Speaker
Description
Synchrotron-based X-ray phase-contrast imaging (X-PCI) enables non-destructive, high-resolution three-dimensional visualisation of soft-tissue microstructure at sub-micron scales, offering unique insights into tissue organisation that are inaccessible with conventional histology [1]. However, the analysis and clinical interpretation of synchrotron datasets in research and clinical contexts require overcoming significant practical barriers: large data volumes, computationally intensive pipelines, complex heterogeneous 3D morphology, and the challenge of integrating imaging findings with clinical data. We present a part of the TransCOR Image Analysis Platform, developed at Universitat Pompeu Fabra, as a cloud-based research environment connected to a high-performance computing cluster, designed to support the researcher from the moment data leaves the synchrotron without requiring local infrastructure or computational expertise. The platform is modality- and organ-agnostic and is currently being developed and used to analyse cardiac and liver tissue imaged at the TOMCAT beamline (Swiss Light Source) and the FaXToR beamline (ALBA Synchrotron).
An overview of the platform is shown in Figure 1 (in attachments). The platform offers several integrated capabilities: (i) preprocessing tools including bit-depth conversion and contrast stretching for inter-sample intensity normalization; (ii) a web-based 3D visualiser for interactive inspection via orthogonal views, volume rotation, and bounding box cropping; (iii) AI-based segmentation models, such as nnUNet [2] or Segment Anything [3]; (iv) quantification of tissue-specific morphological features, including fiber orientation [4] and vascular network morphometry; (v) ML-based analysis for unsupervised representation learning via patch-based UMAP embedding and interactive clustering [5]; (vi) joint analysis linking imaging findings to longitudinal clinical data structured around patient events; and (vii) population analysis combining imaging and clinical data for multi-sample statistical and ML-based studies.
Together, these capabilities provide an online pipeline from reconstructed synchrotron data to interpretable, clinically contextualised results. The TransCOR platform is under active development and is being used to analyse post-transplant endomyocardial biopsies [6] and porto sinusoidal vascular disease liver biopsies [7], to establish a general-purpose system for synchrotron-based tissue research across institutions and organ systems.
References
[1] Dejea H, Scientific Reports. 2019.
[2] Isensee F, Nature Methods. 2021.
[3] Kirillov A, ICCV. 2023.
[4] Brunet J, Journal of Open Source Software. 2026.
[5] Seabra M, CASEIB 2024. 2024.
[6] Planinc I, Transplant International. 2023.
[7] De Gottardi A, The Lancet Gastroenterology & Hepatology. 2019.