Only support windows, GPU with cuda required. Python version should be 3.9

minimum example:

conda activate base
pip install -r requirements.txt
python run_mc.py

• it will call Water(110), calculate 110MeV no size proton beam in water phantom. results save in ./output/totalDose.img, you can read it in matlab

  fileID = fopen([path,'/output/totalDose.img'],'r');
  temp = fread(fileID,'float');
  dose = reshape(temp,51,51,360);

• in water example, it may take 1min at low energy, but it becomes fast when the energy is high, 7 sec at 180MeV
• Things you can modify in Water(E) function:

  Nx = 51
  Ny = 51
  Nz = 360
  dx = 0.1#cm
  dy = 0.1
  dz = 0.1
  energy = E # MeV
  spot_size = [0,0]# beam spot size (cm), sqare beam shape
  num_particle = 1e6# integer, num of particles
  geo.set_header(Nx,Ny,Nz,dx,dy,dz)# see comments
  geo.set_cfg(energy,spot_size,num_particle)# see comments

How to load CT images, modify modify_this.py:

• import lib

  import appgopmc_dose as gopmc
  import mc_setting as mc
  import numpy as np
  import os 
  from scipy.io import savemat
  import matplotlib.pyplot as plt
  import napari
  import time

• prepare MC engine

  pmc = gopmc.appgopmc_dose()
  def change_dir():
      cwd = os.getcwd()
      current_file_path = os.path.dirname(os.path.abspath(__file__))
      if cwd != current_file_path:
          os.chdir(current_file_path)
      return current_file_path
  current_file_path = change_dir()
  geo = mc.geometry(current_file_path)

• read dicom files with read_dicom.py, costumize your own path, filename and index. Then convert dicom files to a single 3d numpy array

  # your dicom names may not formatted, try to rename them in 3rd party software like 3D slicer
  name2 = 'CT.1.3.12.2.1107.5.1.7.130094.30000022051213533424400000'# 214->348
  path2 = 'C:/Users/shuangzhou/Downloads/Head_Phantom/head2/'
  idx2 = np.arange(214,348)
  head,dxyz = read_3d_head_phantom(path2,name2,idx2)
  # save it for later use or don't save
  np.savez('./Phantom/head.npz', head=head, dxyz=dxyz)

• write the 3D numpy array into 1D binary file

  #transpose(2, 1, 0) is required for matlab
  image = np.array(head,dtype=np.short).transpose(2, 1, 0).flatten()
  mc.write_image(image,current_file_path + '\\Phantom\\')

• create a geometry for the CT file (see minimum example)

  geo.set_header(Nx,Ny,Nz,dx,dy,dz)# see comments
  geo.set_cfg(energy,spot_size,num_particle)# see comments

• build your geometry

  geo.create_cfg_file('pencilbeam.cfg')
  geo.create_header_file()

• run

  start = time.time()
  dose = pmc.run(geo.cfg_file,image)
  print("Total time = ",time.time() - start)
  totalDose = np.array(dose).reshape((geo.Nz,geo.Nx,geo.Ny)).transpose(2, 1, 0)

• plot results in python

  viewer = napari.view_image(totalDose, rgb=False)
  napari.run()  # start the event loop and show viewer

• save results to matlab

  matData_filename = 'waterDose'+str(E)
  savemat(geo.outputdir+matData_filename+'.mat',{'totalDose':totalDose, 'dx':geo.dx, 'dy':geo.dy, 'dz':geo.dz})

• load (*.mat)data in matlab, view 3d data with VolumeViewer in Image processing and computer vision toolbox

  load([matData_filename,'.mat']);
  % or load data in struct
  dose = load([matData_filename,'.mat']);
  % then dose.totalDose, dose.dx ...

• read (*.img)data in matlab (see minimum example)

Don't want to run with python

• get geometry and img files(data type is short or int16) ready
• open cmd in current folder
• appgopmc_dose.exe --config ./Phantom/pencilbeam.cfg
• results saved in ./output/, totalDose.img and totalDose.header
• read *.img file in matlab, data type is float

matlab_func

• All files in it are not used for IMPT, they are used for MLSIC
• read .img files

  fileID = fopen([path,'/output/totalDose.img'],'r');
  temp = fread(fileID,'float');
  dose = reshape(temp,51,51,360);

• write .img files

  scene = zeros(51,51,360);
  fileID = fopen([path,'/Phantom/geo_phantom.img'],'w+');
  fwrite(fileID,scene(:),'short');
  fclose(fileID);

• show IDD

  plot(squeeze(sum(totalDose,[1,2])))

Important files

appgopmc_dose.cp39-win_amd64.pyd
# GPU Monte Carlo engine, python module
# import appgopmc_dose
# pmc = appgopmc_dose.appgopmc_dose()
# dose = pmc.run(cfg_file,image) # cfg_file: full path of pencilbeam.cfg file, image: flatten 3d numpy array (placeholder,not used)
appgopmc_dose.exe
# os.system('appgopmc_dose.exe --config ./Phantom/pencilbeam.cfg')
# results saved in, output/totalDose.img and output/totalDose.header

Phantom.cl, ParticleStatus.cl, Macro.h, randomKernel.h
# they will be read by Monte Carlo engine should not change, don't move to other folder