Our work sits where advanced imaging, full-field measurement and machine learning meet. We image specimens while they deform, extract quantitative fields from those volumes, and train models that predict behaviour rather than merely describe it.
Themes
Four directions
Distinct methods, one shared pipeline from specimen to model.
Tissue mechanics from tomography
In‑situ X-ray tomography of bone and soft tissue under load, with digital volume correlation and learned models that infer internal deformation from greyscale alone.
Flagship model · D2IM
Hyperspectral imaging for healthcare
The spectral signature of tissue carries biochemical state alongside structure. We build diffusion-based architectures for classification and tumour detection.
Models · DiffSpectralNet, MedDiffHSI
AI-driven biomaterial design
Predicting and optimising the hierarchical self-assembly of nanoparticles into macroscale soft biomaterials, including formulation with diffusion-language models.
Bio-inspired engineering
Quantum representation of images
Quantum-native representations of tissue images, enabling efficient classification, segmentation, measurement and multimodal fusion in the quantum space.
Quantum-AI synergy
Models
Published architectures
Named models from the lab, with the papers behind them.
Selected figuresFrom the published work below
D2IMDeformation from greyscale
Predicts bone deformation directly from X-ray tomography greyscale, without an explicit correlation step. Under active extension with segmentation models and large language models.
Interdisciplinary work in progress across the lab and its affiliates.
AI digital volume correlation of hard & soft tissueVision transformers for tissue modellingDiffusion architectures for data-driven mechanicsCollaborative diffusion for multi-modal image analysisBiomaterial formulation with diffusion-language modelsBone biomimetic materials via AI sonificationMeasurement in the quantum space
