STROBE - UC Berkeley/Advanced Light Source
Summer Undergraduate Research Scholar
In Summer 2021, I was a STROBE Sunder Undergraduate Research Scholar. During this experience, I developed interactive Jupyter Notebooks that implement Python-based spectromicroscopy and PIC mapping analyses to give x-ray microscope users at the Advanced Light Source (ALS) access to resources at the NERSC supercomputing center. A summary of my work is presented in the poster and description below.
The visual I presented at the ALS User Meeting Poster Slam, a competition for undergraduate and graduate students across the country. I placed second for my presentation, titled "Zooming in On STXM Analysis."
Experiments at the Advanced Light Source’s COSMIC Beamline produce large volumes of data that must be filtered and analyzed to gather meaningful scientific conclusions. Recent exciting examples include investigations into the chemical composition of LiFePO4 batteries used in electric vehicles, as well as the biomineralization process of biogenic materials that will deepen our understanding of evolution-driven materials design and provide biomimetic inspiration. Until now, a pystxm graphical user interface (GUI) developed by STROBE researcher David Shapiro has been used to analyze these images. However, the program is difficult to access and modify by visiting scientists, which takes time from their experiments and limits the depth of their preliminary analysis. To address these limitations, I have developed a new Python-based Jupyter notebook interface utilizing the ipywidgets library and implemented it with NERSC supercomputing resources so that users can easily modify the code and access the program on the web. The new user interface, shown partially in Figure 1, has similar features as the original GUI, including image alignment, pre-edge removal, region of interest selection, and optical density spectrum plotting. More advanced functions, such as Principal Component Analysis and PIC mapping, are also available and have been tested successfully. In the future, other image analysis techniques, such as Non-Negative Matrix Factorization, can be easily added.
The developed Jupyter notebook also allows users to easily adapt their sample-dependent workflow and add their own analysis code if needed. I also added many interactive components, including sliders, buttons, and dropdown menus, as shown in Figure 1. This allows users to manipulate their data easily and compare the results of different analyses. Moving forward, a similar format will be implemented to perform ptychographic tomography analysis, which requires powerful computing resources. In addition, a notebook for a recently developed technique that allows orientational analysis of optical anisotropic materials, such as calcium carbonate-based minerals, has been developed and successfully tested.