I am a postdoctoral researcher at the DASIV SmartState Center, where I am working on methods for the simulation of scanning transmission electron microscopy (STEM) images. This work is part of an ARO grant and in collaboration with Dr. Peter Binev, Dr. Douglas Blom, Dr. Michael Matthews and Dr. Thomas Vogt. Ultimately, the goal of this project is to extract some 3D information about the atomic columns of a specimen from a single 2D STEM image.

During my doctoral studies at RWTH Aachen in Germany, I was working on the mathematical foundations of a variational method for exit wave reconstruction for conventional transmission electron microscopy (TEM). The outcome of this research were new theoretical results as well as new numerical methods for the reconstruction of the exit wave. This work is presented in my thesis.

Education

• Doctorate: Mathematics, RWTH Aachen, Germany, 2020
• M.Sc.: Mathematics, RWTH Aachen, Germany, 2015
• B.Sc.: Mathematics, RWTH Aachen, Germany, 2014

Experience

• 2021 – Present: Postdoc, DASIV Center, University of South Carolina
• 2020: Postdoc, Collaborative research center SFB 1394, RWTH Aachen
• 2016 – 2019: Doctoral student, AICES graduate school, RWTH Aachen
• 2015 – 2016: Research Assistant, AICES graduate school, RWTH Aachen
• 2014 – 2015: Student Assistant, Lehrstuhl D für Mathematik, RWTH Aachen

Teaching

I have been involved in teaching since my graduate studies in 2014. First, as a student assistant leading small exercise groups for linear algebra courses, then as a doctoral student in a course on functional analysis, occasionally as a substitute in image processing courses and regularly as an assistant in oral exams.

In Spring 2022, I will be teaching the class "Introduction to Deep Neural Networks" at UofSC. This is not a regular lecture-type class, but more like an informal seminar. As such, there will be no exams or quizzes; instead, presentations and discussions are an integral part of the class. There will be a strong focus on the application and the main goal of this class is to construct a neural network for the recognition of handwritten digits, a task which is very difficult to solve using classical methods.

Research interests

Mathematical image processing, variational methods and inverse problems, in particular in the context of scanning and conventional transmission electron microscopy (STEM and TEM). I particularly enjoy making use of the mathematical concepts that I learned during my studies in real-world applications.

Publications

• T. Pinetz, E. Kobler, C. Doberstein, B. Berkels, A. Effland. Total Deep Variation for Noisy Exit Wave Reconstruction in Transmission Electron Microscopy, SSVM (2021), 491-502. DOI: https://doi.org/10.1007/978-3-030-75549-2_39
• C. Doberstein. Joint Exit Wave Reconstruction and Multimodal Registration of Transmission Electron Microscopy Image Series (Dissertation), RWTH Aachen (2020). DOI: https://doi.org/10.18154/RWTH-2020-06672
• C. Doberstein, B. Berkels. A least-squares functional for joint exit wave reconstruction and image registration, Inverse Problems 35 (2019), 054004. DOI: https://doi.org/10.1088/1361-6420/ab0b04