def tif_to_png(input, output):
stack = tifread(input)
os.chdir(output)
for i, slice in enumerate(stack):
pictures = Image.fromarray(slice)
pictures.save('%02d.png' % i)
print(str(i + 1) + "/" + str(len(stack)) + " files created")
def membrane(input, output):
stack = tifread(input)
os.chdir(output)
for i, slice in enumerate(stack):
edges_compared = edges(slice)
pictures = Image.fromarray(edges_compared.astype('uint8') * 255)
pictures.save('%02d.png' % i)
print(str(i + 1) + "/" + str(len(stack)) + " files created")
def main():
'reading tif files'
true_label = tifread(a.true)
pred_label = tifread(a.predicted)
'unravel tif files'
true, pred = unravel(true_label, pred_label)
'calculation of ari, split and merge errors and all pa'
splits, merges, ari = matrix(true, pred)
recall, precision = recall_precision(true, pred)
'prints'
print("\nEvaluation results:\n")
print("Splits = %i" % splits)
print("Merges = %i" % merges)
print("Adjusted Rand Index = %0.5f" % ari)
print("Recall = %0.5f" % recall)
print("Precision = %0.5f\n" % precision)
def cytoplasm(input, output):
stack = tifread(input)
os.chdir(output)
for i, slice in enumerate(stack):
edges_compared = edges(slice)
cytoplasm_logical = np.logical_and(np.logical_not(edges_compared), slice > 0)
pictures = Image.fromarray(cytoplasm_logical.astype('uint8') * 255)
pictures.save('%02d.png' % i)
print(str(i + 1) + "/" + str(len(stack)) + " files created")
def split_snemi3d(input, output, prefix):
stack = tifread(input)
os.chdir(output)
'slicing original image stack'
stack_split_a = stack[:, 512:, 512:]
stack_split_b = stack[:, :512, 512:]
stack_split_c = stack[:, :512, :512]
stack_split_d = stack[:, 512:, :512]
'saving all four split image stacks'
tifsave('%s_a.tif' % prefix, stack_split_a)
tifsave('%s_b.tif' % prefix, stack_split_b)
tifsave('%s_c.tif' % prefix, stack_split_c)
tifsave('%s_d.tif' % prefix, stack_split_d)
def consecutive(input, output, size):
stack = tifread(input)
os.chdir(output)
if len(stack.shape) == 4:
stack = stack[:, :, :, 0]
stack1 = stack[:-1]
stack2 = stack[1:]
for i, (slice1, slice2) in enumerate(zip(stack1, stack2)):
zero_array = np.zeros((3, int(size), int(size)))
zero_array[0] = slice1
zero_array[1] = slice2
zero_array = np.ascontiguousarray(zero_array.transpose(1, 2, 0))
consecutive_image = zero_array.astype('uint8')
pictures = Image.fromarray(consecutive_image, 'RGB')
pictures.save('%02d.png' % i)
zero_array[0] = 0
zero_array[1] = 0
print(str(i + 1) + "/" + str(len(stack1)) + " files created")
def overlap(input, output):
stack = tifread(input)
os.chdir(output)
list_for_slices = []
for i, slice in enumerate(stack):
edges_compared = edges(slice)
list_for_slices.append(np.invert(edges_compared))
print(str(i + 1) + "/" + str(len(stack)) + " membrane pictures for overlap detection processed")
membranes = np.asarray(list_for_slices)
membranes_indices = membranes * stack
stack1 = membranes_indices[:-1]
stack2 = membranes_indices[1:]
compared_stack = np.logical_and(np.logical_and(stack1 == stack2, stack1 > 0), stack2 > 0)
for i, slice in enumerate(compared_stack):
overlap_image = slice.astype('uint8') * 255
pictures = Image.fromarray(overlap_image)
pictures.save('%02d.png' % i)
print(str(i + 1) + "/" + str(len(compared_stack)) + " files created")
"to browers running on the local machine, use 0.0.0.0 to permit access from remote browsers.")
parser.add_argument(
"--static-content-url", help="Obtain the Neuroglancer client code from the specified URL.")
parser.add_argument("--em", required=False, help="path to em tif file")
parser.add_argument("--label", required=False, help="path to original labeled tif file")
parser.add_argument("--predicted", required=False, help="path to predicted tif file")
a = parser.parse_args()
if a.bind_address:
neuroglancer.set_server_bind_address(a.bind_address)
if a.static_content_url:
neuroglancer.set_static_content_source(url=a.static_content_url)
em_images = tifread(a.em)
em_array = np.asarray(em_images, dtype="uint8")
label_images = tifread(a.label)
label_array = np.asarray(label_images, dtype="uint16")
predicted_images = tifread(a.predicted)
predicted_array = np.asarray(predicted_images, dtype="uint16")
viewer = neuroglancer.Viewer()
with viewer.txn() as s:
scaling = [5, 5, 30]
offset = [0, 0, 0]
s.layers["EM"] = neuroglancer.ImageLayer(
source=neuroglancer.LocalVolume(em_array, voxel_size=scaling, voxel_offset=offset),