Projects
EasyFiji (Fiji Plugin Suite)
Technologies: Java, Swing, ImageJ, Fiji
A comprehensive GUI-based image processing environment designed to make advanced microscopy analysis accessible to researchers without programming expertise.
Key Features: - User-friendly interface built with Java Swing - Integrated image processing workflows - Deployed on ImageJ update site for easy installation - Supports batch processing and automation - Designed for biological image analysis
Impact: Enables researchers to perform complex image analysis tasks without writing code, accelerating research workflows.
Multi-Agent "AI Scientist" for Bioimaging
Technologies: Python, LangChain, RAG (Retrieval-Augmented Generation), OpenAI API
An intelligent multi-agent system that automates scientific research tasks in bioimaging.
Capabilities: - Literature Search: Automatically searches and summarizes relevant papers - Workflow Design: Suggests optimal image analysis pipelines based on research goals - Scientific Review: Provides critical analysis and recommendations - Knowledge Integration: Uses RAG to combine domain knowledge with current literature
Architecture: Implements multiple specialized AI agents that collaborate to solve complex research questions, similar to having a team of research assistants.
Distributed Training for Radiomics + ViT Fusion
Technologies: PyTorch, PyTorch DDP, MONAI, HPC (DGX A100), SLURM
A high-performance deep learning framework for medical image analysis combining radiomics features with Vision Transformers.
Technical Highlights: - Distributed Data Parallel (DDP) training across multiple GPUs - Integration with MONAI for medical imaging workflows - Deployed on NVIDIA DGX A100 HPC cluster - Fusion architecture combining traditional radiomics with transformer-based features - Optimized for large-scale medical imaging datasets
Performance: Achieves significant speedup through distributed training while maintaining model accuracy for cancer imaging applications.
Deep Learning for Light-Sheet Microscopy
Technologies: Python, PyTorch, TensorFlow, OpenCV
Research projects applying deep learning to improve light-sheet fluorescence microscopy image quality.
Publications: - Autofocus Enhancement: Deep learning-based autofocus that improves image quality in real-time - Illumination Correction: Neural network approach to correct illumination angles during acquisition
These methods are published in Biomedical Optics Express and have been adopted by other research groups.
Open Source Contributions
I actively contribute to the scientific imaging community through: - Fiji/ImageJ plugin development - Python packages for image analysis - Documentation and tutorials for researchers - Code sharing on GitHub
Interested in collaboration? Get in touch!