Asst. Prof. Berna Morova, from Physics Engineering Department, was an invited speaker at the 2025 International Conference on Light and Light-Based Technologies (ICLLT), where she presented recent advancements in fiber-based fluorescence imaging systems.

The research centers on the use of optical fiber bundles—thin, flexible tools made of thousands of optical fibers—to transfer images from small or hard-to-access environments. These bundles were produced using the stack-and-draw method and included thermally matched, custom-made glass materials that provide high numerical apertures (up to 0.61), allowing for better light collection and sharper imaging.

Seven different fiber bundle designs were fabricated and tested. Their performance was evaluated using brightfield and fluorescence imaging of biological and patterned samples. It was shown that bundles with slightly larger core sizes and spacing produced better image clarity and contrast.

To go beyond the physical limits of fiber-based imaging, a deep learning-based enhancement method was introduced. A convolutional neural network was trained to convert standard widefield fluorescence images into enhanced versions, matching the quality of those produced with more complex and costly structured illumination systems. The contrast-to-mean ratio (σ/μ) of the images improved significantly—demonstrating that artificial intelligence can effectively upgrade image quality without the need for additional hardware.

This work presents a promising direction in the development of portable, high-resolution, and low-cost fluorescence imaging tools. The combination of specially designed optical fiber bundles and AI-powered image processing could lead to new capabilities in biomedical research, diagnostics, and compact endoscopic systems. Future efforts aim to extend this platform to live tissue imaging and to refine the AI algorithms for real-time applications.