InTheLoop | 10.28.2013
CRD’s Peter Nugent One of Eight Speakers at Tonight’s 'Science at the Theater'
Peter Nugent, co-leader of CRD’s Computational Cosmology Center, as well as NERSC users Kristin Persson, Ben Bowen and Ian Hinchliffe, are among the eight speakers on tap for tonight’s Science at the Theater presentation. “Eight Big Ideas” will begin at 7 p.m. at the Berkeley Rep Theater, 2025 Addison St., in downtown Berkeley. Admission is free, but seating is extremely limited. The program will also be streamed live at: http://www.lbl.gov/LBL-PID/fobl/.
Nugent will discuss “Supercomputing and the search for supernovae.” Persson will talk about her Materials Project in a talk on “A Google for materials,” Bowen will present “Coming to a hospital near you: mass spectrometry imaging” and Hinchliffe will discuss “The Higgs and all that. How the universe works and why we should care.”
CRD’s John Shalf Delivers Opening Talk at Energy Efficient Systems Symposium
John Shalf, head of CRD’s Computer and Data Sciences Department and a leader of the GreenFlash project, gave the opening talk today at the Third Berkeley Symposium on Energy Efficient Electronic Systems. The symposium, held every other year, brings together researchers who are working on breakthrough improvements in energy efficiency for information processing systems. The meeting is being held in Sutardja Dai Hall on the UC Berkeley campus.
ESnet and Globus Online to Present Nov. 7 Webinar on Managing Big Data in Academia
Staff from ESnet and Globus Online are holding the second of a series of webinars aimed at helping HPC administrators at laboratories, colleges and universities tackle the problems and solutions to managing big data in academic settings. The one-hour webinar starts at 11 a.m. Thursday, November 7 (CST).
Presenters include David Lifka, Director of the Cornell University Center for Advanced Computing; Brock Palen, HPC system administrator and user consultant at CAEN-HPC at the University of Michigan; Steve Tuecke, Deputy Director of the Computation Institute at the University of Chicago and Argonne National Laboratory and Globus Project Co-Lead; and Jason Zurawski, Science Engagement Engineer at ESnet.
For more information, go to http://goo.gl/GXKqvB.
ESnet, Infinera and Brocade Demonstrate 100G Multilayer Network Optimization
ESnet has collaborated with networking vendors Infinera and Brocade in a successful demonstration of multi-layer networking using Software-Defined Networking (SDN) technologies. The demonstration, announced Oct. 15 at the SDN and OpenFlow World Congress in Bad Homburg, Germany, shows how SDN can be used to automatically provision services and optimize network resources, such as dynamically increasing or rerouting data center to data center interconnect bandwidth services, across a multi-layer network as traffic demands change.
SDN is an emerging field that makes it easier for software applications to automatically configure and control the various layers of the network. ESnet has been conducting localized tests to innovate, experiment and demonstrate the value of SDN when applied towards end-to-end support of data-intensive science collaborations and applications. If successful, SDN would give users a measure of predictability by giving them more control over their data flows.
Clarification: Brain Visualization Project a Team Effort by CRD Staff
An item in the Oct. 14 InTheLoop incorrectly gave sole credit to Daniela Ushizima for the lab’s work on a brain visualization project with UC San Francisco and Oblong Industries. The project was initially developed by Gunther Weber of the Visualization Group, and Aydın Buluç and Lenny Oliker of the Future Technologies Group. Weber did initial work on the prototype after obtaining the data and using VisIt to visualize the data. The work is also part of a recently approved LDRD project entitled "Graph-based analysis and visualization of multimodal multi-resolution large-scale neuroimaging data," led by Buluç. We apologize for the omission.
This Week’s Computing Sciences Seminars
Advances in Designing and Evaluating Ultrasound Imaging Systems for Breast Cancer Diagnosis
Wednesday, October 30, 9:30 - 10:30 a.m., Bldg. 50F, Room 1647
Nghia Q. Nguyen, Geophysics Group, Los Alamos National Laboratory
Abstract: Assessment of any medical imaging systems should be based on the task – the reason for the image being acquired. In this objective approach, the best performance of an imaging system can be measured using the Bayesian ideal observer analysis. Comparing this ideal performance with that of human observers, it is possible to predict the efficiency of a system for achieving its clinical design goals. Rigorous and objective assessments of image quality are now well developed for x-ray and gamma-ray imaging modalities but not for ultrasound.
The first part of the talk summarizes my Ph.D. thesis in ultrasonic instrument design and performance assessment for the task of breast cancer diagnosis. We first measure the task information using Kullback-Leibler divergence by combining ideas from information theory and acoustic scattering with medical practice. We then relate its performance metrics to fundamental engineering properties such as system noise, image contrast and spatial resolution, which has direct applications in system optimization. The ideal observer analysis also helps identify the greatest loss of task information at demodulation step of the image formation. Therefore, adaptive filtering algorithms of the echo signals were derived to improve the diagnostic performance for lesion feature discriminations.
The second part describes our recent work on ray-tracing reconstructions for the breast ultrasound tomography system. The method accounted for the ray bending inside the breast by solving the Eikonal equation using a finite-difference scheme. The ultimate goal is to develop a safer and more comfortable tomographic imaging device, using ultrasound waves instead of X-ray, to detect breast cancer in its earliest stages, which boosts the effectiveness of breast cancer screening modalities.
Segmentation of Complex Images: Detection of Thin and Ramified Structures using Markov Random Fields and Perceptual Information
Thursday, October 31, 9:30 - 10:30 a.m., Bldg. 50F, Room 1647
Talita Perciano Costa Leite, University of São Paulo, Brazil
Abstract: Line- curve-like, elongated and ramified structures are commonly found inside many known ecosystems. In biomedicine and biosciences, for instance, different applications can be observed. Therefore, the process to extract this kind of structure is a constant challenge in image analysis problems. However, various difficulties are involved in this process. Their spectral and spatial characteristics are usually very complex and variable. Considering specifically the thinnest ones, they are very “fragile” to any kind of process applied to the image, and then, it becomes easy the loss of crucial data. Another very common problem is the absence of part of the structures, either because of low image resolution and image acquisition problems or because of occlusion problems. This work aims to explore, describe and develop techniques for detection/segmentation of thin and ramified structures. Different methods are used in a combined way, aiming to reach a better topological and perceptual representation of the structures and, therefore, better results. Graphs are used to represent the structures.
This data structure has been successfully used in the literature for the development of solutions for many image processing and analysis problems. Because of the fragility of the kind of structures we are dealing with, some computer vision principles are used besides usual image processing techniques. In doing so, we search for a better "perceptual understanding" of these structures in the image. This perceptual information along with contextual information about the structures are used in a Markov random field, searching for a final detection through an optimization process. Lastly, we propose the combined use of different image modalities simultaneously. A software is produced from the implementation of the developed framework and it is used in two applications in order to evaluate the proposed approach: extraction of road networks from satellite images and extraction of plant roots from soil profile images. Results using the proposed approach for the extraction of road networks show a better performance if compared with an existent method from the literature. Besides that, the proposed fusion technique presents a meaningful improvement according to the presented results. Original and promising results are presented for the extraction of plant roots from soil profile images.