ELogs/JeremyOcampo: Difference between revisions

From PBTWiki
Jump to navigation Jump to search
No edit summary
 
(15 intermediate revisions by the same user not shown)
Line 3: Line 3:
==To do==  
==To do==  


* <b>G4</b> Simulate an image from a 1mm thick scintillator in a non-scintillating plastic slab
* <b>New water tank setup</b>
** compare to image from thick scintillator block
** Scintillator sheet sandwiched between two water tanks replacing plastic blocks.
** this is to test the idea that we are seeing errors due to focal depth or light coming from different depths
** Optical grease in-between for smoothness (will scintillator be contaminated by the grease?)
* <b>G4</b> Get profiles at different depths and compare to data
** Polish outer surfaces and lens
* <b>G4</b> Get output factor curves and compare to data
 
* <b>G4</b> Make code neater, add more macros
* <b>Figure out physics of green dis-coloration in high dose areas of plastic</b>
* <b>Report</b> Finish first draft of start of report.
 
* <b>Camera</b> Familiarize oneself with DSLR camera and software
* Fill out [https://www.hep.ucl.ac.uk/pbt/wiki/Radiotherapy/Experimental_Runs wiki page] for radiotherapy experimental runs
 
* <b>Get ML models to work on real images</b>
 
* <b>Ask Stan for phd viva forms</b>


==Completed==
==Completed==
* <b>Workshop</b> Went to workshop - "Challenges and opportunities within practical small field dosimetry" - 9th may
 
* <b>G4</b> Simulate image from a scintillator and compared it's pdd with image from dslr camera.
=== 2nd year ===
** ~100,000 optical photons collected by lens given ~10,000,000 x-ray photons, which took 3 hours.
 
** reduced computing time using multithreaded mode, and also using the symmetry of the system and adding more lenses to collect more light
<br />
* <b>G4</b> Import phase space file (cyberknife) beam to use as primary generator.
* <big>1st term</big>
* <b>Group placement</b> exploring the simulacrum cancer data from PHE - 12th april
** <b>G4</b> Separate Scintillation and Cerenkov photons in scintillator sheet simulation
* <b>2nd term courses</b> Machine learning with Big data, Inverse problems in imaging
*** Whole beam produces Cerenkov photons in the plastic slabs+scintillator sheet while Scintillation is only produced in the scintillator sheet.
* <b>G4</b> Learning Geant4 with:  
*** A "scintillation only" image produces PDD, profile and Output factor within 2% of actual dose.
** [https://groups.lal.in2p3.fr/ED-geant4/category/2018/page/2/ ED-Geant4] - Lectures with exercises and solutions from an intro G4 workshop in 2018.
** <b>Birmingham Meeting 9th Dec</b> With Adam, Simon, Geoff, Hamid, Spyros
** [http://geant4-userdoc.web.cern.ch/geant4-userdoc/UsersGuides/ForApplicationDeveloper/html/ Geant4 book] - Reference for G4.
*** Presented simulation results
*** Methods to separate Cerenkov and scintillation, concluded with: <br />image plastic block (cerenkov) and then image scintillator sheet sandwiched by plastic slabs (scintillation + cerenkov) and then subtract
*** Plans for experiment in January with scintillator sheet sandwiched between plastic
*** Talked about using the system for treatment plan verification by comparing a video to simulation.
** <b>PhD transfer Report</b>
** <b>PhD transfer Viva</b> - passed yay
** <b>CDT presentation</b> of 1st year work
<br />
 
=== 1st year ===
<br />
* <big>Summer</big>
** <b>Summer School</b> STFC Data Intensive, Artificial Intelligence and Machine Learning Summer School 2019
** <b>G4</b> Simulate PDD, profile and Output factor from a scintillator sheet sandwiched between 2 platic slabs:
*** PDD agrees within 2% but profile and Output factor(OF) doesn't.
*** Cerenkov coming from plastic slabs distorts OF and profile.
*** This is because as you increase the beam width, the amount of X-ray going into the plastic slab is proportional to (beam width)<sup>2</sup> <br /> but the amount of X-rays going into the sheet is proportional to beam width.
 
<br />
* <big>3rd term</big>
** <b>Workshop</b> Went to workshop - "Challenges and opportunities within practical small field dosimetry" - 9th may
** <b>G4</b> Simulate image of an irradiated scintillator block and compared it's pdd with image from CCD camera.
*** ~100,000 optical photons collected by lens given ~10,000,000 x-ray photons, which took 3 hours.
** <b>G4</b> Speed improvements of scintillator block image:
*** can use 4 lens' as there is 2 planes of symmetry => simulation captures 4x as much photons
*** Use HEP batch farm to run jobs on 100 cores
*** kill optical photons which would be internally reflected by all sides of block at there 1st step => simulation ~10x faster
 
<br />
* <big>2nd term</big>
** <b>Courses</b>: Machine learning with big data, Inverse problems in imaging
** <b>Group placement project</b>: exploring the simulacrum cancer dataset from PHE (partnered with Valtech)
*** presentation
*** report
** <b>G4</b> Import phase space file (cyberknife) beam to use as primary generator.
** <b>G4</b> Simulate PDD, profile dose distributions in water and compare with measured.
 
 
<br />
* <big>1st term</big>
** <b>Courses</b>: Standard model 1 & 2, Prob and stats, Research software (python)
** <b>G4</b> Learning Geant4 with:  
*** [https://groups.lal.in2p3.fr/ED-geant4/category/2018/page/2/ ED-Geant4] - Lectures with exercises and solutions from an intro G4 workshop in 2018.
*** [http://geant4-userdoc.web.cern.ch/geant4-userdoc/UsersGuides/ForApplicationDeveloper/html/ Geant4 book] - Reference for G4.

Latest revision as of 09:15, 10 August 2022

Electronic Log for Jeremy Ocampo.

To do

  • New water tank setup
    • Scintillator sheet sandwiched between two water tanks replacing plastic blocks.
    • Optical grease in-between for smoothness (will scintillator be contaminated by the grease?)
    • Polish outer surfaces and lens
  • Figure out physics of green dis-coloration in high dose areas of plastic
  • Fill out wiki page for radiotherapy experimental runs
  • Get ML models to work on real images
  • Ask Stan for phd viva forms

Completed

2nd year


  • 1st term
    • G4 Separate Scintillation and Cerenkov photons in scintillator sheet simulation
      • Whole beam produces Cerenkov photons in the plastic slabs+scintillator sheet while Scintillation is only produced in the scintillator sheet.
      • A "scintillation only" image produces PDD, profile and Output factor within 2% of actual dose.
    • Birmingham Meeting 9th Dec With Adam, Simon, Geoff, Hamid, Spyros
      • Presented simulation results
      • Methods to separate Cerenkov and scintillation, concluded with:
        image plastic block (cerenkov) and then image scintillator sheet sandwiched by plastic slabs (scintillation + cerenkov) and then subtract
      • Plans for experiment in January with scintillator sheet sandwiched between plastic
      • Talked about using the system for treatment plan verification by comparing a video to simulation.
    • PhD transfer Report
    • PhD transfer Viva - passed yay
    • CDT presentation of 1st year work


1st year


  • Summer
    • Summer School STFC Data Intensive, Artificial Intelligence and Machine Learning Summer School 2019
    • G4 Simulate PDD, profile and Output factor from a scintillator sheet sandwiched between 2 platic slabs:
      • PDD agrees within 2% but profile and Output factor(OF) doesn't.
      • Cerenkov coming from plastic slabs distorts OF and profile.
      • This is because as you increase the beam width, the amount of X-ray going into the plastic slab is proportional to (beam width)2
        but the amount of X-rays going into the sheet is proportional to beam width.


  • 3rd term
    • Workshop Went to workshop - "Challenges and opportunities within practical small field dosimetry" - 9th may
    • G4 Simulate image of an irradiated scintillator block and compared it's pdd with image from CCD camera.
      • ~100,000 optical photons collected by lens given ~10,000,000 x-ray photons, which took 3 hours.
    • G4 Speed improvements of scintillator block image:
      • can use 4 lens' as there is 2 planes of symmetry => simulation captures 4x as much photons
      • Use HEP batch farm to run jobs on 100 cores
      • kill optical photons which would be internally reflected by all sides of block at there 1st step => simulation ~10x faster


  • 2nd term
    • Courses: Machine learning with big data, Inverse problems in imaging
    • Group placement project: exploring the simulacrum cancer dataset from PHE (partnered with Valtech)
      • presentation
      • report
    • G4 Import phase space file (cyberknife) beam to use as primary generator.
    • G4 Simulate PDD, profile dose distributions in water and compare with measured.



  • 1st term
    • Courses: Standard model 1 & 2, Prob and stats, Research software (python)
    • G4 Learning Geant4 with:
      • ED-Geant4 - Lectures with exercises and solutions from an intro G4 workshop in 2018.
      • Geant4 book - Reference for G4.