Tayebeh Bahador

Developing and benchmarking deep learning and foundation-model-based approaches for automated classification of corneal diseases from AS-OCT pachymetry maps.

Adapting foundation models (e.g., SAM) for vessel segmentation using parameter-efficient fine-tuning methods (e.g., LoRA) for domain adaptation in low-data medical imaging settings, developing reproducible and GPU-based pipelines, investigating transferability, domain robustness, and annotation efficiency.

• developed image processing and machine vision solutions, known as "vTools": https://docs.baslerweb.com/vtools • delivered web-based machine vision solutions • Developing and Maintaining CI/CD Pipelines • Implementing unit Tests and end-to-end tests • Coordinating Sprint Plannings • Conducting benchmarking of various libraries for Business decisions • Containerizing the application • Developing cloud solutions • Automating Cloud Infrastructure • Making architectural decisions for Software Solution

I developed and implemented self-supervised learning and transfer learning techniques to reduce the need for large labeled datasets in computer vision applications. By implementing Unsupervised Data Augmentation (UDA), I achieved up to a 5% improvement in test accuracy compared to the baseline self-supervised approach, reaching 85% accuracy with only 2,000 labeled images (vs. 90% with 50,000 labeled images in classical supervised learning). This project demonstrates how advanced learning methods can signifi

Medicine and Digitalization, Department of Neurology, University Hospital Magdeburg Topic: Automated segmentation of the locus coeruleus in aging and Alzheimer’s disease using deep learning - Segmentation of MRI images using ITK-SNAP tool at German Center for Neurodegenerative Diseases (DZNE)

As a research assistant, I worked with cutting-edge convolutional neural networks (CNNs) and various deep learning algorithms in Super-Resolution, enabling improvements in visual fidelity and detail reconstruction using non-medical data.

Medicine and Digitalization, Department of Neurology, University Hospital Magdeburg Topic: Super resolution and data imputation in MRI images using deep learning approaches As a Research Assistant, my work centered on leveraging deep learning techniques for advancements in medical imaging, specifically in super-resolution and data imputation for MRI scans. Key contributions included: - Implementing state-of-the-art deep learning models to enhance the resolution of MRI images, enabling more precise visuali

- Member in both Innovation group and R&D Marketing group - Attended surgeries, for instance Open-heart surgeries, Caesarean operations, Laparascopic surgeries, Urological Surgery and etc in numerous cities and hospitals - Collaboration with the specialists in surgeries in order to do researches on new modules and using the feedback in order to upgrade the modules. - Teach specialists and anesthesiologist how to use new devices and modules which are still under control of R&D Marketing department - Evalu

Endoscopic systems

• Tutored high school students in physics, reinforcing core concepts and helping improve academic performance. • Supported my former teacher post-graduation by reviewing and grading student exam papers. • Contributed to the development of his book in physics, by writing detailed example solutions and supporting content.