Synchrotron Tomography and the use of AI for better 3D Images

Shyam Pulickan

Post-Doctoral Researcher, PSICHE & ANATOMIX, Synchrotron SOLEILs

Le 04 juin 2026 à 13h30, Amphi Manet

Abstract: 

Progress in scientific instrumentation and technology brings a higher flux of data which is not often talked about but forms an increasing challenge in all subjects (“data deluge”). Synchrotron tomography is such a domain which produces a significant amount of data, often measured in terabytes per day of experiments, and the analysis of the resulting images needs significant resources and time.  In addition to such constraints, tomograms (3D images reconstructed from a series of 2D radiographs) can suffer from all the issues you can think of for photography, for example movement of the specimen, dirty lens, inadequate lighting, obstructed field of view etc. While most of these can be corrected semi-automatically to some extent, for the amount of data produced, this is becoming increasingly difficult. One possible solution is to implement Artificially Intelligent algorithms to take care of correcting the artefacts in the data, so that we can focus on the scientific questions. 

At Synchrotron SOLEIL within the scope of the Franco-Allemand ANR project AIQuAM3D we are exploring ways to improve quality assurance in synchrotron tomography. The possibilities of using AI algorithms to fix the problems in tomography are explored, and new solutions are developed which requires least human intervention possible, yielding an efficient workflow from data acquisition to publications. An example showing the potential of an AI algorithm for correcting a tomogram with ‘missing wedge’ artefacts (created due to obstructed field of view) is shown in Figure 1. Measures are taken to ensure the quality and reliability of the AI treated data that we provide to our user community. In this talk we look at some examples from these developments and describe how these tools fit into the workflow of the instrument and the precautionary measures to take while dealing with AI algorithms.

Biography: 

Shyam PULICKAN did his Masters in Aeronautics and Space with specialisation in High Temperature Materials at ISAE-ENSMA, Poitiers in 2021. His Master thesis involved data analysis on lab scale X-Ray tomographic data on 3D printed samples. 

He got his PhD in 2024 in Mechanics and Material Science on the topic ‘Wire Arc Additive Manufacturing (WAAM) Characterisation and Performance Assessment Modelling under Uncertainties’ from the Université de Technologie de Troyes in collaboration with ENSAM of Metz. The thesis involved the development of statistical models to predict the geometry of WAAM beads with data from experimental fabrication. 

He is currently employed at Synchrotron SOLEIL as a post-doctoral researcher and is working on developing algorithms to correct the artefacts in Synchrotron X-Ray µ-CT at the beamlines PSICHE & ANATOMIX.