BE-LIGHT Improving BiomEdical diagnosis through LIGHT-based technologies and machine learning

Projektinformationen

BE-LIGHT

ID Finanzhilfevereinbarung: 101119924

DOI

10.3030/101119924

EK-Unterschriftsdatum 10 Juli 2023

Startdatum 1 Oktober 2023

Enddatum 30 September 2027

Finanziert unter

• Marie Skłodowska-Curie Actions (MSCA)

Gesamtkosten

Keine Daten

EU-Beitrag

€ 2 525 220,00

Koordiniert durch

UNIVERSITAT POLITECNICA DE CATALUNYA
Spain

Periodic Reporting for period 1 - BE-LIGHT (BE-LIGHT Improving BiomEdical diagnosis through LIGHT-based technologies and machine learning)

Berichtszeitraum: 2023-10-01 bis 2025-09-30

Zusammenfassung vom Kontext und den Gesamtzielen des Projekts

Photonics is crucial for maintaining and increasing the competitiveness of EU industry, and light-based instruments are fundamental for non-invasive diagnostics and treatment of diseases, delivering huge economic benefits for national health systems. Artificial intelligence (AI) and machine learning (ML) have demonstrated their capability to classify and identify patterns in data, achieving reliability levels that are comparable to those of experienced physicians.
To fully develop the potentials of these technologies, the BE-LIGHT project will provide top level training to 11 researchers in photonic technologies (multispectral imaging, eye-tracking, optical coherence tomography (OCT), multispectral optoacoustic tomography (MSOT), super-resolution microscopy and optogenetics) that will be complemented with AI and ML, computational imaging and modelling. A close-to-the-clinic approach will allow them to advance the field of biomedicine, with the ultimate goal of improving the understanding, diagnosis and treatment of diseases in 3 main research areas: visual health, cardiovascular health and analysis of cells and tissues (related to Parkinson’s and rare diseases) using super-resolution microscopy.
BE-LIGHT brings together an interdisciplinary consortium of several academic institutions, internationally recognized hospitals and an ecosystem of private companies with complementary know-how in photonics, microscopy, AI, medical instruments and clinical research, to promote inter-sectoral synergies.
BE-LIGHT’s multi-skill training program will provide the researchers with the complete set of skills that are nowadays key for their success in both academic and non-academic sectors. They will gain a broad understanding of how state-of-the-art light-based technologies work, a solid knowledge of machine and deep learning (DL) and data processing, and will obtain experience of clinical and commercialization processes. BE-LIGHT will open for them a wide range of high-quality job opportunities.

Arbeit, die ab Beginn des Projekts bis zum Ende des durch den Bericht erfassten Berichtszeitraums geleistet wurde, und die wichtigsten bis dahin erzielten Ergebnisse

All 11 researchers have been successfully recruited. During the reporting period, the consortium has advanced substantially on both training and research objectives. Two major schools (S1 in Göttingen, Germany, and S2 in Gdańsk, Poland) and a Hands-on Training on DL (HoT4) were organized, complemented by outreach events such as the Baltic Science Festival. An Ethics Advisor was appointed and the first Ethics Report was submitted, ensuring compliance with the required standards.
At the scientific level, significant progress has been achieved in all 3 Work Packages:
WP1 (Eye diseases and vision): New retinal processing models were developed; a webcam-based eye tracker and system for glaucoma screening were validated; multispectral imaging and ML improved retinal disease classification; and a first prototype of a tomographic micro-vibrography system combining OCT and ultrasound was built.
WP2 (Cardiovascular diseases): Optogenetics + ML methods for arrhythmia control were demonstrated; key nonlinear features for early cardiac event prediction were identified; a multimodal RGB–thermal imaging system and STREAM-NET for remote pulse estimation were created; and advanced MSOT reconstruction, plaque segmentation, and physics-informed ML methods were developed.
WP3 (Microscopy): Advances included new super-resolution image-formation frameworks, improved cryogenic/expansion microscopy workflows, and a GAN-based fluorescence-restoration method, alongside new high-quality microscopy datasets.
The scientific production has been consolidated through several publications in high-quality journals, including Scientific Reports, Knowledge-Based Systems, Nature Photonics, Advanced Materials, and Annals of Clinical and Translational Neurology, reflecting the strong scientific progress of the consortium.
Several congress communications and early publications highlight the project’s substantial scientific progress and innovative methodologies. Results have been actively disseminated through BE-LIGHT’s website, social media channels, and a dedicated Zenodo subcommunity within the EU Open Research Repository, ensuring broad visibility and accessibility.

Fortschritte, die über den aktuellen Stand der Technik hinausgehen und voraussichtliche potenzielle Auswirkungen (einschließlich der bis dato erzielten sozioökonomischen Auswirkungen und weiter gefassten gesellschaftlichen Auswirkungen des Projekts)

The project has already delivered results that significantly advance the current state of the art across its three scientific domains.
• WP1 has advanced vision science by creating more efficient computational frameworks for retinal modelling and introducing innovative diagnostic methods that combine multispectral imaging with phasor analysis.
• WP2 has shown that integrating optogenetics with ML enables new strategies for predicting and controlling arrhythmias, complemented by multimodal imaging and DL models that improve non-invasive cardiovascular assessment.
• WP3 has pushed microscopy to nanometre-scale resolution through expansion methods and MIET (metal-induced energy transfer), while pioneering generative ML models that restore fluorescence channels and expand rare disease datasets, addressing the challenge of limited annotated data.
The project’s advances open pathways for future clinical applications—including glaucoma screening, oximetry, cardiovascular monitoring, and rare disease diagnostics—while providing new computational and imaging tools that strengthen Europe’s leadership in biophotonics. The results have strong potential for further uptake through continued research, industrial collaboration, and integration into clinical workflows, with standardization playing a key role. Early publications and conference contributions ensure strong international visibility, and public engagement activities have helped communicate the project’s outcomes to wider audiences.

3D reconstruction of a cataract in the eye of a patient.

Computational super-resolution images and 3D reconstructions of collagen VI in the fibroblasts.

Multispectral optoacoustic tomography for non-invasive diagnostics of carotid artery disease.

OCT analysis of the eye at IMO.

Fluorescently labeled tubulin network in a fixed HELA cell.