. Different numbers of major and minor phases can be handled using appropriate algorithms. An acceptable result is usually reached after only a few iterations. This strategy has been implemented using the CT (Computed Tomography) algorithms of PyHST2 [3].
3.    Results and Conclusion
The figure (B, C) shows the effective correction obtained using four iterations of our method. This uses classical CT tools allowing it to be easily implemented in existing workflows [4], and with a computational cost less than artificial intelligence-based solutions.
4.    Acknowledgements
We would like to thank L. Turpin for valuable discussions. The UToPEc ongoing development programme is supported by DIM OxyMORE Île de France, Synchrotron SOLEIL and a CNRS-MITI grant.