BASE Lab BASE LAB
  • Research
  • People
  • Publications
  • Articles
  • Opportunities
  • Contact

People

Principal Investigator

James Le Houx

Senior Lecturer, University of Greenwich · Faraday Institution / STFC Emerging Leader Fellow

James founded the BASE Lab in 2025. His research develops autonomous X-ray imaging platforms for energy materials, combining synchrotron experiments with AI and high-performance simulation. He holds a PhD from the University of Southampton and is a Chartered Engineer (CEng, IMechE). Before academia, he was a design engineer in motorsport (Xtrac Ltd.) and a beamline scientist at Diamond Light Source.

He leads the UK’s Battery Imaging BAG at Diamond Light Source, serves on the ISIS Science Council, and advises DSIT on AI data standards for materials science. He developed the Faraday Engine, a national initiative connecting UK Exascale computing to national facilities.

· · ·


PhD Students

Calum Green

Applications of Deep Learning for Super-Resolution and Data Fusion of 3D X-ray Imaging Datasets for Materials Characterisation

Imperial College London · Co-supervised with Prof. Daniele Dini and Dr Sharif Ahmed · Completing July 2026

Calum develops generative super-resolution methods and multi-modal data fusion pipelines for synchrotron X-ray imaging. His work includes EXG-GAN, a generative adversarial network for upsampling spatially resolved 3D tomographic data, and methods for fusing XRD-CT and XCT datasets for autonomous materials segmentation. He is building OpenLSR-X, an open-source PyTorch package for learnable super-resolution of XCT data.

Available for postdoctoral or industry positions from July 2026. Contact Calum or James to discuss.

Key publication: Three-dimensional, multimodal synchrotron data for machine learning applications — Scientific Data (2025)


Jon Valijonov

Data-Driven Image Mechanics (D2IM): A Deep Learning Approach to Predict Displacement and Strain Fields in Biological Tissues from X-ray Tomography

University of Greenwich · Co-supervised with Prof. Gianluca Tozzi and Dr Peter Soar

Jon is developing D2IM, a deep learning framework that predicts full-field displacement and strain directly from undeformed X-ray tomography images. His initial results show that direct strain prediction significantly outperforms displacement-derived approaches, reducing false-positive high-strain classifications by 75%. The project is expanding D2IM to a range of hard and soft biological tissues and loading scenarios, using Greenwich’s InCiTe 3D X-ray microscope for sub-micron resolution phase-contrast imaging.

First preprint: D2IM-Strain (bioRxiv, 2026)


Isabella Florez

Quantum-AI Synergy for Next-Generation Imaging of Biological Tissues

University of Greenwich · Co-supervised with Prof. Gianluca Tozzi and Dr Peter Soar

Isabella is investigating quantum-native representations of biological tissue images to enable AI operations — classification, segmentation, measurement, and multi-modal data fusion — that are intractable with classical approaches alone. Her work surveys hybrid quantum-classical architectures for biomedical imaging and is developing toward Quantum Image Correlation (QuIC), a quantum extension of digital volume correlation for high-dimensional volumetric data.


Conor Wright

Leveraging Plant Biomechanics under Hostile Environments: Modelling Root Penetration and Fracture with SRXCT

University of Southampton · Co-supervised with Dr Siul Ruiz

Conor’s research investigates the underpinning biomechanical processes of plant roots growing in highly compacted, high-strength environments to inform regenerative farming practices. To capture these dynamics in situ, he combines Synchrotron X-ray Computed Tomography (SRXCT) and coupled, spatially resolved X-ray diffraction (XRD) with Digital Volume Correlation. This multimodal approach allows him to infer root pressures and map localised soil mechanics, which feed into the development of predictive finite element (FE) models. His work simulates both direct root tip penetration and crack propagation in surrogate and natural soils, modelling them as elastoplastic von Mises materials with finite strain theory to create digital twins of root-soil interactions.

Key presentation: Understanding root biomechanics in high-strength environments- assessing the feasibility of penetration and fracture FE models with SRXCT — EGU General Assembly (2026)


Grad Students

Harriet Jones

Massively Parallel Workflows for Image-based Modelling

STFC Daresbury Lab · Co-supervised with Dr Charles Moulinec and Dr Stefano Rolfo

Harriet is integrating code_saturne, a highly optimised, open-source engineering modelling framework, into our semi-automated image-based modelling workflow. Her research focuses on utilizing Compatible Discrete Operator (CDO) schemes to simulate complex physics within porous materials directly from X-ray tomography data. By automatically generating body-fitted meshes from voxel datasets via a mesh blanking process, she bypasses the need for external meshing software. For her early-impact science case, Harriet is simulating flow fields to understand the optimisation of bilayer electrodes for Lithium-ion batteries, utilizing tomography data acquired at Diamond Light Source.

Key presentation: Massively Parallel Workflows for Image-based Modelling — Scientific Computing Department Poster


MSc Students (2025–26)

Asif Ali

Generative AI for Digital Twins of Battery Electrodes

University of Greenwich

Asif is training diffusion models on 3D X-ray tomograms of anisotropic battery electrodes (carbon felt) to generate synthetic microstructures that are statistically indistinguishable from real data. The generated volumes will serve as inputs for physics-based simulation with OpenImpala, addressing the data scarcity bottleneck in computational materials science. Running on Isambard-AI.


Ali Butt

The Battery Explorer: Interactive Web-Based Dashboard for 4D Scientific Data

University of Greenwich

Ali is building an interactive web application for exploring 4D battery tomography data — scrubbing through time, rotating 3D volumes, and viewing synchronised derived quantities (e.g., crack volume vs. cycle number). The tool is designed to make large-scale operando datasets accessible to experimentalists without specialist visualisation software.


Insha Amreen

Guided Super-Resolution for Multimodal Scientific Imaging

University of Greenwich

Insha is developing guided super-resolution methods that use high-resolution structural data (XCT) to intelligently upsample low-resolution chemical maps (XRD-CT), producing unified high-resolution datasets containing both structural and chemical information. The work builds on the FeatUp framework and our group’s multimodal synchrotron dataset.


Alumni

Emily Lu

Heuristic Operando Experimentation and Standardisation of Battery Sample Environments

University of Cambridge · Faraday Institution FUSE Intern (Summer 2025)

Emily is an Engineering undergraduate at the University of Cambridge who joined the BASE Lab as a Faraday Institution FUSE intern. Her research addressed the critical lack of standardisation, reliability, and reproducibility in battery characterisation across national facilities. During her project, she conducted a comprehensive review of in situ/operando sample environments at the Rutherford Appleton Laboratory (RAL). To bridge the gap toward commercial applications, she built the Operando Library, an open-source, interactive web registry of electrochemical cells available across UK central facilities. This machine-readable database serves as the hardware ground truth for our group’s heuristic operando framework, which proposes using AI and digital twins to actively steer beamlines to predict and capture transient failure events.

Interactive tool: Operando Sample Environment Registry

Key publication: Autonomous battery research: Principles of heuristic operando experimentation — arXiv (2025)

Back to top

© 2025 BASE Lab · Made with Quarto

  • Accessibility

  • License

  • · Design inspired by Silvia Canelón