totraindir.py

   1 from os import listdir as ls
   2 import png
   3 import pydicom
   4 from os.path import join
   5
   6 # locals
   7 from topng import topng
   8 from regularjson import search, RT_PATH, JSON_GTS, INFA_STR
   9
  10 # constants
  11 GLOB_DIR = '../../Data/Images_anonymous/'
  12 OUT_DIR = './train/'
  13 INDICE_CASE = 'Case'
  14
  15 START = None # to manage the number of Patient Case to use in training + validation, if START == None => 0, if END == None => last index, it will use all in GLOB_DIR
  16 END = None
  17
  18
  19 def get(l, i, r):
  20         if len(l) <= 1:
  21                 return sorted( ls(r) )
  22
  23 if __name__ == '__main__':
  24         l = sorted(ls(GLOB_DIR))
  25         for cas in l[START:END]:# cas like 'Case0002'
  26                 caspath = join(GLOB_DIR, cas)
  27
  28                 if not INDICE_CASE in cas:# if the directory name doesn't sound familiar
  29                         continue
  30
  31                 l1 = sorted( ls( caspath ) )# just ls one CaseXXXX and sort the result
  32                 # $$NOTRUSTREF here we need to sort because there is no trusted ref or id
  33
  34                 r = search(RT_PATH, cas)
  35
  36                 print(cas, end=' ', flush=True) # log CaseXXXX
  37
  38                 if not r: # if the feature doesn't yet labeled by an expert go to next
  39                         continue
  40                 r = search(r['path'], '.') # to be more dynamic, here can just be '.' instead of '1.2.3.4.5.6'
  41
  42                 if r:
  43                         r = r['path']
  44                         try:
  45                                 l2 = sorted( ls(r) ) # $$NOTRUSTREF
  46                         except NotADirectoryError as e:
  47                                 print("WARN", "a file in unexcepted place")
  48                                 continue
  49
  50                         if 0 < len(l2) < len(l1):
  51                                 r = join(r, l2[0])
  52                                 l2 = sorted( ls(r) ) # $$NOTRUSTREF
  53                                 # Try once : check subdirectory if i'am not the right
  54
  55                         if 0 < len(l2) < len(l1):
  56                                 # TODO: log, json doesn't match with images (labels != features)
  57                                 print("WARN", "json doesn't match with images (labels != features), Case", cas)
  58                                 continue
  59                                 pass
  60
  61
  62                         for i, dic in enumerate(l1):
  63                                 # print('join', r, l2[i])
  64                                 ref = join(r, l2[i]) # logically, should be the json ref of i dicom image
  65
  66                                 infarctus = search(ref, INFA_STR)
  67                                 if infarctus:
  68                                         # print("infarctus:", infarctus) # Testing..
  69                                         # topng(join(caspath, dic), '%/%-%' % (join(OUT_DIR, 'infarctus'), cas, dic)
  70                                         # print(join(caspath, dic), '{}/{}-{}'.format(join(OUT_DIR, 'infarctus'), cas, dic)) # Testing..
  71                                         topng(join(caspath, dic), '{}/{}-{}'.format(join(OUT_DIR, 'infarctus'), cas, dic))
  72                                 else:
  73                                         # print("no infarctus:", infarctus) # Testing..
  74                                         # print(join(caspath, dic), '{}/{}-{}'.format(join(OUT_DIR, 'noinfarctus'), cas, dic)) # Testing..
  75                                         topng(join(caspath, dic), '{}/{}-{}'.format(join(OUT_DIR, 'noinfarctus'), cas, dic))
  76
  77         print('Ended!')