I am a researcher working at the intersection of deep learning, microscopy, and optical imaging. My work spans OCT multilayer simulation, physics-informed neural models and SRGAN-based super-resolution for sub-resolution structure estimation.
- Implemented forward simulation of multilayer samples
- Used geometric optics + synthetic A-scan/B-scan generation
- Simulated reflectivity, refractive index changes, and layer thickness
- Created training datasets for deep learning boundary detection
- Model predicting number of layers (0–5 or 0–9)
- Sub-resolution boundary localization
- Refractive index estimation using both TensorFlow & PyTorch
- Classical + DL hybrid models with genetic algorithms
- Built SRGAN from scratch in TensorFlow 2.x
- Trained on bright-field microscopy of fly wings
- Implemented VGG19 perceptual loss + adversarial loss
- GPU-accelerated training on NVIDIA RTX 3060
- Achieved PSNR: 26 dB & SSIM: 0.85
Python, TensorFlow, PyTorch, MATLAB, NumPy, SciPy,
OpenCV, scikit-image, Git, Linux, Jupyter.
MATLAB + Python model for multilayer OCT forward simulation
Deep models for structural analysis in OCT
Super-Resolution of microscopy images using SRGAN
➡️ https://github.com/NargesRezaei/Microscopy-SRGAN-FlyWing-GPU.git
- Biomedical imaging
- Microscopy
- Super-resolution & GANs
- OCT & optical modeling
- Physics-informed deep learning
- Inverse problems
- Improving the quality of microscopic images using generative adversarial network
Poster presentation, Iranian Optics & Photonics Conference (2024)
📧 nargesrezaei654@gmail.com
🔗 LinkedIn: