Convert to Python 3

segmentation-of-nuclei/color_norm/color_normalize.py

@@ -1,4 +1,4 @@
-import stainNorm_Reinhard
+from . import stainNorm_Reinhard
 from scipy import misc
 
 class reinhard_normalizer():

segmentation-of-nuclei/color_norm/stainNorm_Reinhard.py

@@ -1,8 +1,8 @@
-from __future__ import division
+
 
 import cv2 as cv
 import numpy as np
-import stain_utils as ut
+from . import stain_utils as ut
 
 
 ### Some functions ###

segmentation-of-nuclei/color_norm/stain_utils.py

@@ -6,7 +6,7 @@
 Use with python via e.g https://anaconda.org/conda-forge/python-spams
 """
 
-from __future__ import division
+
 
 import numpy as np
 

segmentation-of-nuclei/convert_polygon_to_mask.py

@@ -74,7 +74,7 @@ def process_poly(poly_path):
         path_mkdir(path.join(cancer_type + '_mask', slide))
         mask.save(output_path)
     except:
-        print 'exception for', poly_path
+        print('exception for', poly_path)
 
 poly_paths = glob('*_polygon/*/*-features.csv')
 

segmentation-of-nuclei/data/nuclei_data.py

@@ -8,7 +8,7 @@
 
 import numpy as np
 
-from utils import imread, imwrite
+from .utils import imread, imwrite
 
 DATA_FNAME = 'real_pair.npz'
 
@@ -56,7 +56,7 @@ def maybe_preprocess(config, data_path, sample_path=None):
 
 def load(config, data_path, sample_path, rng):
   if not os.path.exists(data_path):
-    print('creating folder', data_path)
+    print(('creating folder', data_path))
     os.makedirs(data_path)
 
   synthetic_image_path = maybe_preprocess(config, data_path, sample_path)
@@ -69,7 +69,7 @@ def load(config, data_path, sample_path, rng):
   if not os.path.exists(sample_path):
     os.makedirs(sample_path)
 
-  print("[*] Save samples images in {}".format(data_path))
+  print(("[*] Save samples images in {}".format(data_path)))
   random_idxs = rng.choice(len(real_data), 100)
   for idx, random_idx in enumerate(random_idxs):
     image_path = os.path.join(sample_path, "real_{}.png".format(idx))

segmentation-of-nuclei/data/utils.py

@@ -91,8 +91,8 @@ def img_tile(imgs, aspect_ratio=1.0, tile_shape=None, border=1,
 def save_config(model_dir, config):
   param_path = os.path.join(model_dir, "params.json")
 
-  print("[*] MODEL dir: %s" % model_dir)
-  print("[*] PARAM path: %s" % param_path)
+  print(("[*] MODEL dir: %s" % model_dir))
+  print(("[*] PARAM path: %s" % param_path))
 
   with open(param_path, 'w') as fp:
     json.dump(config.__dict__, fp,  indent=4, sort_keys=True)

segmentation-of-nuclei/extract_patch_segmentation_mask.py

@@ -59,17 +59,17 @@ def patch_xy2mask(patch_xy, scale_to_40X=True):
 
 def extract_segmentation_mask(
         segmentation_polygon_folder, x, y, patch_width, scale_to_40X=True):
-'''
-  Extract segmentation mask
+    '''
+    Extract segmentation mask
 
-  Args:
-    segmentation_polygon_folder: path to a segmentation output folder.
-    x: x coordinate of the patch you want to extract.
-    y: y coordinate of the patch.
-    patch_width: size of the patch.
-  Returns:
-    Instance-level mask as numpy array.
-'''
+    Args:
+        segmentation_polygon_folder: path to a segmentation output folder.
+        x: x coordinate of the patch you want to extract.
+        y: y coordinate of the patch.
+        patch_width: size of the patch.
+    Returns:
+        Instance-level mask as numpy array.
+    '''
 
     patch_xy = info2patch_xy(
             segmentation_polygon_folder, x, y, patch_width)

segmentation-of-nuclei/fix_old_data/one_pass_fix.py

@@ -70,7 +70,7 @@ def fix_probability_map_png(in_folder):
         im.save(fn)
 
 if len(sys.argv) != 2:
-    print('Usage: python {} segmentation_folder'.format(sys.argv[0]))
+    print(('Usage: python {} segmentation_folder'.format(sys.argv[0])))
     exit(1)
 in_folder = sys.argv[1]
 fix_polygon_csv(in_folder)

segmentation-of-nuclei/layers.py

@@ -85,10 +85,10 @@ def resnet_block(
        inputs.get_shape()[3] == layer.get_shape()[3]:
       outputs = tf.add(inputs, layer)
     else:
-      print "Cannot tf.add: shapes do not match {}-{} {}-{} {}-{}".format(
+      print("Cannot tf.add: shapes do not match {}-{} {}-{} {}-{}".format(
               inputs.get_shape()[1], layer.get_shape()[1],
               inputs.get_shape()[2], layer.get_shape()[2],
-              inputs.get_shape()[3], layer.get_shape()[3]);
+              inputs.get_shape()[3], layer.get_shape()[3]));
       outputs = layer;
 
   _update_dict(layer_dict, scope, outputs)

segmentation-of-nuclei/model.py

@@ -394,7 +394,7 @@ def run(sess, feed_dict, fetch,
         fetch['output'] = output_op
 
       result = sess.run(fetch, feed_dict=feed_dict)
-      if result.has_key('summary'):
+      if 'summary' in result:
         summary_writer.add_summary(result['summary'], result['step'])
         summary_writer.flush()
       return result

segmentation-of-nuclei/postprocess.py

@@ -40,7 +40,7 @@ def touch(fname, times=None):
             time.sleep(1)
         else:
             touch(os.path.join(folder, 'finish'))
-            print 'Segmentation finished under {}'.format(folder)
+            print('Segmentation finished under {}'.format(folder))
 
 
 class MultiProcWatershed:
@@ -51,7 +51,7 @@ def __init__(self, n_proc):
 
     def start(self):
         self.workers = [Process(target=queue_waiter, args=(i, self.queue,))
-                        for i in xrange(self.n_proc)]
+                        for i in range(self.n_proc)]
         for w in self.workers:
             w.start()
 

segmentation-of-nuclei/preprocess.py

@@ -41,7 +41,7 @@ def stain_normalized_tiling(slide_name, patch_size, do_actually_read_image=True)
         width = oslide.dimensions[0]
         height = oslide.dimensions[1]
     except:
-        print 'Error in {}: exception caught exiting'.format(slide_name)
+        print('Error in {}: exception caught exiting'.format(slide_name))
         raise Exception('{}: exception caught exiting'.format(slide_name))
         return
 
@@ -65,7 +65,7 @@ def stain_normalized_tiling(slide_name, patch_size, do_actually_read_image=True)
                 try:
                     patch = oslide.read_region((x, y), 0, (pw_x, pw_y)).convert('RGB')
                 except:
-                    print '{}: exception caught'.format(slide_name)
+                    print('{}: exception caught'.format(slide_name))
                     continue
             else:
                 patch = Image.new('RGB', (pw_x, pw_y), (255, 255, 255))

segmentation-of-nuclei/trainer.py

@@ -21,6 +21,7 @@
 from utils import imwrite, imread, img_tile, synthetic_to_refer_paths
 from preprocess import stain_normalized_tiling
 from postprocess import MultiProcWatershed
+from functools import reduce
 
 class Trainer(object):
   def __init__(self, config, rng):
@@ -169,7 +170,7 @@ def train_refiner(push_buffer=False):
 
     def train_discrim():
       a, b, c = self.history_buffer.sample()
-      d, e = self.data_loader.next()
+      d, e = next(self.data_loader)
 
       feed_dict = {
         self.model.synthetic_batch_size: self.data_loader.batch_size/2,
@@ -203,21 +204,21 @@ def train_learner():
       train_discrim()
 
     for step in trange(self.max_step_d_g, desc="Train refiner+discrim"):
-      for k in xrange(self.K_g):
+      for k in range(self.K_g):
         train_refiner(push_buffer=True)
 
-      for k in xrange(self.K_d):
+      for k in range(self.K_d):
         train_discrim()
 
     for k in trange(self.initial_K_l, desc="Train learner"):
       train_learner()
 
     for step in trange(self.max_step_d_g_l, desc="Train all Three"):
-      for k in xrange(self.K_g):
+      for k in range(self.K_g):
         train_refiner(push_buffer=True)
-      for k in xrange(self.K_l):
+      for k in range(self.K_l):
         train_learner()
-      for k in xrange(self.K_d):
+      for k in range(self.K_d):
         train_discrim()
 
     for k in trange(self.after_K_l, desc="Train learner"):
@@ -238,8 +239,8 @@ def get_image_id(svs_path, delimitor='.'):
       try:
         self.cnn_pred_mask(wsi, image_id)
       except Exception as e:
-        print traceback.format_exc()
-        print 'Segmentation failed for {}'.format(wsi)
+        print(traceback.format_exc())
+        print('Segmentation failed for {}'.format(wsi))
         continue
 
     if self.do_cpu_postprocess:
@@ -270,15 +271,15 @@ def cnn_pred_mask(self, wsi_path, image_id):
 
       # Check if skip the CNN step
       if self.do_gpu_process:
-        print "CNN segmentation on", outf
+        print("CNN segmentation on", outf)
 
         pred_m = np.zeros((patch.shape[0], patch.shape[1], 3), dtype=np.float32);
         num_m = np.zeros((patch.shape[0], patch.shape[1], 1), dtype=np.float32) + 4e-6;
 
         net_inputs = [];
         xy_indices = [];
-        for x in range(0, pred_m.shape[0]-PS+1, step_size) + [pred_m.shape[0]-PS,]:
-          for y in range(0, pred_m.shape[1]-PS+1, step_size) + [pred_m.shape[1]-PS,]:
+        for x in list(range(0, pred_m.shape[0]-PS+1, step_size)) + [pred_m.shape[0]-PS,]:
+          for y in list(range(0, pred_m.shape[1]-PS+1, step_size)) + [pred_m.shape[1]-PS,]:
             pat = patch[x:x+PS, y:y+PS, :]
             wh = pat[...,0].std() + pat[...,1].std() + pat[...,2].std();
             if wh >= 0.18:

segmentation-of-nuclei/utils.py

@@ -91,8 +91,8 @@ def img_tile(imgs, aspect_ratio=1.0, tile_shape=None, border=1,
 def save_config(model_dir, config):
   param_path = os.path.join(model_dir, "params.json")
 
-  print("[*] MODEL dir: %s" % model_dir)
-  print("[*] PARAM path: %s" % param_path)
+  print(("[*] MODEL dir: %s" % model_dir))
+  print(("[*] PARAM path: %s" % param_path))
 
   with open(param_path, 'w') as fp:
     json.dump(config.__dict__, fp,  indent=4, sort_keys=True)

segmentation-of-nuclei/watershed/seg_with_watershed.py

@@ -8,8 +8,8 @@
 from skimage.color import label2rgb
 import time
 
-import detection_binarize
-from gen_json import gen_meta_json
+from . import detection_binarize
+from .gen_json import gen_meta_json
 
 def apply_segmentation(in_path, image_id, wsi_width, wsi_height, method_description,
         seg_thres=0.33, det_thres=0.07, win_size=200, min_nucleus_size=20, max_nucleus_size=65536):
@@ -70,7 +70,7 @@ def read_instance(im_file, resize_factor):
 
         return det_seg[..., 0], det_seg[..., 1]
 
-    print "Watershed postprocessing on", in_path
+    print("Watershed postprocessing on", in_path)
 
     file_id = os.path.basename(in_path)[:-len('_SEG.png')]
     resize_factor = int(file_id.split('_')[5])
@@ -122,7 +122,7 @@ def read_instance(im_file, resize_factor):
 
     time3 = time.time()
 
-    print "Time in watershed for ",in_path," Padding/Seed: ",(time1-time0)," Watershed: ",(time2-time1)," Fillholes: ",(time3-time2)
+    print("Time in watershed for ",in_path," Padding/Seed: ",(time1-time0)," Watershed: ",(time2-time1)," Fillholes: ",(time3-time2))
 
     gen_meta_json(in_path, image_id, wsi_width, wsi_height, method_description,
             seg_thres, det_thres, win_size, min_nucleus_size, max_nucleus_size)

training-data-real-patch-extraction/extract_patches.py

@@ -32,7 +32,7 @@ def poly2mask(vertex_row_coords, vertex_col_coords, mask):
 
 
 for id in ids:
-    print 'converting xml to png. {}'.format(id)
+    print('converting xml to png. {}'.format(id))
 
     tissue = Image.open('MoNuSeg Training Data/Tissue images/{}.tif'.format(id))
     s0, s1 = tissue.size[0], tissue.size[1]
@@ -56,18 +56,18 @@ def poly2mask(vertex_row_coords, vertex_col_coords, mask):
 
 
 for id in ids:
-    print 'extracting patches from png. {}'.format(id)
+    print('extracting patches from png. {}'.format(id))
 
     he_image = np.array(Image.open('MoNuSeg Training Data/Tissue images/{}.tif'.format(id)).convert('RGB'))[15:-15, 15:-15, :]
     seg_gt = misc.imread('MoNuSeg Training Data/Mask/{}.png'.format(id))[15:-15, 15:-15] / 40
 
-    for x in range(0, he_image.shape[0]-win_size, stride) + [he_image.shape[0]-win_size,]:
-        for y in range(0, he_image.shape[1]-win_size, stride) + [he_image.shape[1]-win_size,]:
+    for x in list(range(0, he_image.shape[0]-win_size, stride)) + [he_image.shape[0]-win_size,]:
+        for y in list(range(0, he_image.shape[1]-win_size, stride)) + [he_image.shape[1]-win_size,]:
             he = he_image[x:x+win_size, y:y+win_size, :]
             seg = seg_gt[x:x+win_size, y:y+win_size]
             if he.shape[0] != win_size or he.shape[1] != win_size or \
                 seg.shape[0] != win_size or seg.shape[1] != win_size:
-                print 'some shape error', he.shape, seg.shape
+                print('some shape error', he.shape, seg.shape)
                 continue
             misc.imsave('real_data_400x400/image_sup/{}.png'.format(patch_id), he)
             misc.imsave('real_data_400x400/mask_sup/{}.png'.format(patch_id), seg)

training-data-synthesis/draw_fake.py

@@ -12,7 +12,7 @@
     try:
         image, mask, detect, contour, refer, source, nucl, cyto, intp_mask, tile_path = get_new_fake_image();
     except Exception as e:
-        print e;
+        print(e);
         continue;
     mask = (mask>0).astype(np.uint8);
     detect = (detect>0).astype(np.uint8);

training-data-synthesis/nuclei_synthesis_40X_online/image_synthesize.py

@@ -1,9 +1,9 @@
 import numpy as np
 import pickle
-import png
+from . import png
 from PIL import Image, ImageDraw
-from polygon import generatePolygon
-from mask2image_otsu import Mask2Image
+from .polygon import generatePolygon
+from .mask2image_otsu import Mask2Image
 from scipy import ndimage
 from sys import stdout
 

training-data-synthesis/nuclei_synthesis_40X_online/mask2image_otsu.py

@@ -6,7 +6,7 @@
 from os import listdir
 from sys import stdout
 from os.path import isfile, join
-from my_canny import canny_edge_on_mask
+from .my_canny import canny_edge_on_mask
 from scipy import ndimage
 from scipy import stats
 

training-data-synthesis/nuclei_synthesis_40X_online/png.py

@@ -1,6 +1,6 @@
 #!/usr/bin/env python
 
-from __future__ import print_function
+
 
 # png.py - PNG encoder/decoder in pure Python
 #
@@ -1608,7 +1608,7 @@ def asvalues(raw):
                 out.extend([mask&(o>>i) for i in shifts])
             return out[:width]
 
-        return map(asvalues, rows)
+        return list(map(asvalues, rows))
 
     def serialtoflat(self, bytes, width=None):
         """Convert serial format (byte stream) pixel data to flat row
@@ -1905,8 +1905,8 @@ def iterdecomp(idat):
             arraycode = 'BH'[self.bitdepth>8]
             # Like :meth:`group` but producing an array.array object for
             # each row.
-            pixels = map(lambda *row: array(arraycode, row),
-                       *[iter(self.deinterlace(raw))]*self.width*self.planes)
+            pixels = list(map(lambda *row: array(arraycode, row),
+                       *[iter(self.deinterlace(raw))]*self.width*self.planes))
         else:
             pixels = self.iterboxed(self.iterstraight(raw))
         meta = dict()
@@ -2042,11 +2042,11 @@ def itertrns(pixels):
                     # True/False to 0/maxval (by multiplication),
                     # and add it as the extra channel.
                     row = group(row, planes)
-                    opa = map(it.__ne__, row)
-                    opa = map(maxval.__mul__, opa)
+                    opa = list(map(it.__ne__, row))
+                    opa = list(map(maxval.__mul__, opa))
                     opa = list(zip(opa)) # convert to 1-tuples
                     yield array(typecode,
-                      itertools.chain(*map(operator.add, row, opa)))
+                      itertools.chain(*list(map(operator.add, row, opa))))
             pixels = itertrns(pixels)
         targetbitdepth = None
         if self.sbit: