A Review of Some Data-Driven Modeling Applications Fueled by Optical Microscopy
Speaker: Christopher Calderon, PhD; Ursa Analytics, Inc.
Host: Ted Randolph
Tuesday, May 2, 2023 — 2:45 p.m., JSCBB A108
Abstract
The common theme underlying my research over the last 20+ years has been the development and application of new data-driven modeling techniques to help understand complex systems encountered in science and engineering. Here, “data-driven modeling” refers to combining computational statistics and function approximation to calibrate models from observed measurements. In this talk, I will review some of my recent efforts related to extracting quantitative descriptive information from large collections of image data. In the first part of this talk, I will review techniques for estimating physics-based models from time-ordered image stacks of fluorescently tagged molecules measured in live cell environments. More specifically, I will review data-driven stochastic differential equation (SDE) modeling techniques where nonlinear SDEs are calibrated from in vivo single particle tracking measurements in crowded sub-cellular structures like the primary cilium and endoplasmic reticulum. In the second part of this talk, I will discuss some neural network-based approaches for characterizing particles (e.g., protein aggregates) commonly encountered in industrial biopharmaceutical formulations; applications here range from quality control to formulation screening. Throughout this talk, I will highlight and discuss some of my personal experiences in interdisciplinary data-driven modeling.
Biosketch
Christopher Calderon received his BS in chemical engineering from Purdue University and his PhD in chemical engineering from Princeton University. He was an NSF VIGRE postdoctoral fellow jointly appointed to the Departments of Statistics and Computational & Applied Mathematics at Rice University. Calderon is currently the founder and president of Ursa Analytics, Inc., he is a co-founder and current board member of the Quantitative Bioimaging Society and also currently has an affiliate appointment at CU Boulder. Calderon currently directs an R&D company focused on developing new computational statistics and machine learning algorithms applied to image analysis and signal processing applications. His company participates in academic collaborations focused on developing new algorithms and software for characterizing microscopy data. Industrial efforts at Ursa Analytics are focused on developing supervised and unsupervised machine learning algorithms deployed on embedded hardware.