Fan-Yu (Ivy) Yen

PhD student at Department of Bioengineering, Northeastern University

prof_pic.jpg

All conditioned phenomena are like dreams, illusions, bubbles, shadows; Like dew and lightning. One should contemplate them in this way. ── from Vajracchedika Prajnaparamita Sutra (Diamond Sutra)

一切有為法，如夢幻泡影

如露亦如電，應作如是觀

──《金剛經》

Hello! I am Ivy, a PhD student in the Department of Bioengineering at Northeastern University, advised by Professor Qianqian Fang in the Computational Optics and Translational Imaging Lab. I hold a Bachelor’s and a Master’s degree in Biomedical Engineering from National Cheng Kung University (Taiwan).

Prior to my doctoral studies, I worked as a Research Assistant at the Biosignal and Neural Engineering Lab at National Cheng Kung University. I also worked as a Graduate Student Intern at the Fetal-Neonatal Neuroimaging Developmental Science Center at Boston Children’s Hospital, where I focused on neuroimaging and cerebral autoregulation research.

Research: My work lies at the intersection of artificial intelligence, optics, and biomedical imaging. I develop computational tools to make neuroimaging more accurate, efficient, and accessible. Highlights include:

I have developed an open-source tool — NeuroNavigatAR — to visualize optode/electrode positions based on the 10–20 (10–10, 10–5) system in real time using augmented reality (AR). This approach aims to revolutionize neuroimaging practices by improving ease of use, reducing setup time, and increasing the precision of optode or electrode placement. This is now published on bioRxiv and the toolbox can be found here.

In another line of my research, I have been working on extending Monte Carlo photon simulation techniques to better capture the complexities of biological tissues. By introducing curvature-aware modeling into our mesh-based Monte Carlo (MMC) toolbox, I made the simulations more realistic for tissues with complex boundaries.

I am also interested in applying deep learning models to solve practical questions. For example, I have used large language models (LLMs) to interpret user queries for the MCX toolbox, making it easier for entry-level users to run simulations by describing their setup in natural language. The abstract of this work can be found here. Moreover, I implemented a pipeline that uses LLMs as an interface to search across databases on NeuroJSON.io and automatically process the downloaded neuroimaging datasets. This work was presented here.

Course highlights:

  • High performance computing (CUDA, POSIX Threads, OpenMP, MPI)
  • Deep learning
  • Computational physics
  • Biomedical optics
  • Finite element analysis