Division of Scientific Computing
Department of Information Technology
X-ray free electron lasers can be used to image individual biological particles, by exposing them to a pulse strong enough to turn the sample into a plasma. Thanks to the short pulse length, the images represent an almost undamaged structure despite this fact.
My research has been focused on understanding noise in this context. There is noise from the experimental setup, but also inherent physical noise from the fact that a discrete number of photons are sampled. I will discuss the matter of noise and the modeling needed in two main contexts: identifying the correct image frames when the signal-to-noise ratio is far below 0 dB (more noise than signal), and reconstructing 2D projection images based on constrained linear systems.
I spent two years in the Molecular Biophysics group and their partners, but this project also highlights more general issues in statistical modeling and data processing for high-volume scientific infrastructures, such as genome sequencing and high-throughput optical imaging.