Home About NAKFI Conferences Grants Communication Awards Contact Us My Account

Login to manage your account and access the NAFKI Alumni Network.

Password Reminder








Dr. Tom Vogt Publishes “Smart Data Acquistions for Nanoscale Imaging”
Dr. Vogt has provided an extract for his paper that describes without mathematics why optimized imaging using non-linear registration is important.
 
The advances made over the past decades in modern imaging devices offer access to unprecedented physical and chemical information. However, widely used image processing methodologies by far fall short of exploiting the full breadth of information offered by numerous types of scanning probe, optical, and electron microscopies. Due to spatial uncertainties caused by complex motion and image distortion during the acquisition process, the insufficiencies of commonly used off-the-shelf methods are usually attempted to be compensated by human visual inspection.
 
In order to extract more reliable information from images of biological or chemical materials, where the need to minimize specimen damage due to ionizing radiation or prolonged measurement protocols results in intrinsically noisy data with low signal-to-noise ratio, new and more sophisticated data processing methodologies are needed. Here we outline how data assembly of several low dose frames based on a noise-robust non-rigid pixel-to-pixel registration method morphing pixel from one frame to the next allows us to optimally average noisy low-dose image frames, thereby retrieving substantially more information from the full image area.
 
We have developed a strategy for extracting an increased level of information from a series of low dose STEM images, rather than using single high dose images, in order to circumvent or at least significantly ameliorate a buildup of unwanted physical artifacts, contortions, and damage caused during the acquisition process. We have applied the methodology to beam sensitive materials like siliceous zeoliteY. A crucial ingredient is a non-linear registration process that removes visual inspection and human interaction. This represents an important step forward since the huge amounts of data created by many modern imaging techniques employed in astrophysics, medical imaging, process control and various forms of microscopy call for an early and reliable triage before storage. In particular, in many of these areas change detection is crucial and critically hinges on an accurate and reliable registration. We provide algorithms and metrics that allow researchers to extract larger amounts of meaningful information from data which often are costly to acquire. The quality of the final reconstruction hinges on the quality of the initial input data. In the case considered here of a beam sensitive zeolite under low-dose conditions, significant improvement in the information content of the reconstruction was demonstrated relative to the individual input frames without artificially “improving” the areas in the field of view which either were not in focus in the input frames or suffered significant beam damage during the series acquisition.
Optimized imaging using non-rigid registration Ultramicroscopy
(to appear in ULTRAMICRSCOPY)
Benjamin Berkels a, Peter Bineva,b, Douglas A. Blomc, Wolfgang Dahmena,d, Robert C. Sharpleya,b, Thomas Vogt a,c,e
a Interdisciplinary Mathematics Institute, 1523 Greene Street, University of South Carolina, Columbia, SC 29208, USA
bDepartment of Mathematics, 1523 Greene Street, University of South Carolina, Columbia, SC 29208, USA
cNanoCenter, 1212 Greene Street, University of South Carolina, Columbia, SC 29208, USA
d Institut für Geometrie und Praktische Mathematik, RWTH Aachen, Templergraben 55, 52056 Aachen, Germany
eDepartment of Chemistry and Biochemistry, 631 Sumter Street, University of South Carolina, Columbia, SC 29208, USA