Image Analysis of High-Resolution and High-Throughput Experiments

• Activity Number: 415
• Type: Topic Contributed
• Date/Time: Tuesday, August 6, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Imaging
• Author(s): Daniela Ushizima*+ and Andrea Bianchi and Hari Krishnan
• Companies: LBNL and Universidade Federal de Ouro Preto and LBNL
• Keywords: computer vision ; micro-CT ; 3D data ; MRI ; software

Abstract

Experimental science is facing hundred to thousand-fold increases in data volumes. While technologies with high-throughput emerge, reasoning and analysis of massive image datasets are lagging behind. In order to tackle it, analyses must be as automated as possible, and often tailored to specific science domains. This talk will describe a computational analysis workflow to access internal and external properties of geological samples using synchrotron X-ray tomography. Our workflow, Quant-CT, can differentiate porous media, and extract volume, surface area, granularity spectrum, porosity, among others. Quant-CT leverages algorithms from Fiji for preprocessing, segmentation, and analysis, while most of the visualization takes place on VisIt. Quant-CT supports quick user interaction, so that the user trains algorithms for automated parameterization. Quant-CT plugin is part of research on porous materials related to geologic sequestration of CO2. We will overview the material science problem, describe our implementation, and demonstrate our plugin on microCT. In addition, schemes used in Quant-CT can be extended to analyze other 3D imaging modalities as MRI and EM.