|
The Visual Computing Forum, or VCF, is a
series of seminars organized by the visualization
group with selected talks from the fields of
visualization, image processing, computer graphics,
and so on. The individual seminars are arranged
approximately once a month, on Fridays from 11am to
12am, and they will be interleaved with the MedViz seminars.
They will be held either at the Høyteknologisenteret
or at the VilVite
Science Center. If you wish to be informed about upcomming VCF events, please write an e-mail to "vcf.bergen@gmail.com", "Helwig.Hauser@UIB.no" or "Sergej.Stoppel@UIB.no".
|
|
December 12, 2014
Place: MedViz Facilities., Møllendalsbakken 7, 5th floor
Time: Friday December 12, 2014, from 12.00 to 13.30
Talk 1: Fast volumetric super-resolution reconstruction of organs from
moving subjects
Speaker: Bernhard Kainz
Abstract:
Abstract: Magnetic Resonance Imaging (MRI) is a primary tool for
clinical investigation of the brain and fetal organs. High resolution
imaging with volumetric coverage using stacks of slices or true three
dimensional (3D) methods is widely available and provides rich data for
image analysis. However such detailed volumetric data generally takes
everal minutes to acquire and requires the subject to remain still or
move only small distances during acquisition. Fetal organ imaging
introduces a number of additional challenges. Maternal breathing may
move the fetus and the fetus itself can and does spontaneously move
during imaging. These movements are unpredictable and may be large,
particularly involving substantial head and body rotations. Motion
correction methods have revolutionized MRI of the fetus by
reconstructing a high-resolution 3D volume of fetal organs from such
motion corrupted stacks of 2D slices. Such reconstructions are valuable
for both clinical and research applications. However, reconstruction is
computationally expensive and can only be performed off line.
Information about the accuracy of the scan and potential uncertainties
is unknown or not considered in the clinical practice. In this talk I
will discuss the fundamentals of fetal MRI reconstruction and it's
parallelization and hardware acceleration for a future on-line
application during the scan. Furthermore, I am looking forward to a
discussion about potential application of novel visualization
techniques to communicate varying uncertainties of the reconstruction
to examining radiologists and scientists.
Talk 2: New perfusion analysis algorithms for MR and ultrasound
Speakers: Torfinn Taxt and Radovan Jirik
Abstract:
Good estimates of the microvascular functional (perfusion) parameters
in normal and pathological tissues is based on a) good estimates of the
local arterial input functions (aifs), b) adequate tissue residual
function models (trfs) and c) tissue contrast signals with reasonable
signal to noise ratios. This talk will present some recent developments
to obtain reliable aif estimates in MR and ultrasound using blind
deconvolution methods. A presentation of pharmacokinetic trf models for
leaky capillaries will also be given.
Talk 3: Deriving Anatomical Context from 4D Ultrasound
Speaker: Ivan Viola
Abstract:
Real-time three-dimensional (also known as 4D) ultrasound imaging using
matrix array probes has the potential to create large-volume
information of entire organs such as the liver without external
tracking hardware. This information can in turn be placed into the
context of a CT or MRI scan of the same patient. However for such an
approach many image processing challenges need to be overcome and
sources of error addressed, including reconstruction drift, anatomical
deformations, varying appearance of anatomy, and imaging artifacts. In
this work, we present a fully automatic system including robust
image-based ultrasound tracking, a novel learning-based global
initialization of the anatomical context, and joint mono- and
multi-modal registration. In an evaluation on 4D US sequences and MRI
scans of eight volunteers we achieve automatic reconstruction and
registration without any user interaction, assess the registration
errors based on physician-defined landmarks, and demonstrate realtime
tracking of free-breathing sequences.
Additional material:
Flyer
|
|
|
November 28, 2014
Video Visualization: An Overview
Speaker: Andrea Brambilla (Visualization Group, University of Bergen)
Place: Room 3137, floor 3, HIB (data blokk), Thormøhlensgate 55
Time: Friday November 28, 2014, from 10.15am to 11.15am
Abstract:
Videos are one of the most widespread media for
collecting, communicating and archiving
information. Nowadays, acquiring videos is a
relatively straightforward process, and this
explains their success in the context of
entertainment, surveillance, sport events, and so
on. On the other hand, watching and extracting
information from a video stream is a lengthy
process. Automatic techniques are only partially
sucessful because of the intrinsic complexity of
this kind of data. Video visualization is a
growing research field which aims at easying the
study of video data. It relies on both automatic
techniques and user interaction, exploiting the
best of both worlds. In this talk, I will
introduce this field, focusing on its evolution
from computer vision. I will discuss the main
challenges and present an overview of the
state-of-the-art. The talk will conclude with a
discussion of the open problems and the expected
future developments...
Additional material:
Flyer,
Slides
|
|
|
October 31, 2014
Interactive Visual Steering Analysis of Complex Engineering-Simulation
Ensembles
Speaker: Kresimir Matkovic (VRVis Research Center, Vienna, Austria)
Place: Room 3137, floor 3, HIB (data blokk), Thormøhlensgate 55
Time: Friday October 31, 2014, from 10.00am to 11.00am
Abstract: In this talk we propose
a novel approach to interactive visual steering
of simulation ensembles. A simulation ensemble is
a collection of simulation runs of the same
simulation model using different sets of control
parameters. Due to increasing pressure on car
manufacturers, for example, to meet new emission
regulations, to improve efficiency, and to reduce
noise, both simulation and visualization are
pushed to their limits. We describe a successful
realization of a tightly coupled steering loop,
integrating new simulation technology and
interactive visual analysis in a prototyping
environment for automotive industry system
design. By coupling interactive visualization
with the simulation back-end - computational
steering, it is now possible to quickly prototype
a new system, starting from a non-optimized
initial prototype and the corresponding
simulation model. We introduce two kinds of
simulation steering: a model refinement approach,
and a hybrid visual steering. The ability to
early see the first results from a
multidimensional simulation space - thousands of
simulations are run for a multidimensional
variety of input parameters - and to quickly go
back into the simulation and request more runs in
particular parameter regions of interest
significantly improves the prototyping process
and provides a deeper understanding of the system
behavior. The hybrid steering adds an automatic
optimization in order to support a region of
interest selection in the high-dimensional
parameter space. The excellent results and a very
positive feedback from domain experts which we
achieved for the common rail injection system
strongly suggest that our approach has a great
potential of being generalized to other, similar
scenarios.
After the seminar, Geir Smestad
(master student in visualization at UiB) will defend his master thesis,
entitled Interactive Visual Analysis of Streaming Data.
Additional material:
Flyer
|
|
|
September 19, 2014
Place: Store Auditorium, floor 2, HIB (data blokk), Thormøhlensgate 55
Time: Friday September 19, 2014, from 14.00 on
Talk 1: Collaboration in Multi-User Virtual Reality
Speaker: Prof. Dr. Bernd Fröhlich (Bauhaus-Universität in Weimar, Germany)
Abstract:
Immersive telepresence allows distributed groups
of users to meet in a shared virtual 3D world.
Our approach uses two coupled projection-based
multi-user setups, each providing multiple users
with perspectively correct stereoscopic images.
At each site, the users and their local
interaction space are continuously captured using
a cluster of registered depth and color cameras.
The captured 3D information is transferred to the
respective other location, where the remote
participants are virtually reconstructed in
life-size. Local and remote users can jointly or
independently explore virtual environments and
virtually meet face-to-face for discussions. We
structure collaborative activities of collocated
and remote users using Photoportals. Virtual
photos and videos serve as threedimensional
references to objects, places, moments in time
and activities of users. They can be shared among
users and serve as portals to the captured
information. Our Photoportals also provide access
to intermediate or alternative versions of a
scenario and allow the review of recorded task
sequences that include life-size representations
of captured users.
Talk 2: Parallel Computing Towards Exascale
Speaker: Dr. André Brodtkorb (SINTEF, Oslo)
Abstract:
Today's computers are increasingly parallel, and
mapping algorithms to these architectures has
become increasingly difficult. Failing to adapt
to the architecture leaves our performance in the
dark ages, with little hope of scaling on today's
and tomorrows hardware. This tutorial-like talk
discusses the underlying parallel hardware
constraints, different approaches to parallelism,
a hands-on example for computing PI, and the use
of domain specific languages to hide complexity.
Additional material:
Flyer,
Prof. Fröhlich's webpage,
Dr. Brodtkorb's slides
|
|
|
August 29, 2014
Visualization of Functional Medical Data
Speaker: Mattia Natali (Visualization Group, University of Bergen)
Place: Room 3137, floor 3, HIB (data blokk), Thormøhlensgate 55
Time: Friday August 29, 2014, from 10.15am to 11.15am
Abstract:
Medicine is more and more supported by
technology, for instance in clinical analysis and
medical intervention. Medical imaging, that
non-invasively reveals internal anatomy and
physiology of a body, plays an important role
since radiography was introduced. In the last
thirty years, structural modalities have been
placed side by side new image modalities that
detect physiological activities within a certain
tissue or organ, providing what is known as
functional medical data.
Nowadays, a large amount
of data is obtained with different modalities. To
offer a deeper insight into this data,
visualization can be helpful. In this talk, an
overview on different medical image modalities is
presented, as well as corresponding techniques to
turn the data into visual information that a
physician can interpret for diagnosis. We will go
through well known modalities such as computed
tomography (CT) and magnetic resonance imaging
(MRI), but also more specific medical image
technologies for functional imaging, where the
activation of body regions is measured during
metabolic or cognitive processes. Finally,
multi-modality visualization is investigated to
provide functional medical data within its
anatomical context. Therefore we will see an
improvement in tissue characterization due to
simultaneous imaging of morphological and
functional information, such as the integration
of MRI and Positron Emission Tomography (PET)
into one single hybrid system.
Additional material:
Flyer
|
|
|
May 26, 2014
searchCrystal and Set Visualization
Speaker: Prof. Anselm Spoerri (School of Communication and Information, Rutgers University)
Place: Room 3137, floor 3, HIB (data blokk), Thormøhlensgate 55
Time: Monday May 26, 2014, from 14.15am to 15.15am
Abstract:
This talk will address how set data can be visualized using
searchCrystal, which is a toolset of coordinated visualizations that
can be used to compare, remix and share set data. It will be shown how
searchCrystal has been used to visualize meta search data and what is
popular or controversial in Wikipedia.
Additional material:
Flyer,
Anselm Spoerri's webpage
|
|
|
April 04, 2014
Molecular dynamics simulations for biomolecules: principles and
examples
Speaker: Prof. Nathalie Reuter (Dept. of Molecular Biology, UiB)
Place: Room 3137, floor 3, HIB (data blokk), Thormøhlensgate 55
Time: Friday April 04, 2014, from 10.15am to 11.15am
Abstract:
Molecular Dynamics (MD) simulations are widely used to model
biomolecules in conditions that mimic the ones of in vitro experiments.
The raw result of a MD simulation is a trajectory file containing the
position of each atom in the molecule at every time step of the
simulation, similar to a movie representing how the biomolecule is
wiggling and giggling along simulation time. MD simulations are
sometimes referred to as a 'computational microscope' because they
allow the investigation of biomolecules dynamics at the atomic level of
detail, something that is virtually impossible with the experimental
methods currently available.
I will introduce the basic assumptions and principles behind MD
simulations for biomolecules; molecular mechanics force fields,
equations of motion and their integration, treatment of long-range
interactions, etc... Further I will present the practicalities of
running MD simulations and the different steps from system setup to
trajectory post-processing.
Additional material:
Flyer,
Reuter Group's webpage
|
|
|
March 07, 2014
State-of-the-art ultrasound blood flow imaging
Speaker: Lasse Løvstakken (Dept. Circulation and Medical imaging, NTNU)
Place: Room 3137, floor 3, HIB (data blokk), Thormøhlensgate 55
Time: Friday March 07, 2014, from 10.15am to 11.15am
Abstract:
The introduction of real-time ultrasound color-Doppler imaging in the
mid-eighties was a major breakthrough for bedside diagnosis of
cardiovascular disease. Currently this technique allows for real-time
2D and 3D imaging of blood flow, both for detection of blood and for
the quantification of the blood velocity. The use of color-Doppler
imaging in clinical practice is, however, mostly qualitative, used to
localize but not quantify abnormal flow patterns. The reason is mostly
related to the current limitations of color-Doppler related to low
frame rates, beam-to-flow angle-dependencies, and a limited measurable
velocity span. Further, the fundamental information and color
visualization has not changed substantially since its introduction 30
years ago. Currently a technological leap is on the verge in medical
ultrasound imaging. The possibility of real-time transfer and
processing of channel data and software image formation allows for
significantly improved image quality and frame rates in general, as
well as a higher accuracy in blood flow imaging. This includes improved
possibilities for imaging low flow in small vessels, and the estimation
of the blood velocity vector as shown in the image above showing
circular flow patterns in the left ventricle of a neonate. I will in
this lecture introduce the current color-Doppler imaging modality and
its limitations, and present on-going research projects on the future
state of the art in ultrasound blood flow imaging.
Additional material:
Flyer,
Lasse Løvstakken webpage
|
|
|
February 07, 2014
Applied Visualization and modeling of environment and geology
Speakers: Veronika Šoltészová (CMR Computing / UiB) and Saman Tavakoli (CMR Computing)
Place: Room 3137, floor 3, HIB (data blokk), Thormøhlensgate 55
Time: Friday February 07, 2014, from 10.15am to 11.15am
Abstract:
Computing is a research department of Christian Michelsen Researc (CMR)
for applied research in visualization, data analysis and decision
support. CMR computing works closely together with industry and
government and develops innovative solutions and software prototypes in
application areas such as energy, oil&gas, maritime, marine and
medicine. In this talk, we will present selected projects that are
related to modeling and visualization for geology and environment.
* Decision support for offshore wind turbine installation (DECOFF)
Offshore operations such as the installation, maintenance and repairs
of wind turbines are complex and to a large extent weather sensitive.
The cost of such operations is to a high degree determined by waiting
for convenient time for weather-sensitive phases (transportation of
equipment, mooring, crane operations, etc.). In this project, we are
developping a decision support tool which is based on real physical
limitations of the equipment being used and which takes into account
uncertainties such as weather conditions.
* Virtual CO2 Laboratory (VIRCOLA)
The rapidly advancing deployment of geological CO2 storage requires a
better comprehension of the CO2 storage reservoirs, CO2 injection, and
long-term fate of geologically stored CO2. In this cross-disciplinary
project, we address the challenge by building a virtual CO2 laboratory
for co-visualizing and visual analysis of different data types involved.
* Geoillustrator
This project focuses on illustrative geology. We will present a
prototype that allows for intuitive and quick creation of geological
models via sketching. The prototype supports modeling of layers,
faulting.
Additional material:
Flyer,
Slides,
CMR Computing Website
|
VCF seminars in
2017,
2016,
2015,
2014,
2013,
2012,
2011
|
|