Internal:CNF Workshop: Difference between revisions

From crtc.cs.odu.edu
Jump to navigation Jump to search
Pthomadakis (talk | contribs)
CNF, external, internal
1,027 edits
Created page with "# CNF Workshop August 19-20/2019"
 
Ctsolakis (talk | contribs)
BrainMapping, CNF, HPC, I2MConversion, NDAM, Segmentation, UltraSound3D, Bureaucrats, external, internal, Administrators
2,152 edits
No edit summary
 
(20 intermediate revisions by 2 users not shown)
Line 1: Line 1:
# CNF Workshop August 19-20/2019
= 08/19/19 Presentations =
''' Slides https://odu.app.box.com/s/4yfdokkhyyqge2zx1jlv8gctw4zw6w2q '''
== Nikos Chrisochoides ==
 
* Leverage from computational methods from medical imaging for processing phase space distributions in Nuclear Femtography
* Enable next generation process simulations
* Bring jobs to Virginia
* “Usefulness of Useless Knowledge” by Flexner
** Look for standards and learn lessons
 
== Bob McKeown ==
 
* Femtography 2019 ML & Revealing the internal structure of Nucelon & Nuclei
* See pic JLab Mission
* Nuclear Femtography
** High luminosity
** Polarized beams and targets
** Sophisticated detector systems
* Nature publications
** Ultrafast Nucleons in Asymmetic Nuclei
** …
* Virginia Center for Nyclear Femtography
** 7/14 joint lab/university initiatives funded
* Proposed JLab electron Ion Collider (see pic)
* EIC Users Group (see pic)
* EICUG 2019
* Current Status and Path forward of EIC (see pic
* JLab has electron collider, NY Brookhaven has ion collider
** Both need to construct the other half
* Imaging the Nucleon is the future of Nuclear Physics
 
== Gail Dodge ==
 
* Partnership between ODU and JLab in 90’s
* 12 tenure track faculty
* 6 JLab professors
* ODU faculty intellectual leadership in research programs in JLab
* Bring together scientists from different fields
** Bridging the gap between physics, math, and computer science to reach the potential
* ODU has hired ~6 data science faculty
* Need help to learn how to involve people from different disciplines
** Different fields working on analyzing data, imaging protons/neutrons, and extracting scientific insights
 
== David Richards ==
 
* “Understanding the Glue That Binds Us All: The Next QCD Frontier in Nuclear Physics”
* See slide 6 for those that submitted proposals
* Dealing with incomplete data and the importance of visualization
* Synergy across fields
** Astrophysics
** Visualization
** Machine Learning
* Call for proposal late September
** External Evaluation Committee
* CNF Symposium at ODU early 2020
 
== Christian Weiss ==
 
* Imaging technologies in computation
* Subnuclear systems
** Scales
** Scattering experiments
* Workflow inmeasurements
* Tesselation
* Scales
** Organisms- Cells
** Molecular - Atoms
** Nuclear - Subnuclear
* Mission: Explore subnuclear systems
** Strong interaction : quarks,gluons
** Understand proton neutron as emergent phenomena
*** interacting large systems develop complex structure that is not present in parts
* Interest
** Basic building blocks of matter
** nuclear structure and reactions, nuclear energy
* Subnuclear systems
** Unusual dynamical system
*** contents of proton can't be localized because they are in "perpetual movement"
** Relativity: Energy = mass - relativity, conversion processes. Proton is ‘cauldron’ filled with radiation energy and matter/antimatter
** Only average properties are measurable: densities,correlations
*** formulation/visualization possible with appropriate tools
* Scattering experiments
** Electrons collide on protons
** Scatterings are gathered from detectors
** Incident energy determines spatial resolution
** JLab 12GeV accelerators
*** needed for resolving proton's internal structure
*** 1GeV is typical spatial size of a proton
** Information Recorded:
*** Distribution based on type and momentum -> Phase space distributions
*** The “expression” of proton’s internal structure
*** 10^6 - 10^9 experiments in JLab
* Workflow: Measurements
** Theory: Model of proton structure (parameters that need to be verified experimentally)
** Predictions for scattering probabilities (cross section)
** Phase space distributions (meeting ground between theory and experiments)
** Reconstruction based on the raw data (detector effects)
** Scattering Experiments -> Raw Data (kinematic coverage statistics)
** Simulations (can be run with actual or mock data to test reconstruction /feasibility)
** Modify theoretical model based on observations
** "Final" picture of proton from synthesis of multiple experiments and theoretical models
* Phase space distributions
** Use imaging techniques(e.g tessellation, slicing, viewing)
** 10^6 -10^9 experiments
** View distributions, projections, cuts, features
** Compare distributions
** Interpolate refine distributions for simulations
** With ODU
*** Tesselated representation
*** compact memory-efficient representations
*** Efficient interpolation/refinement for MC simulations
*** Visualize the gamma ray points that were detected by the detector + the density of energy
* Proton Structure:
** Imaging for vizualitation of subnuclear system
** Image concept for relativistic quantum system
** Synthetic image from theory and multiple experiments
** Spatial size, internal motion of quarks
** Tesselated representation
 
== Ron Kikinis ==
* Making data machine readable
** Critical for effective/accessible data
* DICOM (standard for medical images)
* 3D Slicer
* Browser based versions
** ohif.org
** openanatomy.org
* Look at data from different vantage points
* SlicerAstro
** colors to show data most helpful to doctors, students, etc.
** Use of Slicer3D in astronomy
* Punzo et al 2015
 
== Alexandra Golby ==
 
* In the past doctors would infer where the problem wass by the symptoms. They would then open the brain and see if they were right
* Now: Brain organization -> functional connectivity
* Functional MRI measures blood flow
* Diffusion Tensor Imaging the only method for imaging tensors
* Combine fMRI with MEG or fMRI with MRI
* VR to prepare for operation
* AR to help annotate points of interest on the brain
* The relationship in the brain of where things are locate and how they relate to other places in the brain is important
 
== Steve Pieper ==
* New software - Keep using the best software we have
* Open source - Reproducible science
* 3D Slicer with
** Python
** Tensorflow
** Numpy, pandas, scipy…
** Wrapper API
** Extensions marketplace
** VR Support
** Remote Rendering
** Jupyter Notebook support
* OHIF Viewer
** Browser based medical image viewer
** VTK.js
** Crowds-cure.org
** GPU Segmentation (Grow cut algorithm)
 
 
== Tina Kapur ==
 
* Discourse Forum
* Project Week Event
** [https://projectweek.na-mic.org/ projectweek.na-mic.org]
** upcoming one takes place January 20-24, 2020 in Las Palmas, Gran Canaria, Spain
** next one takes place June 22-26,2020 at MIT in Boston, USA
 
== Andriy Fedorov ==
 
* “Your source code is your data”, NVIDIA CEO
** For reproducibility for example
* National Cancer Institutr Quantitative Imaging Network
** Goal: Transalte quantitative imaging methods and algorithms as clinical decision support tools into clinical utility
* DICOM includes metadata that can help to interpret images
** Unique identifiers
** Helps to query your data
** TCIA (Cancer Imaging Archive) :Collection of medical images
*** can submit image sets for analysis
*** need big data to be accessible
* NCI Cancer Research Data Commons
 
= 08/20/19 Presentations =
 
== Nikos Chrisochoides ==
 
* Lessons from yesterday
** DICOM type
** 4D Meshes for JLab, Medical imaging
** Very nice presentation by Christian Weiss
** Cross-department funding for another workshop?
 
== James V. Koch ==
 
* Risk: You know the probabilities
* Uncertainty: You don’t
* Privately profitable vs. society
* Recent inovations started from individuals, not big companies
* National Labs work better when in cooperation with universities
* Lesson learnt: Get people together and let them work on what they think
* Profit motive makes people more productive
* When to sell intellectual property, since imitations will arise soon
* Lessons
** Outside funders are skeptical
*** they look at the diffusion curve (uncertainty)
*** they want equity
** Need substantial amount of equity to participate
** Very high discount rates because of uncertainty
** Investors require control
** Try to get some benefits for the decision makers early on since they will not be there for too long
*** Don't sell out before benefits
 
== James Cheng ==
 
* VA institutions very spread around
* Companies also
* Angel investments recently appeared in south VA
** 757 Angels - Hampton Roads Investing
** CVA, 1717, Startup Virginia, NRV - Richmond
** CAN, CAV Angels - Charlottesville
* Resources on the slides
** TEConomy Study
** SBIG - small business investments
** VEDP
** VRIF Awards
** OTA - way Gov't gets funding
* Want scalable technology
** tie software to a machine/product for intellectual property protection

Latest revision as of 16:15, 21 October 2019

08/19/19 Presentations

Slides https://odu.app.box.com/s/4yfdokkhyyqge2zx1jlv8gctw4zw6w2q

Nikos Chrisochoides

  • Leverage from computational methods from medical imaging for processing phase space distributions in Nuclear Femtography
  • Enable next generation process simulations
  • Bring jobs to Virginia
  • “Usefulness of Useless Knowledge” by Flexner
    • Look for standards and learn lessons

Bob McKeown

  • Femtography 2019 ML & Revealing the internal structure of Nucelon & Nuclei
  • See pic JLab Mission
  • Nuclear Femtography
    • High luminosity
    • Polarized beams and targets
    • Sophisticated detector systems
  • Nature publications
    • Ultrafast Nucleons in Asymmetic Nuclei
    • …
  • Virginia Center for Nyclear Femtography
    • 7/14 joint lab/university initiatives funded
  • Proposed JLab electron Ion Collider (see pic)
  • EIC Users Group (see pic)
  • EICUG 2019
  • Current Status and Path forward of EIC (see pic
  • JLab has electron collider, NY Brookhaven has ion collider
    • Both need to construct the other half
  • Imaging the Nucleon is the future of Nuclear Physics

Gail Dodge

  • Partnership between ODU and JLab in 90’s
  • 12 tenure track faculty
  • 6 JLab professors
  • ODU faculty intellectual leadership in research programs in JLab
  • Bring together scientists from different fields
    • Bridging the gap between physics, math, and computer science to reach the potential
  • ODU has hired ~6 data science faculty
  • Need help to learn how to involve people from different disciplines
    • Different fields working on analyzing data, imaging protons/neutrons, and extracting scientific insights

David Richards

  • “Understanding the Glue That Binds Us All: The Next QCD Frontier in Nuclear Physics”
  • See slide 6 for those that submitted proposals
  • Dealing with incomplete data and the importance of visualization
  • Synergy across fields
    • Astrophysics
    • Visualization
    • Machine Learning
  • Call for proposal late September
    • External Evaluation Committee
  • CNF Symposium at ODU early 2020

Christian Weiss

  • Imaging technologies in computation
  • Subnuclear systems
    • Scales
    • Scattering experiments
  • Workflow inmeasurements
  • Tesselation
  • Scales
    • Organisms- Cells
    • Molecular - Atoms
    • Nuclear - Subnuclear
  • Mission: Explore subnuclear systems
    • Strong interaction : quarks,gluons
    • Understand proton neutron as emergent phenomena
      • interacting large systems develop complex structure that is not present in parts
  • Interest
    • Basic building blocks of matter
    • nuclear structure and reactions, nuclear energy
  • Subnuclear systems
    • Unusual dynamical system
      • contents of proton can't be localized because they are in "perpetual movement"
    • Relativity: Energy = mass - relativity, conversion processes. Proton is ‘cauldron’ filled with radiation energy and matter/antimatter
    • Only average properties are measurable: densities,correlations
      • formulation/visualization possible with appropriate tools
  • Scattering experiments
    • Electrons collide on protons
    • Scatterings are gathered from detectors
    • Incident energy determines spatial resolution
    • JLab 12GeV accelerators
      • needed for resolving proton's internal structure
      • 1GeV is typical spatial size of a proton
    • Information Recorded:
      • Distribution based on type and momentum -> Phase space distributions
      • The “expression” of proton’s internal structure
      • 10^6 - 10^9 experiments in JLab
  • Workflow: Measurements
    • Theory: Model of proton structure (parameters that need to be verified experimentally)
    • Predictions for scattering probabilities (cross section)
    • Phase space distributions (meeting ground between theory and experiments)
    • Reconstruction based on the raw data (detector effects)
    • Scattering Experiments -> Raw Data (kinematic coverage statistics)
    • Simulations (can be run with actual or mock data to test reconstruction /feasibility)
    • Modify theoretical model based on observations
    • "Final" picture of proton from synthesis of multiple experiments and theoretical models
  • Phase space distributions
    • Use imaging techniques(e.g tessellation, slicing, viewing)
    • 10^6 -10^9 experiments
    • View distributions, projections, cuts, features
    • Compare distributions
    • Interpolate refine distributions for simulations
    • With ODU
      • Tesselated representation
      • compact memory-efficient representations
      • Efficient interpolation/refinement for MC simulations
      • Visualize the gamma ray points that were detected by the detector + the density of energy
  • Proton Structure:
    • Imaging for vizualitation of subnuclear system
    • Image concept for relativistic quantum system
    • Synthetic image from theory and multiple experiments
    • Spatial size, internal motion of quarks
    • Tesselated representation

Ron Kikinis

  • Making data machine readable
    • Critical for effective/accessible data
  • DICOM (standard for medical images)
  • 3D Slicer
  • Browser based versions
    • ohif.org
    • openanatomy.org
  • Look at data from different vantage points
  • SlicerAstro
    • colors to show data most helpful to doctors, students, etc.
    • Use of Slicer3D in astronomy
  • Punzo et al 2015

Alexandra Golby

  • In the past doctors would infer where the problem wass by the symptoms. They would then open the brain and see if they were right
  • Now: Brain organization -> functional connectivity
  • Functional MRI measures blood flow
  • Diffusion Tensor Imaging the only method for imaging tensors
  • Combine fMRI with MEG or fMRI with MRI
  • VR to prepare for operation
  • AR to help annotate points of interest on the brain
  • The relationship in the brain of where things are locate and how they relate to other places in the brain is important

Steve Pieper

  • New software - Keep using the best software we have
  • Open source - Reproducible science
  • 3D Slicer with
    • Python
    • Tensorflow
    • Numpy, pandas, scipy…
    • Wrapper API
    • Extensions marketplace
    • VR Support
    • Remote Rendering
    • Jupyter Notebook support
  • OHIF Viewer
    • Browser based medical image viewer
    • VTK.js
    • Crowds-cure.org
    • GPU Segmentation (Grow cut algorithm)


Tina Kapur

  • Discourse Forum
  • Project Week Event
    • projectweek.na-mic.org
    • upcoming one takes place January 20-24, 2020 in Las Palmas, Gran Canaria, Spain
    • next one takes place June 22-26,2020 at MIT in Boston, USA

Andriy Fedorov

  • “Your source code is your data”, NVIDIA CEO
    • For reproducibility for example
  • National Cancer Institutr Quantitative Imaging Network
    • Goal: Transalte quantitative imaging methods and algorithms as clinical decision support tools into clinical utility
  • DICOM includes metadata that can help to interpret images
    • Unique identifiers
    • Helps to query your data
    • TCIA (Cancer Imaging Archive) :Collection of medical images
      • can submit image sets for analysis
      • need big data to be accessible
  • NCI Cancer Research Data Commons

08/20/19 Presentations

Nikos Chrisochoides

  • Lessons from yesterday
    • DICOM type
    • 4D Meshes for JLab, Medical imaging
    • Very nice presentation by Christian Weiss
    • Cross-department funding for another workshop?

James V. Koch

  • Risk: You know the probabilities
  • Uncertainty: You don’t
  • Privately profitable vs. society
  • Recent inovations started from individuals, not big companies
  • National Labs work better when in cooperation with universities
  • Lesson learnt: Get people together and let them work on what they think
  • Profit motive makes people more productive
  • When to sell intellectual property, since imitations will arise soon
  • Lessons
    • Outside funders are skeptical
      • they look at the diffusion curve (uncertainty)
      • they want equity
    • Need substantial amount of equity to participate
    • Very high discount rates because of uncertainty
    • Investors require control
    • Try to get some benefits for the decision makers early on since they will not be there for too long
      • Don't sell out before benefits

James Cheng

  • VA institutions very spread around
  • Companies also
  • Angel investments recently appeared in south VA
    • 757 Angels - Hampton Roads Investing
    • CVA, 1717, Startup Virginia, NRV - Richmond
    • CAN, CAV Angels - Charlottesville
  • Resources on the slides
    • TEConomy Study
    • SBIG - small business investments
    • VEDP
    • VRIF Awards
    • OTA - way Gov't gets funding
  • Want scalable technology
    • tie software to a machine/product for intellectual property protection
Retrieved from "https://crtc.cs.odu.edu/index.php?title=Internal:CNF_Workshop&oldid=4164"