[DeepSee] Deep learning single-shot in-line quantitative phase imaging
![[DeepSee] Deep learning single-shot in-line quantitative phase imaging](/var/www_mchtr/storage/images/badania-i-nauka/projekty/projekty-aktualne/deepsee-zastosowania-uczenia-glebokiego-w-jednoramkowych-poosiowych-technikach-ilosciowego-obrazowania-fazowego/44153-4-eng-GB/DeepSee-Deep-learning-single-shot-in-line-quantitative-phase-imaging_medium.png "[DeepSee] Deep learning single-shot in-line quantitative phase imaging")
project manager: dr inż. Maria Cywińska
project Implementation Period: 24.06.2025 r. - 23.06.2029 r.
allocated Funds for the Project: 1 195 356 zł.
funding Source: National Science Centre, Poland (Opus 28)
team: dr inż. Julianna Winnik (PW), mgr inż. Wiktor Forjasz (SD PW), inż. Wojciech Ogonowski (PW)
Partners:
- prof. Balpreet Ahluwalia group from the Arctic University of Norway in Tromso
- prof. Rosario Porras-Aguilar group from University of North Carolina in Charlotte
project description:
The project aims to push the boundaries of quantitative phase imaging (QPI) by developing single-shot phase retrieval algorithms for in-line interferometric microscopy. By shifting the complexity from optical hardware to software, the project introduces deep learning (CNN) and Bayesian inference methods for accurate and robust phase estimation from intensity-only data. The outcomes will enable compact, cost-efficient, and stable optical setups capable of real-time imaging of dynamic biological processes. DeepSee pioneers a universal AI-based approach to single-shot in-line QPI, independent of specific experimental systems or sample types.
