Development of robust tomographic model-based reconstruction-segmentation methods for X-ray macromolecular crystallography

The Rosalind Franklin Institute has significant interest in the fields of automated or semi-automated image segmentation (partitioning a digital image into multiple segments). This is primarily due to it being a crucial technique in many practical fields such as medical imaging, material science, object detection, recognition and others. This project focuses on a collaboration between the Franklin and Diamond Light Source to deliver a reliable automated image reconstruction and segmentation pipeline for routine processing of tomographic data at Diamond’s long-wavelength macromolecular crystallography beamline. The main difficulty resides in segmenting very weak contours of image data with low contrast and image artifacts. The project will focus on various aspects of iterative image reconstruction, edge detection, denoising and other image processing tasks required to deliver more accurate segmentation. The developed segmentation pipeline will be integrated into the Diamond software infrastructure, making it possible to process the significant quantities of data collected as fast as possible.