def main():
args = getArguments(getParser())
# prepare logger
logger = Logger.getInstance()
if args.debug: logger.setLevel(logging.DEBUG)
elif args.verbose: logger.setLevel(logging.INFO)
# loading input images
b0img, b0hdr = load(args.b0image)
bximg, bxhdr = load(args.bximage)
# convert to float
b0img = b0img.astype(numpy.float)
bximg = bximg.astype(numpy.float)
# check if image are compatible
if not b0img.shape == bximg.shape:
raise ArgumentError('The input images shapes differ i.e. {} != {}.'.format(b0img.shape, bximg.shape))
if not header.get_pixel_spacing(b0hdr) == header.get_pixel_spacing(bxhdr):
raise ArgumentError('The input images voxel spacing differs i.e. {} != {}.'.format(header.get_pixel_spacing(b0hdr), header.get_pixel_spacing(bxhdr)))
# check if supplied threshold value as well as the b value is above 0
if args.threshold is not None and not args.threshold >= 0:
raise ArgumentError('The supplied threshold value must be greater than 0, otherwise a division through 0 might occur.')
if not args.b > 0:
raise ArgumentError('The supplied b-value must be greater than 0.')
# compute threshold value if not supplied
if args.threshold is None:
b0thr = otsu(b0img, 32) / 4. # divide by 4 to decrease impact
bxthr = otsu(bximg, 32) / 4.
if 0 >= b0thr:
raise ArgumentError('The supplied b0image seems to contain negative values.')
if 0 >= bxthr:
raise ArgumentError('The supplied bximage seems to contain negative values.')
else:
b0thr = bxthr = args.threshold
logger.debug('thresholds={}/{}, b-value={}'.format(b0thr, bxthr, args.b))
# threshold b0 + bx DW image to obtain a mask
# b0 mask avoid division through 0, bx mask avoids a zero in the ln(x) computation
mask = binary_fill_holes(b0img > b0thr) & binary_fill_holes(bximg > bxthr)
# perform a number of binary morphology steps to select the brain only
mask = binary_erosion(mask, iterations=1)
mask = largest_connected_component(mask)
mask = binary_dilation(mask, iterations=1)
logger.debug('excluding {} of {} voxels from the computation and setting them to zero'.format(numpy.count_nonzero(mask), numpy.prod(mask.shape)))
# compute the ADC
adc = numpy.zeros(b0img.shape, b0img.dtype)
adc[mask] = -1. * args.b * numpy.log(bximg[mask] / b0img[mask])
adc[adc < 0] = 0
# saving the resulting image
save(adc, args.output, b0hdr, args.force)
def proc_roi_region(self,add_region=True):
mpossx = self.roi_item.getArrayRegion(self.possx,self.img).astype(int)
mpossx = mpossx[np.nonzero(binary_fill_holes(mpossx))]#get the x pos from ROI
mpossy = self.roi_item.getArrayRegion(self.possy,self.img).astype(int)
mpossy = mpossy[np.nonzero(binary_fill_holes(mpossy))]# get the y pos from ROI
xLims = [np.min(mpossx)-10,np.max(mpossx)+10]
yLims = [np.min(mpossy)-10,np.max(mpossy)+10]
#xLims = [np.mean(mpossx)-20,np.mean(mpossx)+20]; yLims = [np.mean(mpossy)-20,np.mean(mpossy)+20]
xIn = np.all(np.logical_and(xLims[0]>1,xLims[1]<510))
yIn = np.all(np.logical_and(yLims[0]>1,yLims[1]<510))
if np.logical_and(xIn,yIn):
self.temp_mask[mpossx,mpossy] = 1
self.temp_mask = binary_fill_holes(self.temp_mask).T
if add_region:
self.vidTimer.stop()
self.ROI_attrs['centres'].append([np.mean(mpossx),np.mean(mpossy)])
self.ROI_attrs['patches'].append(self.mean_image[yLims[0]:yLims[1],xLims[0]:xLims[1]])
self.ROI_attrs['idxs'].append([mpossx,mpossy])
self.ROI_attrs['masks'].append(self.temp_mask[yLims[0]:yLims[1],xLims[0]:xLims[1]])
if online_trace_extract:
temp = areaFile[:,yLims[0]:yLims[1],xLims[0]:xLims[1]] *self.ROI_attrs['masks'][-1]
temp = temp.astype('float64')
temp[temp==0] = np.nan
self.ROI_attrs['traces'].append(np.nanmean(temp,axis=(1,2)))
#self.ROI_attrs['mask_arr'].append(temp_mask)
self.Gplt.clear()
for i in self.stimattrs['clicks']:
clt_L = pg.InfiniteLine(pos=i,angle=90,movable=False)
self.Gplt.addItem(clt_L)
self.Gplt.addItem(self.timeLine)
self.Gplt.plot(self.ROI_attrs['traces'][-1])
else:
self.ROI_attrs['traces'].append([0])
self.vidTimer.start(self.IFI)
self.mask[:,:,0] += self.mask[:,:,1]
self.mask[:,:,1] = 0
self.mask[:,:,1] = self.temp_mask.T
#self.mask[:,:,0] += self.temp_mask.T
self.mask[:,:,3] += self.temp_mask.T
self.nROIs += 1
self.roi_idx = self.nROIs - 1
else:
self.mask[mpossx,mpossy,0] = 0
self.mask[mpossx,mpossy,3] = 0
else:
print 'Cannot draw ROI, out of bounds.'
self.temp_mask = np.zeros(self.temp_mask.shape)
def get_mask_boundary(masks, radius_dil=2, radius_ero=2):
mask_boundaries = np.zeros(masks.shape)
for i, mask in enumerate(masks):
mask_img = sitk.GetImageFromArray(mask, isVector=False)
mask_dilated = sitk.GrayscaleDilate(mask_img, radius_dil)
mask_dilated = binary_fill_holes(
sitk.GetArrayFromImage(mask_dilated)).astype(mask.dtype)
mask_eroded = sitk.GrayscaleErode(mask_img, radius_ero)
mask_eroded = binary_fill_holes(
sitk.GetArrayFromImage(mask_eroded)).astype(mask.dtype)
mask_boundaries[i] = mask_dilated - mask_eroded
return mask_boundaries
def load_data(patient_path, mask_path):
ct_scan = read_ct_scan(patient_path)
mask = load(mask_path)
patient = get_pixels_hu(ct_scan)
patient, spacing = resample(patient, ct_scan, SPACING)
mask, spacing = resample(mask, ct_scan, SPACING)
mask = morphology.binary_fill_holes(
morphology.binary_dilation(morphology.binary_fill_holes(mask > 0),
iterations=4))
return patient, mask
def norm_spray(test_path, flat_path, dark_path, save_fld, mask_fld):
if 'SC' in test_path:
rot = -1.5
otsu_scale = 0.1
erode_iter = 7
dilat_iter = 10
else:
rot = -1
otsu_scale = 0.01
data = np.array(Image.open(test_path))
flat = np.array(Image.open(flat_path))
dark = np.array(Image.open(dark_path))
data_norm = (data - dark) / (flat - dark)
# Calculate and apply offset
ofst = np.nanmedian(data_norm[52:52 + 52, 52:52 + 52])
data_norm /= ofst
data_norm = rotate(data_norm, rot, preserve_range=True, cval=np.nan)
# Apply median filter to remove spurious pixels
data_norm = median_filter(data_norm, 3)
mask = np.zeros(data_norm.shape, dtype='single')
mask[20:, :] = 1
data_filt = median_filter(data_norm, 13)
data_atten = 1 - data_filt
data_atten[data_atten < 0] = 0
float2int = 10000
int2float = 1 / float2int
data_atten16 = np.rint(data_atten * float2int).astype(np.uint16)
data_thresh = data_atten16 * mask.astype(np.uint16)
thresh_otsu = threshold_otsu(data_thresh[~np.isnan(data_thresh)])
spray_mask = data_thresh > (otsu_scale * thresh_otsu)
if 'SC' in test_path:
spray_mask = binary_erosion(spray_mask, iterations=erode_iter)
spray_mask = binary_erosion(spray_mask, iterations=dilat_iter)
else:
spray_mask = binary_erosion(spray_mask, iterations=5)
spray_mask = binary_dilation(spray_mask, iterations=11)
spray_mask = binary_fill_holes(spray_mask)
spray_mask = binary_dilation(spray_mask, iterations=11)
spray_mask = binary_fill_holes(spray_mask)
# Save Transmission images
im = Image.fromarray(data_norm)
im.save(save_fld + '/' + split(test_path)[1])
# Save mask images
im = Image.fromarray(spray_mask)
im.save(mask_fld + '/' + split(test_path)[1])
def compute_volume_pixels(lung):
col = lung.shape[1]
left_volume = 0
right_volume = 0
struct_opening_post = disk(2)
struct_closing = disk(10)
filled = np.copy(lung)
label_image = np.zeros(lung[..., 0].shape)
max_area = 0
for i in range(lung.shape[2]):
if np.count_nonzero(lung[..., i]) == 0:
continue
label_image_first, num = label(lung[..., i], return_num=True)
coords = regionprops(label_image_first)[0].coords
threshold = coords[0, 0] + ((coords[-1, 0] - coords[0, 0]) // 8)
filled_test = (binary_fill_holes(lung[..., i])).astype(np.uint8) * 255
closed_test = filled_test.copy()
closed_test = (binary_closing(filled_test, struct_closing)).astype(np.uint8) * 255
closed_test = (binary_opening(closed_test, struct_opening_post)).astype(np.uint8) * 255
label_image, num = label(closed_test, return_num=True)
size = 5
opened_high = closed_test.copy()
while size <= 40 and num == 1 and regionprops(label_image)[0].area > 10000:
struct_opening = disk(size)
opened_high = (binary_opening(opened_high, struct_opening)).astype(np.uint8) * 255
label_image, num = label(opened_high, return_num=True)
size += 5
if size <= 40:
closed_test = opened_high
filled_test_second = (binary_fill_holes(closed_test)).astype(np.uint8) * 255
#filled_test_second = (binary_opening(filled_test_second, struct_opening_post)).astype(np.uint8) * 255
label_image, num = label(filled_test_second, return_num=True)
filled[..., i] = filled_test_second
area = 0
for region in regionprops(label_image):
area += region.area
if np.sum(region.coords[:, 1] > col // 2) <= np.sum(region.coords[:, 1] <= col // 2):
left_volume += region.area
else:
right_volume += region.area
return left_volume, right_volume, filled
def makeCervixAndChannelMask(pmapCervixU8, pmapChannelU8):
try:
# (1) load images
mskChn = pmapChannelU8.astype(np.float) / 255.
mskCrv = pmapCervixU8.astype(np.float) / 255.
# (2) preprocess Cervix Mask
thresholdCrv = 0.5
if np.sum(mskCrv>0.5)<50:
thresholdCrv = 0.75 * np.max(mskCrv)
_, R_crv = get_max_blob_mask(mskCrv > thresholdCrv)
msizCRV = math.ceil(R_crv * 0.04)
if msizCRV < 3:
msizCRV = 3
mskCrv_Blob2 = skmorph.closing(mskCrv > thresholdCrv, skmorph.disk(msizCRV))
mskCrv_Blob3, _ = get_max_blob_mask(mskCrv_Blob2 > 0)
mskCrv_Blob4 = binary_fill_holes(mskCrv_Blob3)
# (3) preprocess Channel mask
thresholdChn = 0.5
if np.sum(mskChn>0.5)<50:
thresholdChn = 0.75 * np.max(mskChn)
mskChn_Blob1 = mskChn.copy()
mskChn_Blob1[~mskCrv_Blob4] = 0
# (3.1) check zero-channel-mask
if np.sum(mskChn_Blob1 > thresholdChn) < 1:
mskChn_Blob1 = skmorph.skeletonize(mskCrv_Blob4)
mskChn_Blob1 = skmorph.closing(mskChn_Blob1, skmorph.disk(5))
R_chn = 5
else:
_, R_chn = get_max_blob_mask(mskChn_Blob1 > 0.5)
msizChn = math.ceil(R_chn * 0.1)
if msizChn < 3:
msizChn = 3
mskChn_Blob2 = skmorph.closing(mskChn_Blob1 > 0.5, skmorph.disk(msizChn))
mskChn_Blob2[~mskCrv_Blob4] = 0
# (3.1) check zero-channel-mask
if np.sum(mskChn_Blob2 > 0) < 1:
mskChn_Blob2 = skmorph.skeletonize(mskCrv_Blob4)
mskChn_Blob2 = skmorph.closing(mskChn_Blob2, skmorph.disk(5))
#
mskChn_Blob3, _ = get_max_blob_mask(mskChn_Blob2 > 0)
mskChn_Blob4 = binary_fill_holes(mskChn_Blob3)
# (4) Composing output mask
mskShape = pmapCervixU8.shape[:2]
mskOut = 64 * np.ones(mskShape, dtype=np.uint8)
mskOut[mskCrv_Blob4] = 255
mskOut[mskChn_Blob4] = 128
except Exception as err:
print ('\t!!! ERROR !!! [{0}]'.format(err))
mskOut = np.zeros(pmapCervixU8.shape[:2], dtype=np.uint8)
return mskOut
def segment_lung(imgs_with_hu):
masks = create_mask(imgs_with_hu)
h_masks = morphology.binary_fill_holes(masks > 0)
# dilation = extract pixels, smoothes the boundaires
d1_masks = morphology.binary_dilation(h_masks, iterations=1)
# fill_holes = fill in holes
h1_masks = morphology.binary_fill_holes(d1_masks)
imgs_segmented_lung = imgs_with_hu * h1_masks
return imgs_segmented_lung
def FillHole(bin_image, invert=False):
# files holes
values = np.unique(bin_image)
if len(values) > 2:
print "Not binary image"
return []
background = min(values)
bin_image -= background
bin_image[bin_image > 0] = 1
if invert:
bin_image -= 1
bin_image[bin_image < 0] = 1
result = np.copy(bin_image)
binary_fill_holes(bin_image, output=result)
return result
def load_data(in_folder, pixel_size):
name_add = "" # "_large
mask_traction = np.load(os.path.join(in_folder, "traction_mask.npy"))
mask_traction = binary_fill_holes(mask_traction)
mask_cell_border = np.load(os.path.join(in_folder, "cell_border_mask.npy"))
mask_cell_border = binary_fill_holes(mask_cell_border)
fx = np.load(os.path.join(in_folder, "fx%s.npy"%name_add))
fy = np.load(os.path.join(in_folder, "fy%s.npy"%name_add))
tx = fx / ((pixel_size * 10 ** -6) ** 2)
ty = fy / ((pixel_size * 10 ** -6) ** 2)
u = np.load(os.path.join(in_folder, "u%s.npy"%name_add))
v = np.load(os.path.join(in_folder, "v%s.npy"%name_add))
return mask_traction, mask_cell_border, fx, fy, tx, ty, u, v
def fill_hole(bin_image, invert=False):
# files holes
values = np.unique(bin_image)
if len(values) > 2:
print("Not binary image")
return []
background = min(values)
bin_image -= background
bin_image[bin_image > 0] = 1
if invert:
bin_image -= 1
bin_image[bin_image < 0] = 1
result = np.copy(bin_image)
binary_fill_holes(bin_image, output=result)
return result
def Fg_extract(frame,type = 1): #extract foreground
if type ==1:
mu[:] = alpha*frame + (1.0-alpha)*mu_old
mu_old[:] = mu
sig2[:] = alpha*(1.0*frame-mu)**2 + (1.0-alpha)*sig2_old
sig2_old[:] = sig2
sig = sig2**0.5
lmcs = lmc*sig
bmcs = bmc*sig
fg= np.abs(1.0*frame-mu)[:,:,0]-1*sig[:,:,0]>0.0
elif type == 2:
try:
fg = np.abs(1.0*frame.mean(2)-BG)>50.0
except:
BG = pickle.load(open("bg13-19.pkl","rb"))
BG = cv2.resize(BG,(0,0),fx = 0.5,fy=0.5)
fg = np.abs(1.0*frame.mean(2)-BG)>50.0
fgo = ndm.binary_opening(fg)
fgf = ndm.binary_fill_holes(fgo)
right.set_data(fgf)
plt.draw()
return fgf
def post_processing(image):
image_post = image.copy()
# get texte pixel
illu = 255*(np.sum((image -[0,0,255])**2,axis=2)<10).astype(np.uint8)
# fill holes
illu_out = binary_fill_holes(illu)
image_post[illu_out>0,:] = [0,0,255]
# get illustration pixel
illu = 255*(np.sum((image -[255,0,0])**2,axis=2)<10).astype(np.uint8)
# get bounding-box of connected components
bbox = pymorph.blob(measure.label(illu),'boundingbox','data')
illu_out = illu.copy()
# transform connected components into englobing rectangles
for l in bbox:
x1 = l[0]
y1 = l[1]
x2 = l[2]
y2 = l[3]
if ((y2-y1)<image.shape[0]/2) | ((x2-x1)<image.shape[1]/2):
illu_out[y1:y2,x1:x2]=255
image_post[illu_out>0,:] = [255,0,0]
return image_post
def refine_aseg(aseg, ball_size=4):
"""
First step to reconcile ANTs' and FreeSurfer's brain masks.
Here, the ``aseg.mgz`` mask from FreeSurfer is refined in two
steps, using binary morphological operations:
1. With a binary closing operation the sulci are included
into the mask. This results in a smoother brain mask
that does not exclude deep, wide sulci.
2. Fill any holes (typically, there could be a hole next to
the pineal gland and the corpora quadrigemina if the great
cerebral brain is segmented out).
"""
# Read aseg data
bmask = aseg.copy()
bmask[bmask > 0] = 1
bmask = bmask.astype(np.uint8)
# Morphological operations
selem = sim.ball(ball_size)
newmask = sim.binary_closing(bmask, selem)
newmask = binary_fill_holes(newmask.astype(np.uint8), selem).astype(np.uint8)
return newmask.astype(np.uint8)
def find_ellipse(img, background=None, threshrange=[1,254], sizerange=[10,400], dist_thresh=10, erode=False, check_centers=False, autothreshpercentage=None, show=False):
#print '**img shape** ', img.shape
body = find_object(img, background=background, threshrange=threshrange, sizerange=sizerange, dist_thresh=dist_thresh, erode=erode, check_centers=check_centers, autothreshpercentage=autothreshpercentage)
if body.sum() < 1 and check_centers==True:
body = find_object(img, background=background, threshrange=threshrange, sizerange=sizerange, dist_thresh=dist_thresh, erode=erode, check_centers=check_centers, autothreshpercentage=autothreshpercentage)
body = binary_fill_holes(body)
if body.sum() < 1:
body[body.shape[0] / 2, body.shape[1] / 2] = 1
center, longaxis, shortaxis, ratio = get_ellipse_cov(body, erode=False, recenter=True)
if show:
fig = plt.figure()
ax = fig.add_subplot(111)
ax.imshow(img)
circle = patches.Circle((center[1], center[0]), 2, facecolor='white', edgecolor='none')
ax.add_artist(circle)
ax.plot([center[1]-longaxis[1]*ratio[0], center[1]+longaxis[1]*ratio[0]], [center[0]-longaxis[0]*ratio[0], center[0]+longaxis[0]*ratio[0]], zorder=10, color='white')
return center, longaxis, shortaxis, body, ratio
def detect_sources(snmap, threshold):
hot = (snmap > threshold)
hot = binary_dilation(hot, iterations=2)
hot = binary_fill_holes(hot)
blobs,nblobs = label(hot)
print(nblobs, 'blobs')
#print('blobs min', blobs.min(), 'max', blobs.max())
slices = find_objects(blobs)
px,py = [],[]
for i,slc in enumerate(slices):
blob_loc = blobs[slc]
sn_loc = snmap[slc]
imax = np.argmax((blob_loc == (i+1)) * sn_loc)
y,x = np.unravel_index(imax, blob_loc.shape)
y0,x0 = slc[0].start, slc[1].start
px.append(x0+x)
py.append(y0+y)
#if i == 0:
# plt.subplot(2,2,1)
# plt.imshow(blob_loc, interpolation='nearest', origin='lower')
# plt.colorbar()
# plt.subplot(2,2,2)
# plt.imshow((blob_loc==(i+1))*sn_loc, interpolation='nearest', origin='lower')
# plt.subplot(2,2,3)
# plt.plot(x, y, 'ro')
return np.array(px),np.array(py)
def crop_face(image):
"""takes as input an rgb image of a centered face. crops the skin part out using the HSV distribution of the center
part. Fills in the holes and returns the cropped image as a grey-value matrix"""
assert shape(image)==(300,300,3)
subimage = image[110:190,110:190]
subimage_hsv=cv2.cvtColor(subimage, cv2.COLOR_RGB2HSV)
image_hsv=cv2.cvtColor(image, cv2.COLOR_RGB2HSV)
h,w,colors=shape(subimage)
HSV=zeros([h*w,3])
for i in range(3):
HSV[:,i]=ravel(subimage_hsv[:,:,i])
Mean_hsv=mean(HSV,axis=0)
STD_hsv=std(HSV,axis=0)
notface=sum(((image_hsv-Mean_hsv)/STD_hsv)**2,axis=2)
mask=(1*(notface<10)).astype(int)
label_im, nb_labels = ndimage.label(mask)
sizes = ndimage.sum(mask, label_im, range(nb_labels + 1))
mask=label_im==argmax(sizes)
mask2=binary_fill_holes(mask)
image2=copy(image)
for i in range(3):
image2[:,:,i] *= mask2 #2.astype(int64)
grey_image=sum(image2,axis=2)
return grey_image, image2
def remove_background (image):
# remove frame if any
image = image[5:-5, 5:-5, :]
thresh = np.array([10,10,10]).astype(int)
c = get_dominant_colors(image, bin_size=thresh/4)
if c is None: return None
b, g, r = np.rollaxis(image, axis=-1)
c = c.astype(int)
mask = np.ones(r.shape, dtype=bool)
mask = np.bitwise_and(mask, b.astype(int) >= c[0]-thresh[0])
mask = np.bitwise_and(mask, b.astype(int) <= c[0]+thresh[0])
mask = np.bitwise_and(mask, g.astype(int) >= c[1]-thresh[1])
mask = np.bitwise_and(mask, g.astype(int) <= c[1]+thresh[1])
mask = np.bitwise_and(mask, r.astype(int) >= c[2]-thresh[2])
mask = np.bitwise_and(mask, r.astype(int) <= c[2]+thresh[2])
mask = np.invert(binary_fill_holes(np.invert(mask)))
for i in range(DILATE_ITER):
mask = binary_dilation(mask)
r[mask] = 0
g[mask] = 0
b[mask] = 0
image = np.dstack((b, g, r, np.invert(mask.astype(np.uint8)*255)))
return image
def postproc_masks(im, thrs=[], fill_holes=True, output='', report={}):
"""Apply slicewise thresholds to data and fill holes.
NOTE: zyx assumed
"""
ods = 'mask_thr{:05d}'.format(0)
im.load(load_data=False)
if thrs:
data = im.slice_dataset()
mask = np.zeros(im.dims[:3], dtype='bool')
for slc in range(0, mask.shape[0]):
mask[slc, :, :] = data[slc, :, :] > thrs[slc]
else:
mask = im.slice_dataset()
im.close()
if fill_holes:
for slc in range(0, mask.shape[0]):
mask[slc, :, :] = binary_fill_holes(mask[slc, :, :])
props = im.get_props()
mo = write_data(mask, props, output, ods)
c_slcs = {dim: get_centreslice(mo, '', dim) for dim in 'zyx'}
report['centreslices'][ods] = c_slcs
return mo, report
def main():
# catch parameters
forest_file = sys.argv[1]
case_folder = sys.argv[2]
mask_file = sys.argv[3]
segmentation_file = sys.argv[4]
# loading case features
feature_vector = []
for _file in os.listdir(case_folder):
if _file.endswith('.npy') and _file.startswith('feature.'):
with open(os.path.join(case_folder, _file), 'r') as f:
feature_vector.append(numpy.load(f))
feature_vector = join(*feature_vector)
if 1 == feature_vector.ndim:
feature_vector = numpy.expand_dims(feature_vector, -1)
# load and apply the decision forest
with open(forest_file, 'r') as f:
forest = pickle.load(f)
classification_results = forest.predict(feature_vector)
# preparing image
m, h = load(mask_file)
m = m.astype(numpy.bool)
o = numpy.zeros(m.shape, numpy.uint8)
o[m] = numpy.squeeze(classification_results).ravel()
# applying the post-processing morphology
#o = binary_dilation(o, iterations=2)
#o = keep_largest_connected_component(o)
o = binary_fill_holes(o)
# savin the results
save(o, segmentation_file, h, True)
def get_body_mask(ct_data, win_min=1200, morphology_radius=2, connectivity=3):
"""
Get body mask from CT image.
Argument `win_min` should approx to air intensity value.
"""
body_mask = (ct_data>=win_min)# & (ct_data<=win_max)
#print(' {} of {} voxels masked.'.format(np.sum(body_mask),np.size(body_mask)))
if np.sum(body_mask)==0:
raise ValueError('BODY could not be extracted!')
# Find largest connected component in 3D
struct = generate_binary_structure(3,connectivity)
body_mask = binary_erosion(body_mask,structure=struct,iterations=morphology_radius)
if np.sum(body_mask)==0:
raise ValueError('BODY mask disappeared after erosion!')
# Get the largest connected component
labeled_array, num_features = label(body_mask, structure=struct)
component_sizes = np.bincount(labeled_array.ravel())
max_label = np.argmax(component_sizes[1:-1])+1
# only keep largest, dilate again and fill holes
body_mask = binary_dilation(labeled_array==max_label,structure=struct,iterations=morphology_radius)
# Fill holes slice-wise
for z in range(0,body_mask.shape[2]):
body_mask[:,:,z] = binary_fill_holes(body_mask[:,:,z])
return body_mask
def Fg_extract(frame): #extract foreground
mu[:] = alpha*frame + (1.0-alpha)*mu_old
mu_old[:] = mu
sig2[:] = alpha*(1.0*frame-mu)**2 + (1.0-alpha)*sig2_old
sig2_old[:] = sig2
sig = sig2**0.5
lmcs = lmc*sig
bmcs = bmc*sig
fg= np.abs(1.0*frame-mu)[:,:,0]-2*sig[:,:,0]>0.0
fgo = ndm.binary_opening(fg)
fgf = ndm.binary_fill_holes(fgo)
right.set_data(fgf)
plt.draw()
'''
if imsave:
im = zeros(frame.shape)
a = fgf.astype(np.uint8)*255
im[:,:,0]=a
im[:,:,1]=a
im[:,:,2]=a
im = Image.fromarray(im.astype(np.uint8))
im.save('/home/andyc/image/tracking VIDEO0004/binary/b%.3d.bmp'%vid_idx)
'''
return fgf
def segment_roi(roi):
# step 1. phase congruency (edge detection)
Mm = phasecong_Mm(roi)
# step 2. hysteresis thresholding (of edges)
B = hysthresh(Mm,HT_T1,HT_T2)
# step 3. trim pixels off border
B[B[:,1]==0,0]=0
B[B[:,-2]==0,-1]=0
B[0,B[1,:]==0]=0
B[-1,B[-2,:]==0]=0
# step 4. threshold to find dark areas
dark = dark_threshold(roi, DARK_THRESHOLD_ADJUSTMENT)
# step 5. add dark areas back to blob
B = B | dark
# step 6. binary closing
B = binary_closing(B,SE3)
# step 7. binary dilation
B = binary_dilation(B,SE2)
# step 8. thinning
B = bwmorph_thin(B,3)
# step 9. fill holes
B = binary_fill_holes(B)
# step 10. remove blobs smaller than BLOB_MIN
B = remove_small_objects(B,BLOB_MIN,connectivity=2)
# done.
return B
def get_segmented_lungs(im, plot=False, THRESHOLD=-320):
binary = im < THRESHOLD
cleared = clear_border(binary)
label_image = measure.label(cleared)
areas = [r.area for r in measure.regionprops(label_image)]
areas.sort()
#print areas
if len(areas) > 2:
for region in measure.regionprops(label_image):
if region.area < areas[-2]:
for coordinates in region.coords:
label_image[coordinates[0], coordinates[1]] = 0
binary = label_image > 0
selem = morphology.disk(2)
binary = morphology.binary_erosion(binary, selem)
selem = morphology.disk(15)
binary = morphology.binary_closing(binary, selem)
edges = filters.roberts(binary)
binary = binary_fill_holes(edges)
get_high_vals = binary == 0
im[get_high_vals] = 0
return im
def _label_fill_holes(im_label):
im_label_out = np.copy(im_label)
for j in regionprops(im_label):
mask = (im_label == j.label)
mask = binary_fill_holes(mask)
im_label_out[mask] = j.label
return im_label_out
def clean_seg(seg):
labels = label(seg)
labels = remove_small_objects(labels)
seg = labels > 0.5
seg = binary_closing(seg, structure=np.ones((10, 10)))
seg = binary_fill_holes(seg)
return seg
def mask_polar_to_cart(mask, center, min_radius, max_radius, output_shape, zoom_factor=1):
'''Converts a polar binary mask to Cartesian and places in an image of zeros'''
# Account for upsampling
if zoom_factor != 1:
center = (center[0]*zoom_factor + zoom_factor/2, center[1]*zoom_factor + zoom_factor/2)
min_radius = min_radius * zoom_factor
max_radius = max_radius * zoom_factor
output_shape = map(lambda a: a * zoom_factor, output_shape)
# new image
image = np.zeros(output_shape)
# coordinate conversion
theta, r = np.meshgrid(np.linspace(0, 2*np.pi, mask.shape[1]),
np.arange(0, max_radius))
x, y = coord_polar_to_cart(r, theta, center)
x, y = np.round(x), np.round(y)
x, y = x.astype(int), y.astype(int)
x = np.clip(x, 0, image.shape[0]-1)
y = np.clip(y, 0, image.shape[1]-1)
ix,iy = np.meshgrid(np.arange(0,mask.shape[1]), np.arange(0,mask.shape[0]))
image[x,y] = mask
# downsample image
if zoom_factor != 1:
zf = 1/float(zoom_factor)
image = zoom(image, (zf, zf), order=4)
# ensure image remains a filled binary mask
image = (image > 0.5).astype(int)
image = binary_fill_holes(image)
return image
def get_FOV(around_lung, lung):
FOV = np.where((around_lung + lung) > 0, 1, 0)
for idx in range(FOV.shape[0]):
FOV[idx, :, :] = binary_fill_holes(FOV[idx, :, :],
structure=np.ones(
(5, 5))).astype(FOV.dtype)
return FOV
def binary(self, img="", tol=15):
if isinstance(img, str): img = self.data
gray = self.grayscale(img, type='max')
gray[gray < tol] = 0
gray[gray > tol] = 255
gray = binary_fill_holes(gray).astype(np.uint8)*255
return gray
def _fill_label_holes(lbl_img, **kwargs):
lbl_img_filled = np.zeros_like(lbl_img)
for l in (set(np.unique(lbl_img)) - set([0])):
mask = lbl_img == l
mask_filled = binary_fill_holes(mask, **kwargs)
lbl_img_filled[mask_filled] = l
return lbl_img_filled
def proc_np_dst(pred):
"""
Process Nuclei Prediction with Distance Map
Args:
pred: prediction output, assuming
channel 0 contain probability map of nuclei
channel 1 containing the regressed distance map
"""
blb_raw = pred[...,0]
dst_raw = pred[...,1]
blb = np.copy(blb_raw)
blb[blb > 0.5] = 1
blb[blb <= 0.5] = 0
blb = measurements.label(blb)[0]
blb = remove_small_objects(blb, min_size=30)
blb[blb > 0] = 1
dst_raw[dst_raw < 0] = 0
dst = np.copy(dst_raw)
dst = dst * blb
dst[dst > 0.5] = 1
dst[dst <= 0.5] = 0
marker = dst.copy()
marker = binary_fill_holes(marker)
marker = measurements.label(marker)[0]
marker = remove_small_objects(marker, min_size=30)
proced_pred = watershed(-dst_raw, marker, mask=blb)
return proced_pred
def refine_aseg(aseg, ball_size=4):
"""
Refine the ``aseg.mgz`` mask of Freesurfer.
First step to reconcile ANTs' and FreeSurfer's brain masks.
Here, the ``aseg.mgz`` mask from FreeSurfer is refined in two
steps, using binary morphological operations:
1. With a binary closing operation the sulci are included
into the mask. This results in a smoother brain mask
that does not exclude deep, wide sulci.
2. Fill any holes (typically, there could be a hole next to
the pineal gland and the corpora quadrigemina if the great
cerebral brain is segmented out).
"""
from skimage import morphology as sim
from scipy.ndimage.morphology import binary_fill_holes
# Read aseg data
bmask = aseg.copy()
bmask[bmask > 0] = 1
bmask = bmask.astype(np.uint8)
# Morphological operations
selem = sim.ball(ball_size)
newmask = sim.binary_closing(bmask, selem)
newmask = binary_fill_holes(newmask.astype(np.uint8), selem).astype(np.uint8)
return newmask.astype(np.uint8)
def hide_image_elements(img, bin_mask):
"""
where mask, change pixels' brightness min(img) value
"""
bin_mask = bin_mask.astype(bool)
bin_mask = binary_fill_holes(bin_mask)
return bin_mask * np.min(img) + img * (~bin_mask.astype(bool))
def get_chest_boundary(im, plot=False):
size = im.shape[1]
if plot == True:
f, plots = plt.subplots(6, 1, figsize=(5, 30))
binary = im < -320
if plot == True:
plots[0].axis('off')
plots[0].imshow(binary, cmap=plt.cm.bone)
cleared = clear_border(binary)
temp_label = label(cleared)
for region in regionprops(temp_label):
if region.area < 300:
for coordinates in region.coords:
temp_label[coordinates[0], coordinates[1]] = 0
cleared = temp_label > 0
label_img = label(cleared)
for region in regionprops(label_img):
if region.eccentricity > 0.99 \
or region.centroid[0] > 0.90 * size \
or region.centroid[0] < 0.12 * size \
or region.centroid[1] > 0.88 * size \
or region.centroid[1] < 0.10 * size \
or (region.centroid[1] > 0.46 * size and region.centroid[1] < 0.54 * size and region.centroid[
0] > 0.75 * size) \
or (region.centroid[0] < 0.2 * size and region.centroid[1] < 0.2 * size) \
or (region.centroid[0] < 0.2 * size and region.centroid[1] > 0.8 * size) \
or (region.centroid[0] > 0.8 * size and region.centroid[1] < 0.2 * size) \
or (region.centroid[0] > 0.8 * size and region.centroid[1] > 0.8 * size):
for coordinates in region.coords:
label_img[coordinates[0], coordinates[1]] = 0
if plot == True:
plots[1].axis('off')
plots[1].imshow(label_img, cmap=plt.cm.bone)
region_n = np.max(label_img)
selem = disk(10)
filled = np.zeros(cleared.shape, np.uint8)
for i in range(1, region_n + 1):
curr_region = np.zeros(cleared.shape, np.uint8)
curr_region[label_img == i] = 1
curr_region = binary_closing(curr_region, selem)
curr_region = binary_fill_holes(curr_region)
filled[curr_region == 1] = 1
if plot == True:
plots[2].axis('off')
plots[2].imshow(filled, cmap=plt.cm.bone)
filled_edge = misc.imfilter(filled.astype(np.float64), 'find_edges') / 255
if plot == True:
plots[3].axis('off')
plots[3].imshow(filled_edge, cmap=plt.cm.bone)
return filled_edge
def extract_foreground(img):
"""
Extracts the single largest object from grayscale image img.
Returns a boolean mask and a skimage RegionProperty for that object.
"""
thresholds = threshold_multiotsu(img, classes=3)
true_fg = np.digitize(img, [thresholds[0]])
mask_edge = get_edge_mask(true_fg, 100, 20)
masked_img = extract_mask(img, mask_edge)
new_true_fg = new_edge_threshold(masked_img, mask_edge, true_fg)
new_mask_edge = get_edge_mask(new_true_fg, 50, 10)
new_masked_img = extract_mask(img, new_mask_edge)
last_true_fg = new_edge_threshold(
new_masked_img,
new_mask_edge,
new_true_fg,
)
labels = label(last_true_fg)
props = regionprops(labels, img)
areas = np.asarray([prop.area for prop in props])
ind = areas.argmax()
prop = props[ind]
mask = binary_fill_holes(labels == prop.label)
return mask, prop
def postprocess(X, min_area=0):
"""
performs non_max_suppression within connected components
parameters
__________
X : np.array
the image you are trying to suppress
percentile : int
the fraction of images you are accepting
min_area : int
the minimum area of a component you are accepting
return
__________
predictions : np.array
the thresholded image
"""
otsu_predictions = otsu(X)
labeled, num_labels = label(otsu_predictions)
flattened_labels = labeled.reshape((256 * 256, 1))
flattened_otsu_predictions = otsu_predictions.reshape((256 * 256, 1))
for label_num in range(0, num_labels + 4):
indices = np.where(flattened_labels == label_num)[0]
component = flattened_otsu_predictions[indices]
if (len(indices) > min_area):
flattened_otsu_predictions[indices] = component
else:
flattened_otsu_predictions[indices] = 0
return binary_fill_holes(flattened_otsu_predictions.reshape((256, 256)))
def extract_foreground_biofilms(img, area_threshold=9000):
"""
Extracts the single largest object from grayscale image img.
Returns a boolean mask and a skimage RegionProperty for that object.
"""
thresholds = threshold_multiotsu(img, classes=3)
true_fg = np.digitize(img, [thresholds[0]])
mask_edge = get_edge_mask(true_fg, 100, 20)
masked_img = extract_mask(img, mask_edge)
new_true_fg = new_edge_threshold(masked_img, mask_edge, true_fg)
new_mask_edge = get_edge_mask(new_true_fg, 50, 10)
new_masked_img = extract_mask(img, new_mask_edge)
last_true_fg = new_edge_threshold(
new_masked_img,
new_mask_edge,
new_true_fg,
)
labels = label(last_true_fg)
props = regionprops(labels, img)
areas = np.asarray([prop.area for prop in props])
mask = np.zeros_like(labels)
inds = np.arange(len(areas))[areas > area_threshold]
for i, ind in enumerate(inds):
prop = props[ind]
mask[binary_fill_holes(labels == prop.label)] = i + 1
return mask
def find_all_children(labels):
mask = binary_fill_holes(labels < 0)
mask[labels < 0] = False
clabelnums = np.unique(labels[mask]).tolist()
if 0 in clabelnums:
clabelnums.remove(0)
return clabelnums
def find_ellipse(img,
background=None,
threshrange=[1, 254],
sizerange=[10, 400],
dist_thresh=10,
erode=False,
check_centers=False,
autothreshpercentage=None,
show=False):
#print '**img shape** ', img.shape
body = find_object(img,
background=background,
threshrange=threshrange,
sizerange=sizerange,
dist_thresh=dist_thresh,
erode=erode,
check_centers=check_centers,
autothreshpercentage=autothreshpercentage)
if body.sum() < 1 and check_centers == True:
body = find_object(img,
background=background,
threshrange=threshrange,
sizerange=sizerange,
dist_thresh=dist_thresh,
erode=erode,
check_centers=check_centers,
autothreshpercentage=autothreshpercentage)
body = binary_fill_holes(body)
if body.sum() < 1:
body[body.shape[0] / 2, body.shape[1] / 2] = 1
center, longaxis, shortaxis, ratio = get_ellipse_cov(body,
erode=False,
recenter=True)
if show:
fig = plt.figure()
ax = fig.add_subplot(111)
ax.imshow(img)
circle = patches.Circle((center[1], center[0]),
2,
facecolor='white',
edgecolor='none')
ax.add_artist(circle)
ax.plot([
center[1] - longaxis[1] * ratio[0],
center[1] + longaxis[1] * ratio[0]
], [
center[0] - longaxis[0] * ratio[0],
center[0] + longaxis[0] * ratio[0]
],
zorder=10,
color='white')
return center, longaxis, shortaxis, body, ratio
def _compute_sample_params(data, label, rim, rate_mg, logits_g):
nonmissing = label != 0
zero_count_spots = 1 + torch.where(data.sum(1) == 0)[0]
nonpartial = binary_fill_holes(
np.isin(label.cpu(), [0, *zero_count_spots.cpu()]))
nonpartial = torch.as_tensor(nonpartial).to(nonmissing)
mask = nonpartial & nonmissing
if not mask.any():
return (
data[[]],
torch.zeros(0, num_genes).to(rim),
logits_g.expand(0, -1),
)
label = label[mask] - 1
idxs, label = torch.unique(label, return_inverse=True)
data = data[idxs]
rim = rim[:, mask]
labelonehot = sparseonehot(label)
rim = torch.sparse.mm(labelonehot.t().float(), rim.t())
rgs = rim @ rate_mg.exp()
return data, rgs, logits_g.expand(len(rgs), -1)
def postprocess_img(img, close_ksize=5, size_factor=0.05):
""" Postprocess the binary mask from kmeans clustering in order to fill holes and remove small elements to grab the main epithelial sheet.
Parameters
-----------
img : numpy array
(n_rows x n_cols) binary image
close_ksize : int
kernel size for binary_closing with a disk kernel.
size_factor : float
0-1, the proportion of image area below which are regarded as small objects and removed from the binary mask
Returns
-------
filt : numpy array
(n_rows x n_cols), post-processed binary image.
"""
from scipy.ndimage.morphology import binary_fill_holes
from skimage.morphology import disk, binary_closing, remove_small_objects
img_rows, img_cols = img.shape
filt = binary_closing(img, disk(close_ksize))
filt = remove_small_objects(filt, size_factor * img_rows * img_cols)
filt = binary_fill_holes(filt)
return filt
def get_breast_mask(shape, breast, debug_verbose=True):
# Fill all contour points
mask = np.zeros(shape)
x, y = dists.spline(breast, n_points=8000)
x = np.clip(np.round(x).astype(int), 0, mask.shape[0] - 1)
y = np.clip(np.round(y).astype(int), 0, mask.shape[1] - 1)
mask[x, y] = 1
# Fill all points at the top
x = np.clip(np.linspace(breast[0, 0], breast[-1, 0], 2000), 0,
mask.shape[0] - 1)
y = np.clip(np.linspace(breast[0, 1], breast[-1, 1], 2000), 0,
mask.shape[1] - 1)
x = np.round(x).astype(int)
y = np.round(y).astype(int)
mask[x, y] = 1
# Fill black holes of the mask
mask_full = morpho.binary_fill_holes(mask)
if debug_verbose:
plt.clf()
plt.imshow(mask_full)
plt.savefig(config.debug_path + "mask.png")
return mask_full
def match_contour_frames(self, desired_frames):
"""
returns a 3D np array of filled in contours of desired frames
"""
wdf = self.df.copy()
wdf = wdf.reset_index()
wdf = wdf.set_index('frame')
wdf = wdf.loc[desired_frames]
wdf[wdf['index'] >= len(self.contours)]
wdf[wdf['index'] < 0]
desired_contours = np.array(wdf.dropna('index')['index'])
print(np.min(wdf['index']), 'min index')
print(np.min(desired_contours))
shape = list(self.contours.shape)
shape[0] = len(desired_contours)
# xshape = list(self.x_mid.shape)
# xshape[0] = len(desired_contours)
# x_mid2 = np.zeros(xshape)
# yshape = list(self.y_mid.shape)
# yshape[0] = len(desired_contours)
# y_mid2 = np.zeros(yshape)
contours2 = np.zeros(shape)
for i, j in enumerate(desired_contours):
if j > len(self.contours):
print(j, 'wtf')
if j < 0:
print(j, 'wtf')
contours2[i] = morph.binary_fill_holes(self.contours[j])
# x_mid2[i] = self.x_mid[j]
# y_mid2[i] = self.y_mid[j]
return contours2
def postprocess_prediction(pred, gaussian_std=1, threshold=0.5, fill_holes=True, connected_component=True):
pred = gaussian_filter(pred, gaussian_std) > threshold
if fill_holes:
pred = binary_fill_holes(pred)
if connected_component:
pred = get_main_connected_component(pred)
return pred
def draw_segm_mask(mask, vertices, triangles, color):
m = np.zeros_like(mask)
for t in triangles:
coord = np.array([vertices[t[0]][:2], vertices[t[1]][:2], vertices[t[2]][:2]], dtype=np.int32)
cv2.fillConvexPoly(m, coord, 1)
m = binary_fill_holes(m).astype(np.uint8)
mask[m.astype(bool)] = color
def remove_abundant(image,mask):
edges = cv2.bitwise_and(image,image,mask=mask);
red_mean = np.average(edges[:,:,0], weights=edges[:,:,0].astype(bool));
green_mean = np.average(edges[:,:,1], weights=edges[:,:,1].astype(bool));
blue_mean = np.average(edges[:,:,2], weights=edges[:,:,2].astype(bool));
im_temp = copy.copy(image);
im = np.zeros(im_temp.shape,dtype='int');
im[:,:,0] = abs(im_temp[:,:,0] - red_mean);
im[:,:,1] = abs(im_temp[:,:,1] - green_mean);
im[:,:,2] = abs(im_temp[:,:,2] - blue_mean);
im = im.astype('uint8');
im = cv2.bitwise_and(im,im,mask=mask)
red_mean = np.average(im[:,:,0], weights=im[:,:,0].astype(bool));
green_mean = np.average(im[:,:,1], weights=im[:,:,1].astype(bool));
blue_mean = np.average(im[:,:,2], weights=im[:,:,2].astype(bool));
ret1, grad0 = cv2.threshold (im[:,:,0],red_mean, 255, cv2.THRESH_BINARY);
ret1, grad1 = cv2.threshold (im[:,:,1],green_mean, 255, cv2.THRESH_BINARY);
ret1, grad2 = cv2.threshold (im[:,:,2],blue_mean, 255, cv2.THRESH_BINARY);
grad = cv2.bitwise_and(grad0,grad1);
grad = cv2.bitwise_and(grad,grad2);
grad = morphology.binary_closing(grad,iterations=5);
grad = morphology.binary_fill_holes(grad);
grad = grad.astype('uint8');
im = cv2.bitwise_and(image,image,mask=grad);
return (im,grad);
def _fill2d(arr, structure = None, dimension = 2):
r"""
Fill holes along a certain dimension only.
"""
res = numpy.zeros(arr.shape, numpy.bool)
for sl in range(arr.shape[dimension]):
res[:,:,sl] = binary_fill_holes(arr[:,:,sl], structure)
return res
def __morph_ops__(self,mask): mask = cv2.medianBlur(np.uint8(mask),3) mask = cv2.morphologyEx(mask, cv2.MORPH_OPEN, KERNEL) mask = cv2.medianBlur(mask,3) mask = cv2.morphologyEx(mask, cv2.MORPH_CLOSE,KERNEL) mask = binary_fill_holes(mask) mask = remove_small_objects(mask,min_size=128,connectivity=2) return np.uint8(mask);
def find_circle(img, npts=25, nstart=0, navg=20, plot=False):
filled_img = binary_fill_holes(img)
dil_img = binary_dilation(filled_img)
edges = dil_img-filled_img
pts = np.transpose(np.nonzero(edges))
# select an evenly spaced subset of points (to speed up computation):
if len(pts) > npts:
indices = np.linspace(nstart, len(pts)-1, npts)
indices = [int(indices[i]) for i in range(len(indices))]
else:
indices = np.arange(0, len(pts), 1).tolist()
pts_subset = pts[indices,:]
len_pts_diff = np.arange(1,len(pts_subset), 1)
pts_diff = np.zeros(np.sum(len_pts_diff))
pts_diff_arr = np.zeros([np.sum(len_pts_diff), 2])
iarr = 0
for i in range(len(pts_subset)):
indices = np.arange(i+1, len(pts_subset), 1)
pts_diff_arr[iarr:len(indices)+iarr, 1] = indices
pts_diff_arr[iarr:len(indices)+iarr, 0] = np.ones_like(indices)*i
d_arr = pts_subset[indices.tolist(), :] - pts_subset[i,:]
d = np.array( [np.linalg.norm(d_arr[n]) for n in range(len(d_arr))] )
pts_diff[iarr:len(indices)+iarr] = d
iarr += len(indices)
ordered_pairs = np.argsort(pts_diff)[::-1]
best_pairs = pts_diff_arr[(ordered_pairs[0:navg]).tolist()]
center_arr = np.zeros([len(best_pairs), 2])
radius_arr = np.zeros([len(best_pairs), 1])
#centers = np.zeros([len(best_pairs)])
for i, pair in enumerate(best_pairs):
pt1 = np.array(pts_subset[ pair[0] ], dtype=float)
pt2 = np.array(pts_subset[ pair[1] ], dtype=float)
pt_diff = pt2 - pt1
radius_arr[i] = np.linalg.norm( pt_diff ) / 2.
center_arr[i] = pt1 + pt_diff/2
center = np.mean(center_arr, axis=0)
radius = np.mean(radius_arr)
if plot:
fig = plt.figure(None)
ax = fig.add_axes([.1,.1,.8,.8])
circle = patches.Circle( center, radius=radius, facecolor='none', edgecolor='green')
ax.add_artist(circle)
ax.imshow(edges)
return center, radius
def proc_roi_region(self,add_region=True):
mpossx = self.roi_item.getArrayRegion(self.possx,self.img).astype(int)
mpossx = mpossx[np.nonzero(binary_fill_holes(mpossx))]#get the x pos from ROI
mpossy = self.roi_item.getArrayRegion(self.possy,self.img).astype(int)
mpossy = mpossy[np.nonzero(binary_fill_holes(mpossy))]# get the y pos from ROI
xLims = [np.min(mpossx)-10,np.max(mpossx)+10]
yLims = [np.min(mpossy)-10,np.max(mpossy)+10]
#xLims = [np.mean(mpossx)-20,np.mean(mpossx)+20]; yLims = [np.mean(mpossy)-20,np.mean(mpossy)+20]
self.temp_mask[mpossx,mpossy] = 1
self.temp_mask = binary_fill_holes(self.temp_mask).T
if add_region:
self.vidTimer.stop()
self.ROI_attrs['centres'].append([np.mean(mpossx),np.mean(mpossy)])
self.ROI_attrs['patches'].append(self.mean_image[yLims[0]:yLims[1],xLims[0]:xLims[1]])
self.ROI_attrs['idxs'].append([mpossx,mpossy])
self.ROI_attrs['masks'].append(self.temp_mask[yLims[0]:yLims[1],xLims[0]:xLims[1]])
if online_trace_extract:
temp = areaFile[:,yLims[0]:yLims[1],xLims[0]:xLims[1]] *self.ROI_attrs['masks'][-1]
temp = temp.astype('float64')
temp[temp==0] = np.nan
self.ROI_attrs['traces'].append(np.nanmean(temp,axis=(1,2)))
#self.ROI_attrs['mask_arr'].append(temp_mask)
self.Gplt.clear()
self.Gplt.addItem(self.timeLine)
self.Gplt.plot(self.ROI_attrs['traces'][-1])
else:
self.ROI_attrs['traces'].append([0])
self.vidTimer.start(self.IFI)
self.mask[:,:,0] += self.mask[:,:,1]
self.mask[:,:,1] = 0
self.mask[:,:,1] = self.temp_mask.T
#self.mask[:,:,0] += self.temp_mask.T
self.mask[:,:,3] += self.temp_mask.T
self.nROIs += 1
self.roi_idx = self.nROIs - 1
else:
self.mask[mpossx,mpossy,0] = 0
self.mask[mpossx,mpossy,3] = 0
self.temp_mask = np.zeros(self.temp_mask.shape)
def get_ellipse_cov(img, erode=False, recenter=True):
# Pattern. Recogn. 20, Sept. 1998, pp. 31-40
# J. Prakash, and K. Rajesh
# Human Face Detection and Segmentation using Eigenvalues of Covariance Matrix, Hough Transform and Raster Scan Algorithms
#eroded_img = binary_erosion(img)
#boundary = img-eroded_img
if img is not None:
if erode is not False:
try:
e = 0
while e < erode:
e += 1
img = binary_erosion(img)
except:
pass
img = binary_fill_holes(img)
if recenter:
center = center_of_blob(img)
else:
center = np.array([0,0])
if 1:
ptsT = np.transpose(np.nonzero(img))
for pt in ptsT:
pt -= center
pts = (ptsT).T
cov = np.cov(pts)
cov = np.nan_to_num(cov)
e,v = np.linalg.eig(cov)
longaxis = v[:,np.argmax(e)]
shortaxis = v[:,np.argmin(e)]
if len(ptsT) > 2:
dl = [np.dot(longaxis, ptsT[i]) for i in range(len(ptsT))]
longaxis_radius = np.max( np.abs(dl) )
ds = [np.dot(shortaxis, ptsT[i]) for i in range(len(ptsT))]
shortaxis_radius = np.max( np.abs(ds) )
else:
longaxis_radius = None
shortaxis_radius = None
if recenter is False:
return longaxis, shortaxis, [longaxis_radius, shortaxis_radius]
else:
return center, longaxis, shortaxis, [longaxis_radius, shortaxis_radius]
else:
return [0,0],0
def __morphologicalOps__(mask): #_mask = binary_fill_holes(mask) _mask = cv2.medianBlur(np.uint8(mask),3) _mask = cv2.morphologyEx(_mask, cv2.MORPH_CLOSE,KERNEL) _mask = binary_fill_holes(_mask) _mask = remove_small_objects(_mask,min_size=128,connectivity=2) kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(5,5)) _mask = cv2.dilate(np.uint8(_mask),kernel,iterations = 1) return _mask;
def seg_sect(self, img):
img_canny = canny(img, sigma=self.sigma,
low_threshold=self.low_threshold)
img_dilate = binary_dilation(img_canny, square(3))
img_erode = binary_erosion(img_dilate, square(3))
img_fill = binary_fill_holes(img_erode)
return img_fill
def compute_skeleton_from_outline(outline, return_intermediate_steps=False, verbose=False):
'''
accepts a list of points in the outline of a shape and returns the winding centerline 1 pixel wide of that shape
:param outline: a list of x,y tuples containing the points around the shape's outline
:param return_intermediate_steps: a toggle to show internal guts of the process.
'''
# occasional outlines do not form a fully closed shape. this closes it.
xs, ys = zip(*close_outline_border(outline))
# make a box that is big enough to hold the outline + borders + a safety margin
border_size, safety_margin = 1, 1
matrix_x = max(xs) - min(xs) + 2 * border_size + safety_margin
matrix_y = max(ys) - min(ys) + 2 * border_size + safety_margin
outline_matrix = np.zeros([matrix_x, matrix_y], dtype=int)
# add outline to the matrix by shifting the xy coordinates accordingly
xy_shift = [min(xs) - border_size, min(ys) - border_size]
for x, y in zip(xs, ys):
outline_matrix[x - xy_shift[0]][y - xy_shift[1]] = 1
# fill in the inside of the outline to make a solid shape
filled_matrix = binary_fill_holes(outline_matrix)
# if something is wrong, save the offending outline for later
if sum(sum(outline_matrix)) == sum(sum(filled_matrix)):
write_pathological_input(outline.tolist(), input_type='outline', note='outline filling problem',
savename='%sfill_outline_%s.json' %(exception_directory, str(time.time())))
assert False, 'you have an outline filling problem'
# thin the solid shape until it is 1px thick, then remove all shortest branches.
spine_matrix_branched = skeletonize(filled_matrix)
if return_intermediate_steps:
spine_matrix_branched_copy = spine_matrix_branched.copy()
spine_matrix, endpoints = cut_branchpoints_from_spine_matrix(spine_matrix_branched)
# the option to return matrices from intermediate steps is for plotting
if return_intermediate_steps:
return outline_matrix, filled_matrix, spine_matrix_branched_copy, spine_matrix, xy_shift, endpoints
if len(endpoints) < 2:
if sum(sum(spine_matrix)) > 2:
# if spine is long enough to have endpoints, something is wrong.
# this saves the offending outline for later tests
write_pathological_input(outline.tolist(),
input_type='outline', note='less than 2 endpoints. long enough spine.',
savename='%sendpoints_%s.json' %(exception_directory, str(time.time())))
assert False, 'spine is long enough, but does not have endpoints' + str(endpoints)
else:
# short spines can't be helped.
if verbose:
print('warning: spine too short to find endpoints!')
return []
# change spine matrix back to a list of points and reverse the previous coordinate shift
shifted_spine = line_matrix_to_ordered_points(spine_matrix, endpoints)
real_spine = [(pt[0] + xy_shift[0], pt[1] + xy_shift[1]) for pt in shifted_spine]
return real_spine
def neuropil_correct(areaF,roi_attrs):
nROIs = len(roiattrs['idxs'])
len_trace = areaFile.shape[0]
roiattrs['traces'] = np.zeros([nROIs,len_trace])
roiattrs['neuropil_traces'] = np.zeros([nROIs,len_trace])
roiattrs['corr_traces'] = np.zeros([nROIs,len_trace])
for idx in range(nROIs):
sys.stdout.write('\r Extracting_Trace_from roi: %s' %idx)
sys.stdout.flush()
mpossx= roi_attrs['idxs'][idx][0]
mpossy = roi_attrs['idxs'][idx][1]
xLims = [np.clip(np.min(mpossx)-10,0,510),np.clip(np.max(mpossx)+10,0,510)]
yLims = [np.clip(np.min(mpossy)-10,0,510),np.clip(np.max(mpossy)+10,0,510)]
temp_mask = np.zeros(areaFile.shape[1:])
temp_mask[mpossx,mpossy] = 1
temp_mask = binary_fill_holes(temp_mask).T
mask = temp_mask[yLims[0]:yLims[1],xLims[0]:xLims[1]]
print '___',np.sum(mask),
im_mask = np.dstack([mask,mask*0,mask*0,mask*.2])
#plt.imshow(np.mean(areaF[:,yLims[0]:yLims[1],xLims[0]:xLims[1]],axis=0),cmap='binary_r')
#plt.imshow(im_mask)
#plt.show()
temp = areaF[:,yLims[0]:yLims[1],xLims[0]:xLims[1]] *np.abs(mask-1)
temp = temp.astype('float64')
temp[temp==0] = np.nan
neuropil_trace = np.nanmean(temp,axis=(1,2))
temp = areaF[:,yLims[0]:yLims[1],xLims[0]:xLims[1]] *mask
temp = temp.astype('float64')
temp[temp==0] = np.nan
trace = np.nanmean(temp,axis=(1,2))
corrected_trace = trace - .4*neuropil_trace
roiattrs['traces'][idx] = trace
roiattrs['neuropil_traces'][idx] = neuropil_trace
roiattrs['corr_traces'][idx] = corrected_trace
return roiattrs
def gen_tmp_extent(f_wi,f_temp): ref_wi = GR.geo_raster.open(f_wi) bnd_wi = ref_wi.get_band() m_wi = bnd_wi.read() m_out = np.zeros_like(m_wi) m_out[m_wi > -9999] = 1 m_fill = binary_fill_holes(m_out) m_out = lib_amerl_c.remove_small_objects(m_fill.astype(np.int16),100,0,4) GR.write_raster(f_temp, ref_wi.geo_transform, ref_wi.projection, m_out.astype(np.int8),1) return ref_wi.geo_transform,ref_wi.projection
def _is_converged(result, result_previous):
"""Check convergence.
Criterion: exclusion masks unchanged in subsequent iterations.
"""
mask = result["exclusion"].data == result_previous["exclusion"].data
# Because of pixel to pixel noise, the masks can still differ.
# This is handled by removing structures of the scale of one pixel
mask = binary_fill_holes(mask)
return np.all(mask)
def nuclei_detection(img, MinPixel, MaxPixel):
img_f = ski.img_as_float(img)
adjustRed = rescale_intensity(img_f[:,:,0])
roiGamma = rescale_intensity(adjustRed, in_range=(0, 0.5));
roiMaskThresh = roiGamma < (250 / 255.0) ;
roiMaskFill = morphology.remove_small_objects(~roiMaskThresh, MinPixel);
roiMaskNoiseRem = morphology.remove_small_objects(~roiMaskFill,150);
roiMaskDilat = morphology.dilation(roiMaskNoiseRem, morphology.disk(3));
roiMask = smorphology.binary_fill_holes(roiMaskDilat)
hsv = ski.color.rgb2hsv(img);
hsv[:,:,2] = 0.8;
img2 = ski.color.hsv2rgb(hsv)
diffRGB = img2-img_f
adjRGB = np.zeros(diffRGB.shape)
adjRGB[:,:,0] = rescale_intensity(diffRGB[:,:,0],in_range=(0, 0.4))
adjRGB[:,:,1] = rescale_intensity(diffRGB[:,:,1],in_range=(0, 0.4))
adjRGB[:,:,2] = rescale_intensity(diffRGB[:,:,2],in_range=(0, 0.4))
gauss = gaussian_filter(adjRGB[:,:,2], sigma=3, truncate=5.0);
bw1 = gauss>(100/255.0);
bw1 = bw1 * roiMask;
bw1_bwareaopen = morphology.remove_small_objects(bw1, MinPixel)
bw2 = smorphology.binary_fill_holes(bw1_bwareaopen);
bwDist = nd.distance_transform_edt(bw2);
filtDist = gaussian_filter(bwDist,sigma=5, truncate=5.0);
L = label(bw2)
R = regionprops(L)
coutn = 0
for idx, R_i in enumerate(R):
if R_i.area < MaxPixel and R_i.area > MinPixel:
r, l = R_i.centroid
#print(idx, filtDist[r,l])
else:
L[L==(idx+1)] = 0
BW = L > 0
return BW
def _is_converged(self, result, result_previous):
"""
Check convercence by comparing the exclusion masks between two
subsequent iterations.
"""
from scipy.ndimage.morphology import binary_fill_holes
mask = result['exclusion'].data == result_previous['exclusion'].data
# Because of pixel to pixel noise, the masks can still differ.
# This is handled by removing structures of the scale of one pixel
mask = binary_fill_holes(mask)
return np.all(mask)
def _init_data(self, img, mask):
# get image data
data = img.get_data()
self.affine = img.get_affine()
self.shape = data.shape
# masking -- by default, mask out zero intensities
if mask == None:
mask = binary_fill_holes(data > 0)
self.data = data[mask]
X, Y, Z = np.where(mask)
XYZ = np.zeros((X.shape[0], 3), dtype='intp')
XYZ[:, 0], XYZ[:, 1], XYZ[:, 2] = X, Y, Z
self.mask = mask
self.XYZ = XYZ
