Proton Calorimetry/Meetings/2024/11/20

Minutes for UCL Proton Beam Therapy Group Meetings, 20th November

Joe Bateman, Sonia Escribano, Febian, Simon Jolly

Currently occupied with numerous grant applications including Health and tech grant application
Scintillator machining by company in Wales - kept in email loop by Wayne
Crick institute also has a machine we could possibly use
Martin has got sheets to 30um flatness but need to tests on optical clearness
Multiple investigations into sheet manufacturing currently: Harry, Wayne, cricke inst, Martin
Meeting this morning about circuit design with Pete Hastings in Oxford, ddc boards sent out for manufacturing and should be made within 10 days

GUI v3.3
- GUI DAQ: acquires data with FTDI_GUI C++ code
  - Calibration applied to live data but some times live data is shown quicker than calibration function can be applied
  - Occasionally postprocessing calibration and live calibration has 1.1 factor difference
  - Current state of GUI page is now ready
- GUI Calibrated - calibrates raw file and desired frequency: 25 Hz -1000 Hz or raw
- GUI Replay - replays fitted data obtained during DAQ
  - Slight issue on automatically changing y axis value but shall be fixed soon
GUI at final version for now
Put under version control and document all of them
Next main task is when the new DDC boards arrive so will need Matt to explain about what pins do what on new zmod boards
Start inventory for scintillator sheets and start new naming convention e.g. YYMM<Sheet No.> (maybe thickness also).

Looking Saad's QuARC Geant4 simulation up and running over CVMFS and will also look at adding the fibre tracker module to have a full QuADProBe simulation.
Modified ProtonGB tutorial script to score 3D voxelised dose distribution in water phantom.
- General purpose Geant4 is fine for 1D and 2D dose distributions, however when running a 3D voxelised scorer, the available RAM limits the resolution, e.g., even over batch farm it is unlikely to be able to score more than 50x50x50 bins.
- Therefore propose to use TOPAS and FRED for 3D dose simulations rather than spend time on developing improved memory management in standalone Geant4 (especially since TOPAS uses the same physics as Geant4 and has optimised memory management specifically for this purpose and for PBT simulations).
Have simulated first pencil beam from run033 from UCLH in FRED (and is very quick!! less than 1 minute to run 1e6 protons into a water phantom scoring the dose in 100x100x200 bins with 1mm^3 voxel size).
Simulate the UCLH pristine Bragg peak, SOBP and 5x5 cm box field runs between FRED and TOPAS and compare dose distributions.
Take charge of PPRIG 2025 abstract due 31/12.

Gave second part of presentation on FPGAs
Looking at converting training ML models in tensor flow /keras to fpga via hls4ml produces c++ zip files then convert to vivado fpga using vitilis hls
Fpga fit using minimiser using Minuit library