When
Where
Title
“Brain cancer diagnosis using lens-based multicore-fiber endoscopy with a learning-based digital twin”
Abstract
Rapid diagnosis is crucial to cancer treatment. Multi-core fibers (MCF) enable access to deep tissue. Deep neural network (DNN)-based image reconstruction can depixelate structural artifacts and enhance spatial resolution beyond physical limits. However, limited available tissue samples for training (not enough patients for research studies) make DNN-assisted fiber endoscopy for in vivo brain cancer diagnosis challenging. We propose using a digital twin.
A dataset is simulated for pre-training a MCF reconstruction network. Then transfer learning is applied to the network on a few ten experimental image pairs of glioblastoma samples. The results show, even when transfer learning on only 10 experimental image pairs, the mean structural similarity SSIM of the test set can be increased from 0.547 to 0.932. The reconstructed images can achieve a comparable resolution with widefield microscopes when increasing the data size to 50, which are hundreds of times less than the training from scratch. The digital twin provides an efficient and highly feasible approach towards minimally invasive all-optical biopsies in neurosurgeries.
Bio
Juergen W Czarske (Fellow EOS, OPTICA, SPIE, IET, IOP) is director and full chair professor of the TU Dresden, Germany. His awards include the 2019 OPTICA Joseph-Fraunhofer-Award/Robert-M.-Burley-Prize in Optical Engineering and the 2024 SPIE Dennis Gabor Award in Diffractive Optics. Juergen fosters talented students early. The students and members of his lab have won over 100 prizes, including Bertha-Benz award of Daimler Benz Foundation (10,000 Euro). Juergen is Vice President of International Commission for Optics, ICO, and was the general chair of the world congress of ICO with 3 Nobel laureates. Juergen also serves on the editorial boards of Light Science and Applications, Advanced Photonics and Light Advanced Manufacturing.