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
  26         # Initiliaze
  27         wmin = hmin = None # None is important here, confert the 'minimum' algo
  28         wmax = hmax = w = h = 0
  29
  30         for cas in l[START:END]:# cas like 'Case0002'
  31                 caspath = join(GLOB_DIR, cas)
  32
  33                 if not INDICE_CASE in cas:# if the directory name doesn't sound familiar
  34                         continue
  35
  36                 l1 = sorted( ls( caspath ) )# just ls one CaseXXXX and sort the result
  37                 # $$NOTRUSTREF here we need to sort because there is no trusted ref or id
  38
  39                 r = search(RT_PATH, cas)
  40
  41                 print(cas, end=' ', flush=True) # log CaseXXXX
  42
  43                 if not r: # if the feature doesn't yet labeled by an expert go to next
  44                         continue
  45                 r = search(r['path'], '.') # to be more dynamic, here can just be '.' instead of '1.2.3.4.5.6'
  46
  47                 if r:
  48                         r = r['path']
  49                         try:
  50                                 l2 = sorted( ls(r) ) # $$NOTRUSTREF
  51                         except NotADirectoryError as e:
  52                                 print("WARN", "a file in unexcepted place")
  53                                 continue
  54
  55                         if 0 < len(l2) < len(l1):
  56                                 r = join(r, l2[0])
  57                                 l2 = sorted( ls(r) ) # $$NOTRUSTREF
  58                                 # Try once : check subdirectory if i'am not the right
  59
  60                         if 0 < len(l2) < len(l1):
  61                                 # TODO: log, json doesn't match with images (labels != features)
  62                                 print("WARN", "json doesn't match with images (labels != features), Case", cas)
  63                                 continue
  64                                 pass
  65
  66                         for i, dic in enumerate(l1):
  67                                 # print('join', r, l2[i])
  68                                 ref = join(r, l2[i]) # logically, should be the json ref of i dicom image
  69
  70                                 infarctus = search(ref, INFA_STR)
  71                                 if infarctus:
  72                                         # print("infarctus:", infarctus) # Testing..
  73                                         # topng(join(caspath, dic), '%/%-%' % (join(OUT_DIR, 'infarctus'), cas, dic)
  74                                         # print(join(caspath, dic), '{}/{}-{}'.format(join(OUT_DIR, 'infarctus'), cas, dic)) # Testing..
  75                                         w, h = topng(join(caspath, dic), '{}/{}-{}'.format(join(OUT_DIR, 'infarctus'), cas, dic))
  76                                 else:
  77                                         # print("no infarctus:", infarctus) # Testing..
  78                                         # print(join(caspath, dic), '{}/{}-{}'.format(join(OUT_DIR, 'noinfarctus'), cas, dic)) # Testing..
  79                                         w, h = topng(join(caspath, dic), '{}/{}-{}'.format(join(OUT_DIR, 'noinfarctus'), cas, dic))
  80
  81                                 # search maximums
  82                                 if wmax < w: wmax = w
  83                                 if hmax < h: hmax = h
  84
  85                                 # search width minimum
  86                                 if wmin is None: wmin = w
  87                                 elif wmin > w: wmin = w
  88
  89                                 # search height minimum
  90                                 if hmin is None: hmin = h
  91                                 elif hmin > h: hmin = h
  92
  93         print('min-width, max-width:', (wmin, wmax))
  94         print('min-height, max-height:', (hmin, hmax))
  95
  96         print('Ended!')