def run(f):
patient_id = os.path.basename(f)[:-len("_seg.nii.gz")]
# if patient_id != "1585":
# return
print "PATIENT_ID",patient_id
f_img = img_folder + "/" + patient_id + ".nii"
if not os.path.exists(f_img):
f_img += ".gz"
seg = irtk.imread( f, dtype='float32', force_neurological=True )
img = irtk.imread( f_img, dtype='float32', force_neurological=True )
img = irtk.Image(nd.median_filter(img.view(np.ndarray),(3,5,5)),img.get_header())
ga = all_ga[patient_id]
scale = get_CRL(ga)/get_CRL(30.0)
# if all_iugr[patient_id][0] == 1:
# scale = (get_weight(ga,0.02) / get_weight(30,0.5)) ** (1.0/3.0)
# else:
# scale = (get_weight(ga,0.5) / get_weight(30,0.5)) ** (1.0/3.0)
seg = seg.resample( 1.0*scale, interpolation='nearest')
img = img.resample( 1.0*scale, interpolation='bspline' )
irtk.imwrite(output_folder + "/data_resampled_weight/"+patient_id+"_img.nii.gz",img)
irtk.imwrite(output_folder + "/data_resampled_weight/"+patient_id+"_seg.nii.gz",seg)
return
def get_training_data(file_img, file_mask, r):
# create mask
input_mask = irtk.imread(file_mask)
x_min, y_min, z_min, x_max, y_max, z_max = (input_mask == 0).bbox()
background = irtk.zeros(input_mask.get_header(), dtype='uint8')
background[z_min:z_max + 1, y_min:y_max + 1, x_min:x_max + 1] = 1
background = nd.morphological_gradient(background, size=7)
n = background[z_min + 1:z_max, y_min + 1:y_max, x_min + 1:x_max].sum()
z = np.random.randint(low=0, high=input_mask.shape[0], size=1.25 * n)
y = np.random.randint(low=0, high=input_mask.shape[1], size=1.25 * n)
x = np.random.randint(low=0, high=input_mask.shape[2], size=1.25 * n)
background[z, y, x] = 1
background[z_min + 1:z_max, y_min + 1:y_max, x_min + 1:x_max] = 0
foreground = (input_mask == 1).astype('uint8')
new_mask = irtk.zeros(input_mask.get_header(), dtype='uint8')
new_mask[foreground == 1] = 1
new_mask[background != 0] = 2
img = irtk.imread(file_img, dtype='float32')
X = []
Y = []
for z in xrange(img.shape[0]):
YX = np.transpose(np.nonzero(foreground[z]))
if DEBUG:
YX = YX[::10]
else:
YX = YX[::2]
if YX.shape[0] == 0:
continue
patches = extract_patches2D(img[z], r, YX)
patches = np.reshape(
patches, (patches.shape[0], patches.shape[1] * patches.shape[2]))
print patches.shape, YX.shape
X.extend(patches)
Y.extend([1] * len(YX))
for z in xrange(img.shape[0]):
YX = np.transpose(np.nonzero(background[z]))
if DEBUG:
YX = YX[::10]
else:
YX = YX[::2]
if YX.shape[0] == 0:
continue
patches = extract_patches2D(img[z], r, YX)
patches = np.reshape(
patches, (patches.shape[0], patches.shape[1] * patches.shape[2]))
print patches.shape, YX.shape
X.extend(patches)
Y.extend([0] * len(YX))
return X, Y
def show_offline_preprocessing( self, folder ):
if not os.path.exists(folder):
os.makedirs(folder)
all_files = glob( "offline_preprocessing/*_img.nii.gz" )
for f in all_files:
print f
name = os.path.basename(f)[:-len("_img.nii.gz")]
img = irtk.imread(f,dtype='float32')
seg = irtk.imread("offline_preprocessing/"+name+"_seg.nii.gz",dtype='uint8')
irtk.imshow(img,seg,filename=folder+"/"+name+".png",opacity=0.4)
def create_mask_from_all_masks(f_lists,transformations,ga,resolution=0.75):
points = []
for f, t in zip(f_lists,transformations):
m = irtk.imread(f,force_neurological=True)
points.extend( t.apply(get_corners(m)) )
points = np.array(points,dtype='float64')
x_min, y_min, z_min = points.min(axis=0)
x_max, y_max, z_max = points.max(axis=0)
pixelSize = [resolution, resolution, resolution, 1]
orientation = np.eye( 3, dtype='float64' )
origin = [ x_min + (x_max - x_min)/2,
y_min + (y_max - y_min)/2,
z_min + (z_max - z_min)/2,
0 ]
dim = [ (x_max - x_min)/resolution,
(y_max - y_min)/resolution,
(z_max - z_min)/resolution,
1 ]
header = irtk.new_header( pixelSize=pixelSize,
orientation=orientation,
origin=origin,
dim=dim )
mask = irtk.zeros( header, dtype='float32' )
for f, t in zip( f_lists, transformations ):
m = irtk.imread(f,force_neurological=True).transform(t, target=mask,interpolation="linear")
mask += m
irtk.imwrite( "debug_mask1.nii.gz", mask)
mask = irtk.Image( nd.gaussian_filter( mask, 0.5 ),
mask.get_header() )
irtk.imwrite( "debug_mask2.nii.gz", mask)
mask = (mask > 0).bbox(crop=True).astype('uint8')
scale = get_CRL(ga)/get_CRL(30.0)
template = irtk.imread(f_template,force_neurological=True)
template.header['pixelSize'][:3] /= scale
template = template.transform(target=mask,interpolation='nearest')
mask[template==0] = 0
irtk.imwrite( "debug_template.nii.gz", template)
irtk.imwrite( "debug_mask3.nii.gz", mask)
return mask
def crop_data(f,mask,t):
file_id = os.path.basename(f).split('.')[0]
img = irtk.imread( f, dtype='float32',force_neurological=True )
seg = mask.transform(t.invert(), target=img,interpolation='nearest')
x_min,y_min,z_min,x_max,y_max,z_max = seg.bbox()
seg = seg[z_min:z_max+1,
y_min:y_max+1,
x_min:x_max+1]
img = img[z_min:z_max+1,
y_min:y_max+1,
x_min:x_max+1].rescale(0,1000) + 1.0 # +1 to avoid zeros in the heart
img_file = output_dir + '/img_' + file_id + ".nii.gz"
irtk.imwrite( img_file, img )
for z in range(img.shape[0]):
scale = img[z].max()
img[z] = restoration.nl_means_denoising(img[z].rescale(0.0,1.0).view(np.ndarray),
fast_mode=False,
patch_size=5,
patch_distance=7,
h=0.05,
multichannel=False)
img[z] *= scale
img[seg==0] = 0
masked_file = output_dir + '/masked_' + file_id + ".nii.gz"
irtk.imwrite( masked_file, img )
def predict( self,
filename,
nb_autocontext=None,
mask=None,
debug=False,
return_all=False ):
"""
The prediction function must be defined outside of the main class in order to be used in joblib's Parallel.
"""
nb_labels = len(self.params['labels'])+1
if nb_autocontext is None:
nb_autocontext = len(glob(self.params['name'] + "_*"))
if self.params['predict_preprocessing_function'] is None:
img = irtk.imread( filename, dtype="float32" )
img = img.resample( pixelSize=self.params['resample'], interpolation='linear' ).rescale(0,1000)
else:
img = self.params['predict_preprocessing_function'](self, filename).copy()
if mask is None:
mask = irtk.ones( img.get_header(), dtype="uint8" ).as3D()
mask[img==0] = 0
else:
mask = mask.resample( pixelSize=self.params['resample'], interpolation='nearest' ).astype('uint8')
probas = predict_autocontext( self,
img,
mask,
nb_labels,
nb_autocontext,
debug=debug,
return_all=return_all )
return probas
def show(all_files,prefix="",saturate=False):
for f in all_files:
name = os.path.basename(f)[:-len('.nii.gz')]
img = irtk.imread(f,dtype="float32")
if saturate:
img = img.saturate().rescale()
png_name = "img/"+prefix+name+".png"
print png_name
irtk.imshow(img,filename=png_name)
def align_to_template(f,f_template,output_folder,ga):
file_id = f.split('/')[-3]
landmarks = irtk.imread(f,force_neurological=True)
scale = get_CRL(ga)/get_CRL(30.0)
template = irtk.imread(f_template,force_neurological=True)
template.header['pixelSize'][:3] /= scale
points = []
points_template = []
for i,j in zip( [2,8,3,4,5],
[5,4,1,2,3] ):
points_template.append( template.ImageToWorld( nd.center_of_mass(template.view(np.ndarray)==i)[::-1] ) )
points.append( landmarks.ImageToWorld( nd.center_of_mass(landmarks.view(np.ndarray)==j)[::-1] ) )
t,rms = irtk.registration_rigid_points( np.array(points),
np.array(points_template),
rms=True )
print "RMS: ", rms
t.invert().write( output_folder + '/' + file_id + '.dof' )
landmarks = landmarks.transform(t,target=template)
irtk.imwrite( output_folder + '/landmarks_' + file_id + '.nii.gz',landmarks )
return t
def run(f):
patient_id = os.path.basename(f)[:-len("_seg.nii.gz")]
print "PATIENT_ID",patient_id
f_img = data_folder + "/" + patient_id + ".nii"
if not os.path.exists(f_img):
f_img += ".gz"
seg = irtk.imread( f, dtype='float32', force_neurological=True )
img = irtk.imread( f_img, dtype='float32', force_neurological=True )
img = irtk.Image(nd.median_filter(img.view(np.ndarray),(3,5,5)),img.get_header())
ga = all_ga[patient_id]
scale = get_CRL(ga)/get_CRL(30.0)
OFD = get_OFD(30.0)
BPD = get_BPD(30.0)
CRL = get_CRL(30.0)
brain_center = seg.ImageToWorld( np.array(nd.center_of_mass( (seg == 2).view(np.ndarray) ),
dtype='float32')[::-1] )
header = img.get_header()
header['origin'][:3] = brain_center
header['pixelSize'][:3] = 1.0*scale
header['dim'][0] = CRL
header['dim'][1] = CRL
header['dim'][2] = CRL
img = img.transform( target=header, interpolation="bspline" )
seg = seg.transform( target=header, interpolation="nearest" )
img[img<1.0] = 0
irtk.imwrite(output_folder + "brain_center/"+patient_id+"_img.nii.gz",img)
irtk.imwrite(output_folder + "brain_center/"+patient_id+"_seg.nii.gz",seg)
return
def get_center_brain_detection(f,world=True):
input_mask = irtk.imread( f, force_neurological=False )
mask = irtk.ones( input_mask.get_header(), dtype='uint8' )
mask[input_mask == 2] = 0
x_min, y_min, z_min, x_max, y_max, z_max = map( float, mask.bbox() )
center = [ x_min + (x_max-x_min)/2,
y_min + (y_max-y_min)/2,
z_min + (z_max-z_min)/2 ]
if not world:
center = np.array(center,dtype='float64')[::-1]
return center
else:
center = input_mask.ImageToWorld(center)
return center
def score( self,
validation_patients,
nb_autocontext=None ):
gc.collect()
filenames = []
for patient_id in validation_patients:
img_filename = self.params['img_folder'] + "/" + patient_id + self.params['file_extension']
filenames.append(img_filename)
# probas = Parallel(n_jobs=self.params['n_jobs'])(delayed(predict_level)( self,
# img_filename,
# all_ga[patient_id],
# level=level,
# nb_autocontext=nb_autocontext )
# for patient_id,img_filename in zip(validation_patients,filenames) )
probas = []
for patient_id,img_filename in zip(validation_patients,filenames):
print img_filename
probas.append( predict( self,
img_filename,
nb_autocontext=nb_autocontext ) )
print "will compute Dice scores"
score = 0.0
n = 0
for patient_id,proba in zip(validation_patients,probas):
header = proba.get_header()
header['dim'][3] = 1
if self.params['offline_preprocessing']:
seg_filename = "offline_preprocessing/"+patient_id+"_seg"+ self.params['file_extension']
else:
seg_filename = self.params['seg_folder'] + "/" +patient_id+"_seg"+ self.params['file_extension']
seg = irtk.imread( seg_filename, dtype="uint8" )
#seg = seg.resample( self.params['resample'], interpolation="nearest").astype('uint8')
# irtk.imwrite( "debug/"+patient_id+"_proba"+str(nb_autocontext)+".nii.gz",
# proba )
# we skip mother/background as it depends of mask
for i in [1]:#xrange(1,proba.shape[0]):
dice,overlap = (seg==i).dice( self.hard_thresholding( proba[i] ),
verbose=False)
score += dice
n += 1
return score/n
def process_file( f, ga, step=1, DEBUG=False ):
print f
img = irtk.imread( f, dtype='float32', force_neurological=False )
## Resample
img = resampleOFD( img, ga )
## Contrast-stretch with saturation
img = img.saturate(1,99).rescale().astype('uint8')
detector = cv2.SIFT( nfeatures=0,
nOctaveLayers=3,
contrastThreshold=0.04,
edgeThreshold=10,
sigma=0.8)
descriptorExtractor = cv2.DescriptorExtractor_create("SIFT")
points = []
for z in range(0,img.shape[0],step):
keypoints = detector.detect(img[z])
if keypoints is None or len(keypoints) == 0:
continue
(keypoints, descriptors) = descriptorExtractor.compute(img[z],keypoints)
unique_index= np.unique( descriptors.dot(np.random.rand(128)),
return_index=True)[1]
points.extend(descriptors[unique_index])
## For debugging purpose:
if DEBUG:
img_color = cv2.cvtColor( img[z].astype('uint8'), cv2.cv.CV_GRAY2RGB )
for y,x in F.transpose():
cv2.circle( img_color,
(int(x),int(y)),
2,
(0,0,255),
-1)
cv2.imwrite( "/tmp/"
+ os.path.basename(f.rstrip('.nii'))
+ "_" + str(z)
+".png", img_color )
points = np.array(points)
unique_index= np.unique( points.dot(np.random.rand(128)),
return_index=True)[1]
return points[unique_index]
def mask_data(f):
file_id = f.split('/')[-3]
seg = irtk.imread(f,force_neurological=True) > 0
r = 10
x_min,y_min,z_min,x_max,y_max,z_max = seg.bbox()
seg = seg[max(0,z_min-3*r):min(z_max+3*r+1,seg.shape[0]),
max(0,y_min-3*r):min(y_max+3*r+1,seg.shape[1]),
max(0,x_min-3*r):min(x_max+3*r+1,seg.shape[2])]
ball = morphology.ball( 5 )
seg = irtk.Image( nd.binary_dilation(seg,ball), seg.get_header() )
ball = morphology.ball( r )
seg = irtk.Image( nd.binary_closing(seg,ball), seg.get_header() )
seg = seg.bbox(crop=True)
seg_file = output_dir + '/seg_' + file_id + ".nii.gz"
irtk.imwrite( seg_file, seg )
def predict(self,
filename,
ga,
nb_autocontext=None,
mask=None,
debug=False,
return_all=False):
nb_labels = len(self.labels) + 1
if nb_autocontext is None:
nb_autocontext = len(glob(self.params['name'] + "_*"))
img = irtk.imread(filename, dtype="float32")
img = img.resample(pixelSize=self.params['resample'],
interpolation='linear').rescale(0, 1000)
extra_layers = []
if mask is None:
mask = irtk.ones(img.get_header(), dtype="uint8")
mask[img == 0] = 0
else:
mask = mask.resample(pixelSize=self.params['resample'],
interpolation='nearest').astype('uint8')
metadata = None
probas = predict_autocontext(self,
img,
mask,
np.array(extra_layers, dtype="float32"),
metadata,
nb_labels,
ga,
nb_autocontext,
debug=debug,
return_all=return_all)
return probas
def predict( self,
filename,
ga,
nb_autocontext=None,
mask=None,
debug=False,
return_all=False ):
nb_labels = len(self.labels)+1
if nb_autocontext is None:
nb_autocontext = len(glob(self.params['name'] + "_*"))
img = irtk.imread( filename, dtype="float32" )
img = img.resample( pixelSize=self.params['resample'], interpolation='linear' ).rescale(0,1000)
extra_layers = []
if mask is None:
mask = irtk.ones( img.get_header(), dtype="uint8" )
mask[img==0] = 0
else:
mask = mask.resample( pixelSize=self.params['resample'], interpolation='nearest' ).astype('uint8')
metadata = None
probas = predict_autocontext( self,
img,
mask,
np.array( extra_layers, dtype="float32" ),
metadata,
nb_labels,
ga,
nb_autocontext,
debug=debug,
return_all=return_all )
return probas
def get_coordinates( f ):
seg = irtk.imread( f )
u,v,w = get_orientation_training(seg)
M = np.array( [u,v,w], dtype='float32' ) # Change of basis matrix
heart = np.array(nd.center_of_mass( (seg == 5).view(np.ndarray) ),
dtype='float32')
brain = np.array(nd.center_of_mass( (seg == 2).view(np.ndarray) ),
dtype='float32')
left_lung = np.array(nd.center_of_mass( (seg == 3).view(np.ndarray) ),
dtype='float32')
right_lung = np.array(nd.center_of_mass( (seg == 4).view(np.ndarray) ),
dtype='float32')
liver = np.array(nd.center_of_mass( (seg == 8).view(np.ndarray) ),
dtype='float32')
# centering and orient
brain = np.dot( M, brain - heart)
left_lung = np.dot( M, left_lung - heart)
right_lung = np.dot( M, right_lung - heart)
liver = np.dot( M, liver - heart)
return np.array( [brain, left_lung, right_lung, liver], dtype='float32' ).flatten()
def detect_mser(
raw_file,
ga,
vocabulary,
mser_detector,
NEW_SAMPLING,
output_folder="debug",
DEBUG=False,
return_image_regions=False,
):
OFD = get_OFD(ga) / NEW_SAMPLING
BPD = get_BPD(ga) / NEW_SAMPLING
max_e = 0.64
mser = cv2.MSER(
_delta=5,
_min_area=60,
_max_area=14400,
_max_variation=0.15,
_min_diversity=0.1,
_max_evolution=200,
_area_threshold=1.01,
_min_margin=0.003,
_edge_blur_size=5,
)
sift = cv2.SIFT(nfeatures=0, nOctaveLayers=3, contrastThreshold=0.04, edgeThreshold=10, sigma=0.8)
siftExtractor = cv2.DescriptorExtractor_create("SIFT")
voca = np.load(open(vocabulary, "rb"))
classifier = neighbors.NearestNeighbors(1, algorithm="kd_tree")
N = voca.shape[0]
classifier.fit(voca)
# flann = pyflann.FLANN()
# flann.build_index( voca.astype('float32') )
svc = joblib.load(mser_detector)
img = irtk.imread(raw_file, dtype="float32")
img = img.resample2D(NEW_SAMPLING).saturate().rescale().astype("uint8")
detections = []
image_regions = []
for z in range(img.shape[0]):
detections.append([])
image_regions.append([])
# Extract MSER
# print "extracting mser"
contours = mser.detect(img[z, :, :])
# print "mser done"
if DEBUG:
img_color = cv2.cvtColor(img[z, :, :], cv2.cv.CV_GRAY2RGB)
for c in contours:
ellipse = cv2.fitEllipse(np.array(map(lambda x: [x], c), dtype="int32"))
cv2.ellipse(img_color, (ellipse[0], (ellipse[1][0], ellipse[1][1]), ellipse[2]), (0, 0, 255))
cv2.imwrite(output_folder + "/" + str(z) + "_all_mser_.png", img_color)
img_color_mser = cv2.cvtColor(img[z, :, :], cv2.cv.CV_GRAY2RGB)
# Filter MSER
selected_mser = []
mask = np.zeros((img.shape[1], img.shape[2]), dtype="uint8")
# print "fitting ellipses"
for c in contours:
ellipse = cv2.fitEllipse(np.reshape(c, (c.shape[0], 1, 2)).astype("int32"))
# filter by size
if ellipse[1][0] > OFD or ellipse[1][1] > OFD or ellipse[1][0] < 0.5 * OFD or ellipse[1][1] < 0.5 * OFD:
continue
# filter by eccentricity
if math.sqrt(1 - (np.min(ellipse[1]) / np.max(ellipse[1])) ** 2) > max_e:
continue
cv2.ellipse(mask, ellipse, 255, -1)
selected_mser.append((c, ellipse))
# print "ellipses done"
if len(selected_mser) == 0:
continue
# Extract SIFT
# print "extracting SIFT"
keypoints = sift.detect(img[z, :, :], mask=mask)
# print "SIFT done"
if keypoints is None or len(keypoints) == 0:
continue
(keypoints, descriptors) = siftExtractor.compute(img[z, :, :], keypoints)
# words = np.zeros(len(keypoints),dtype="int")
# for i,d in enumerate(descriptors):
# words[i] = classifier.kneighbors(d, return_distance=False)
words = classifier.kneighbors(descriptors, return_distance=False)
# words, dist = flann.nn_index( descriptors.astype('float32') )
for i, (c, ellipse) in enumerate(selected_mser):
# Compute histogram
hist = np.zeros(N, dtype="float")
for ki, k in enumerate(keypoints):
if is_in_ellipse(k.pt, ellipse):
hist[words[ki]] += 1
# Normalize histogram
norm = np.linalg.norm(hist)
if norm > 0:
hist /= norm
cl = svc.predict(hist).flatten()
if DEBUG:
if cl == 1:
opacity = 0.4
img_color = cv2.cvtColor(img[z, :, :], cv2.cv.CV_GRAY2RGB)
for p in c:
img_color[p[1], p[0], :] = (1 - opacity) * img_color[p[1], p[0], :] + opacity * np.array(
[0, 255, 0]
)
cv2.imwrite(output_folder + "/" + str(z) + "_" + str(i) + "_region.png", img_color)
img_color = cv2.cvtColor(img[z, :, :], cv2.cv.CV_GRAY2RGB)
cv2.ellipse(img_color, (ellipse[0], (ellipse[1][0], ellipse[1][1]), ellipse[2]), (0, 0, 255))
for k_id, k in enumerate(keypoints):
if is_in_ellipse(k.pt, ellipse):
if cl == 1:
cv2.circle(img_color, (int(k.pt[0]), int(k.pt[1])), 2, (0, 255, 0), -1)
else:
cv2.circle(img_color, (int(k.pt[0]), int(k.pt[1])), 2, (0, 0, 255), -1)
cv2.imwrite(output_folder + "/" + str(z) + "_" + str(i) + ".png", img_color)
cv2.ellipse(
img_color_mser,
(ellipse[0], (ellipse[1][0], ellipse[1][1]), ellipse[2]),
(0, 255, 0) if cl == 1 else (0, 0, 255),
)
if cl == 1:
ellipse_center = [ellipse[0][0], ellipse[0][1], z]
print c.shape
if return_image_regions:
image_regions[-1].append((ellipse_center, c))
else:
region = np.hstack((c, [[z]] * c.shape[0]))
detections[-1].append((img.ImageToWorld(ellipse_center), img.ImageToWorld(region)))
if DEBUG:
cv2.imwrite(output_folder + "/" + str(z) + "_color_mser.png", img_color_mser)
if return_image_regions:
return image_regions
else:
return detections
#!/usr/bin/python
import irtk
import cv2
mask = irtk.imread("mask.nii", dtype='uint8')
irtk.imwrite("mask.png", mask)
img = irtk.Image(cv2.imread("lena.png", 0))
irtk.imshow(img,
mask,
filename="initialisation.png",
colors={
1: (255, 0, 0),
2: (0, 255, 0)
},
opacity=1.0)
mask2 = irtk.imread("mask2.nii", dtype='uint8')
irtk.imwrite("mask2.png", mask2)
irtk.imshow(img,
mask2,
filename="initialisation2.png",
colors={
1: (255, 0, 0),
2: (0, 255, 0)
},
opacity=1.0)
def getIRTKDtls(self, fileName):
header, dtype = irtk._irtk.get_header(fileName)
img = irtk.imread(fileName, dtype='float32')
irtkDtls = []
irtkDtls.append(
("Image Size",
str(header['dim'][0]) + "\t" + str(header['dim'][1]) + "\t" +
str(header['dim'][2]) + "\t" + str(header['dim'][3]) + "\t"))
irtkDtls.append(
("Voxel Size (mm)", format(header['pixelSize'][0], '.3f') + "\t" +
format(header['pixelSize'][1], '.3f') + "\t" +
format(header['pixelSize'][2], '.3f') + "\t" +
format(header['pixelSize'][3], '.3f') + "\t"))
irtkDtls.append(("Origin", format(header['origin'][0], '.3f') + "\t" +
format(header['origin'][1], '.3f') + "\t" +
format(header['origin'][2], '.3f') + "\t" +
format(header['origin'][3], '.3f') + "\t"))
irtkDtls.append(("X Axis", format(header['orientation'][0, 0], '.3f') +
"\t" + format(header['orientation'][0, 1], '.3f') +
"\t" + format(header['orientation'][0, 2], '.3f')))
irtkDtls.append(("Y Axis", format(header['orientation'][1, 0], '.3f') +
"\t" + format(header['orientation'][1, 1], '.3f') +
"\t" + format(header['orientation'][1, 2], '.3f')))
irtkDtls.append(("Z Axis", format(header['orientation'][2, 0], '.3f') +
"\t" + format(header['orientation'][2, 1], '.3f') +
"\t" + format(header['orientation'][2, 2], '.3f')))
irtkDtls.append(("Ordering", img.order()))
orientation = img.orientation()
irtkDtls.append(
("Orientation",
orientation[0] + "\t" + orientation[1] + "\t" + orientation[2]))
irtkDtls.append(("Data Type", dtype))
irtkDtls.append(("Min-Max", format(float(img.min()), '.3f') + "\t" +
format(float(img.max()), '.3f')))
irtkDtls.append(("Image to World", format(img.I2W[0, 0], '.3f') +
"\t" + format(img.I2W[0, 1], '.3f') + "\t" +
format(img.I2W[0, 2], '.3f') + "\t" +
format(img.I2W[0, 3], '.3f') + "\n" +
format(img.I2W[1, 0], '.3f') + "\t" +
format(img.I2W[1, 1], '.3f') + "\t" +
format(img.I2W[1, 2], '.3f') + "\t" +
format(img.I2W[1, 3], '.3f') + "\n" +
format(img.I2W[2, 0], '.3f') + "\t" +
format(img.I2W[2, 1], '.3f') + "\t" +
format(img.I2W[2, 2], '.3f') + "\t" +
format(img.I2W[2, 3], '.3f') + "\n" +
format(img.I2W[3, 0], '.3f') + "\t" +
format(img.I2W[3, 1], '.3f') + "\t" +
format(img.I2W[3, 2], '.3f') + "\t" +
format(img.I2W[3, 3], '.3f')))
irtkDtls.append(("World to Image", format(img.W2I[0, 0], '.3f') +
"\t" + format(img.W2I[0, 1], '.3f') + "\t" +
format(img.W2I[0, 2], '.3f') + "\t" +
format(img.W2I[0, 3], '.3f') + "\n" +
format(img.W2I[1, 0], '.3f') + "\t" +
format(img.W2I[1, 1], '.3f') + "\t" +
format(img.W2I[1, 2], '.3f') + "\t" +
format(img.W2I[1, 3], '.3f') + "\n" +
format(img.W2I[2, 0], '.3f') + "\t" +
format(img.W2I[2, 1], '.3f') + "\t" +
format(img.W2I[2, 2], '.3f') + "\t" +
format(img.W2I[2, 3], '.3f') + "\n" +
format(img.W2I[3, 0], '.3f') + "\t" +
format(img.W2I[3, 1], '.3f') + "\t" +
format(img.W2I[3, 2], '.3f') + "\t" +
format(img.W2I[3, 3], '.3f')))
return irtkDtls
return noise.std()
def get_noiseZ(img):
img = img.astype('float32')
new_img = np.zeros(img.shape, dtype='float32')
for x in xrange(img.shape[2]):
new_img[:, :, x] = nd.gaussian_filter(img[:, :, x], 2, mode='reflect')
noise = img - new_img
#print "Noise Z:", noise.std(), img.std()
return noise.std()
output_filename = sys.argv[3]
img = irtk.imread(sys.argv[1], dtype='float64').saturate()
mask = irtk.imread(sys.argv[2], dtype='int16')
mask = irtk.Image(mask, img.get_header())
# crop
x_min, y_min, z_min, x_max, y_max, z_max = mask.bbox()
mask = mask[z_min:z_max + 1, y_min:y_max + 1, x_min:x_max + 1]
tmp_img = img[z_min:z_max + 1, y_min:y_max + 1, x_min:x_max + 1]
downsampled_img = tmp_img.resample(2)
mask = mask.transform(target=downsampled_img, interpolation='nearest')
seg = irtk.graphcut(downsampled_img,
mask,
sigma=get_noiseXY(downsampled_img),
sigmaZ=get_noiseZ(downsampled_img))
def detect_mser( f,
ga,
vocabulary,
mser_detector,
output_folder="debug",
DEBUG=False ):
max_e = 0.64
OFD = get_OFD(30.0,centile=50)
BPD = get_BPD(30.0,centile=50)
mser = cv2.MSER( _delta=5,
_min_area=60,
_max_area=14400,
_max_variation=0.15,
_min_diversity=.1,
_max_evolution=200,
_area_threshold=1.01,
_min_margin=0.003,
_edge_blur_size=5)
sift = cv2.SIFT( nfeatures=0,
nOctaveLayers=3,
contrastThreshold=0.04,
edgeThreshold=10,
sigma=0.8)
siftExtractor = cv2.DescriptorExtractor_create("SIFT")
voca = np.load( open(vocabulary, 'rb') )
nn_classifier = neighbors.NearestNeighbors(1,algorithm='kd_tree')
N = voca.shape[0]
nn_classifier.fit(voca)
classifier = joblib.load(mser_detector)
classifier.classes_ = np.array([0,1],dtype='int')
img = irtk.imread( f,
dtype='float32',
force_neurological=False ).saturate(1,99).rescale()
img = resampleOFD( img, ga ).astype('uint8')
detected_centers = []
detected_regions = []
for z in range(img.shape[0]):
# Extract MSER
#print "extracting mser"
contours = mser.detect(img[z])
#print "mser done"
if DEBUG:
img_color = cv2.cvtColor( img[z], cv2.cv.CV_GRAY2RGB )
for c in contours:
ellipse = cv2.fitEllipse(np.array(map(lambda x:[x],
c),dtype='int32'))
cv2.ellipse( img_color, (ellipse[0],
(ellipse[1][0],ellipse[1][1]),
ellipse[2]) , (0,0,255))
cv2.imwrite(output_folder + "/" +str(z) + "_all_mser_.png",img_color )
img_color_mser = cv2.cvtColor( img[z], cv2.cv.CV_GRAY2RGB )
# Filter MSER
selected_mser = []
mask = np.zeros( (img.shape[1],img.shape[2]), dtype='uint8' )
#print "fitting ellipses"
for c in contours:
ellipse = cv2.fitEllipse(np.reshape(c, (c.shape[0],1,2) ).astype('int32'))
# filter by size
if ( ellipse[1][0] > OFD
or ellipse[1][1] > OFD
or ellipse[1][0] < 0.5*OFD
or ellipse[1][1] < 0.5*OFD ) :
continue
# filter by eccentricity
if math.sqrt(1-(np.min(ellipse[1])/np.max(ellipse[1]))**2) > max_e:
continue
cv2.ellipse( mask, ellipse, 255, -1 )
selected_mser.append((c,ellipse))
#print "ellipses done"
if len(selected_mser) == 0:
continue
# Extract SIFT
#print "extracting SIFT"
keypoints = sift.detect(img[z],mask=mask)
#print "SIFT done"
if keypoints is None or len(keypoints) == 0:
continue
(keypoints, descriptors) = siftExtractor.compute(img[z],keypoints)
words = nn_classifier.kneighbors(descriptors, return_distance=False)
for i,(c,ellipse) in enumerate(selected_mser):
# Compute histogram
hist = np.zeros(N, dtype='float')
for ki,k in enumerate(keypoints):
if is_in_ellipse(k.pt,ellipse):
hist[words[ki]] += 1
# Normalize histogram
norm = np.linalg.norm(hist)
if norm > 0:
hist /= norm
cl = classifier.predict(hist)
if DEBUG:
if cl == 1:
opacity = 0.4
img_color = cv2.cvtColor( img[z], cv2.cv.CV_GRAY2RGB )
for p in c:
img_color[p[1],p[0],:] = (
(1-opacity)*img_color[p[1],p[0],:]
+ opacity * np.array([0,255,0])
)
cv2.imwrite(output_folder + "/"+str(z) + '_' +str(i) +"_region.png",img_color)
img_color = cv2.cvtColor( img[z], cv2.cv.CV_GRAY2RGB )
cv2.ellipse( img_color, (ellipse[0],
(ellipse[1][0],ellipse[1][1]),
ellipse[2]) , (0,0,255))
for k_id,k in enumerate(keypoints):
if is_in_ellipse(k.pt,ellipse):
if cl == 1:
cv2.circle( img_color,
(int(k.pt[0]),int(k.pt[1])),
2,
(0,255,0),
-1)
else:
cv2.circle( img_color,
(int(k.pt[0]),int(k.pt[1])),
2,
(0,0,255),
-1)
cv2.imwrite(output_folder + "/"+str(z) + '_' +str(i) +".png",img_color)
cv2.ellipse( img_color_mser, (ellipse[0],
(ellipse[1][0],ellipse[1][1]),
ellipse[2]),
(0,255,0) if cl == 1 else (0,0,255) )
if cl == 1:
ellipse_center = [z,ellipse[0][1],ellipse[0][0]]
detected_centers.append( ellipse_center )
detected_regions.append( c )
if DEBUG:
cv2.imwrite(output_folder + "/"+str(z) + "_color_mser.png",img_color_mser)
return np.array(detected_centers, dtype='int32'), np.array(detected_regions)
#!/usr/bin/python import irtk import sys import scipy.ndimage as nd input_file = sys.argv[1] output_file = sys.argv[2] img = irtk.imread(input_file, dtype='float32') img = irtk.Image(nd.median_filter(img.get_data(), 5), img.get_header()) irtk.imwrite(output_file, img)
import irtk
from glob import glob
import os
def get_ED_ES(patient_id):
f = "/vol/biomedic/users/kpk09/DATASETS/CETUS_data/Training/Images/"+patient_id+"/"+patient_id+"_ED_ES_time.txt"
f = open(f,"rb")
res = []
for line in f:
line = line.rstrip() # chomp
res.append( int(line.split(' ')[-1]))
return res
all_files = glob("nifti/*_seg.nii.gz")
m = 0.0
n = 0.0
for f in all_files:
patient_id = os.path.basename(f).split("_")[0]
frame_id = int(os.path.basename(f).split("_")[1][len("frame"):])
ED_ES = get_ED_ES(patient_id)
print patient_id, frame_id, ED_ES
if frame_id == ED_ES[1]:
mask = irtk.imread(f).resample(0.001, interpolation='nearest')
m += mask.sum()
n += 1
m /= n
print "mean heart volume:",m
import numpy as np
import irtk
def get_center_brain_detection(f,world=True):
input_mask = irtk.imread( f, force_neurological=False )
mask = irtk.ones( input_mask.get_header(), dtype='uint8' )
mask[input_mask == 2] = 0
x_min, y_min, z_min, x_max, y_max, z_max = map( float, mask.bbox() )
center = [ x_min + (x_max-x_min)/2,
y_min + (y_max-y_min)/2,
z_min + (z_max-z_min)/2 ]
if not world:
center = np.array(center,dtype='float64')[::-1]
return center
else:
center = input_mask.ImageToWorld(center)
return center
def get_box_center((z,y,x),(d,h,w),f):
img = irtk.imread( f, empty=True, force_neurological=False )
center = np.array( [ x+w/2,
y+h/2,
z+d/2 ], dtype='float' )
center = img.ImageToWorld(center)
return center
def mask_image(file_img, file_mask, ga, r, neigh, output_dir):
img = irtk.imread(file_img, dtype='float32')
input_mask = irtk.imread(file_mask)
print "predicting..."
res = irtk.zeros(img.get_header(), dtype='float32')
res2 = irtk.zeros(img.get_header(), dtype='float32')
res3 = irtk.zeros(img.get_header(), dtype='float32')
res4 = irtk.zeros(img.get_header(), dtype='uint8')
mask = irtk.ones(input_mask.get_header(), dtype='uint8')
mask[input_mask == 2] = 0
for z in xrange(img.shape[0]):
print z
YX = np.transpose(np.nonzero(mask[z]))
if YX.shape[0] == 0:
continue # this slice does not intersect the box
patches = extract_patches2D(img[z], r, YX)
patches = np.reshape(
patches, (patches.shape[0], patches.shape[1] * patches.shape[2]))
predictions = neigh.predict_proba(patches)[:, 1]
res[z, YX[:, 0], YX[:, 1]] = predictions
x_min, y_min, z_min, x_max, y_max, z_max = mask.bbox()
proba = res[z_min:z_max + 1, y_min:y_max + 1, x_min:x_max + 1]
if args.mass:
BV = get_BV(args.ga)
box_volume = (z_max - z_min) * img.header['pixelSize'][2] * (
y_max - y_min) * img.header['pixelSize'][1] * (
x_max - x_min) * img.header['pixelSize'][0]
ratio = float(BV) / float(box_volume)
print "ratio", ratio
q0, q1 = mquantiles(proba.flatten(),
prob=[0.5 * (1.0 - ratio), 1.0 - 0.5 * ratio])
print "threshold", q0, q1
#threshold = max(0.5,threshold)
# labels = res[z_min:z_max+1,
# y_min:y_max+1,
# x_min:x_max+1] > threshold
#res = 1 / (np.exp(-(res-threshold)/(res.max()-res.min())))
res[res < q0] = q0
res[res > q1] = q1
res -= res.min()
res /= res.max()
labels = res[z_min:z_max + 1, y_min:y_max + 1, x_min:x_max + 1] > 0.5
proba = res[z_min:z_max + 1, y_min:y_max + 1, x_min:x_max + 1]
cropped_img = img[z_min:z_max + 1, y_min:y_max + 1, x_min:x_max + 1]
if args.do_3D:
labels = irtk.crf(cropped_img,
labels,
proba,
l=args.l,
sigma=get_noiseXY(cropped_img),
sigmaZ=get_noiseZ(cropped_img))
# elif args.do_patchZ:
# labels = irtk.crf_patchZ( cropped_img,
# labels,
# proba,
# l=10.0 )
# else:
# for z in xrange(z_min,z_max+1):
# labels[z] = irtk.crf( cropped_img[z],
# labels[z],
# proba[z],
# l=1.0 )
print "MAX LABEL:", labels.max()
irtk.imwrite(output_dir + "/bare_" + os.path.basename(file_img), labels)
tmp = irtk.zeros(img.get_header(), dtype='uint8')
tmp[z_min:z_max + 1, y_min:y_max + 1, x_min:x_max + 1] = labels
(min_x_bare, min_y_bare, min_z_bare, max_x_bare, max_y_bare,
max_z_bare) = tmp.bbox()
if not args.no_cleaning:
# clean by fitting ellipses enlarged of 10%
for z in xrange(labels.shape[0]):
edges = nd.morphological_gradient(labels[z] > 0, size=5)
points = np.transpose(edges.nonzero())[:, ::-1]
if len(points) == 0:
continue
points = np.array(map(lambda x: [x], points), dtype='int32')
ellipse = cv2.fitEllipse(points)
cv2.ellipse(
labels[z],
(ellipse[0],
(1.1 * ellipse[1][0], 1.1 * ellipse[1][1]), ellipse[2]), 1,
-1)
irtk.imwrite(output_dir + "/seg_" + os.path.basename(file_img), labels)
irtk.imwrite(output_dir + "/res_" + os.path.basename(file_img), res)
# re-read the image in case we processed it
img = irtk.imread(file_img, dtype='float32')
cropped_img = img[z_min:z_max + 1, y_min:y_max + 1, x_min:x_max + 1]
cropped_img[labels == 0] = -1
masked = cropped_img.bbox(crop=True)
irtk.imwrite(output_dir + "/masked_" + os.path.basename(file_img), masked)
# re-read the image in case we processed it
img = irtk.imread(file_img, dtype='float32')
x0 = min_x_bare + (max_x_bare - min_x_bare) / 2
y0 = min_y_bare + (max_y_bare - min_y_bare) / 2
ofd = get_OFD(ga) / img.header['pixelSize'][0]
cropped_img = img[min_z_bare:max_z_bare + 1,
max(0, int(round(y0 - ofd / 2))
):min(img.shape[1], int(round(y0 + ofd / 2 + 1))),
max(0, int(round(x0 - ofd / 2))
):min(img.shape[2], int(round(x0 + ofd / 2 +
1)))].copy()
irtk.imwrite(output_dir + "/very_large_" + os.path.basename(file_img),
cropped_img)
cropped_proba = res[min_z_bare:max_z_bare + 1,
max(0, int(round(y0 - ofd / 2))
):min(img.shape[1], int(round(y0 + ofd / 2 + 1))),
max(0, int(round(x0 - ofd / 2))
):min(img.shape[2], int(round(x0 + ofd / 2 +
1)))].copy()
irtk.imwrite(output_dir + "/proba_" + os.path.basename(file_img),
cropped_proba)
def mask_image( file_img, file_mask, ga, r, neigh, output_dir ):
img = irtk.imread( file_img, dtype='float32' )
input_mask = irtk.imread( file_mask )
print "predicting..."
res = irtk.zeros( img.get_header(), dtype='float32' )
res2 = irtk.zeros( img.get_header(), dtype='float32' )
res3 = irtk.zeros( img.get_header(), dtype='float32' )
res4 = irtk.zeros( img.get_header(), dtype='uint8' )
mask = irtk.ones( input_mask.get_header(), dtype='uint8' )
mask[input_mask == 2] = 0
for z in xrange(img.shape[0]):
print z
YX = np.transpose( np.nonzero( mask[z] ) )
if YX.shape[0] == 0:
continue # this slice does not intersect the box
patches = extract_patches2D( img[z], r, YX )
patches = np.reshape( patches, (patches.shape[0],patches.shape[1]*patches.shape[2]) )
predictions = neigh.predict_proba(patches)[:,1]
res[z,YX[:,0],YX[:,1]] = predictions
x_min, y_min, z_min, x_max, y_max, z_max = mask.bbox()
proba = res[z_min:z_max+1,
y_min:y_max+1,
x_min:x_max+1]
if args.mass:
BV = get_BV( args.ga )
box_volume = (z_max-z_min)*img.header['pixelSize'][2]*(y_max-y_min)*img.header['pixelSize'][1]*(x_max-x_min)*img.header['pixelSize'][0]
ratio = float(BV) / float(box_volume)
print "ratio", ratio
q0,q1 = mquantiles( proba.flatten(), prob=[0.5*(1.0-ratio),
1.0-0.5*ratio] )
print "threshold", q0,q1
#threshold = max(0.5,threshold)
# labels = res[z_min:z_max+1,
# y_min:y_max+1,
# x_min:x_max+1] > threshold
#res = 1 / (np.exp(-(res-threshold)/(res.max()-res.min())))
res[res<q0] = q0
res[res>q1] = q1
res -= res.min()
res /= res.max()
labels = res[z_min:z_max+1,
y_min:y_max+1,
x_min:x_max+1] > 0.5
proba = res[z_min:z_max+1,
y_min:y_max+1,
x_min:x_max+1]
cropped_img = img[z_min:z_max+1,
y_min:y_max+1,
x_min:x_max+1]
if args.do_3D:
labels = irtk.crf( cropped_img,
labels,
proba,
l=args.l,
sigma=get_noiseXY(cropped_img),
sigmaZ=get_noiseZ(cropped_img) )
# elif args.do_patchZ:
# labels = irtk.crf_patchZ( cropped_img,
# labels,
# proba,
# l=10.0 )
# else:
# for z in xrange(z_min,z_max+1):
# labels[z] = irtk.crf( cropped_img[z],
# labels[z],
# proba[z],
# l=1.0 )
print "MAX LABEL:", labels.max()
irtk.imwrite(output_dir + "/bare_"+os.path.basename(file_img), labels )
tmp = irtk.zeros( img.get_header(), dtype='uint8' )
tmp[z_min:z_max+1,
y_min:y_max+1,
x_min:x_max+1] = labels
( min_x_bare, min_y_bare, min_z_bare,
max_x_bare, max_y_bare, max_z_bare ) = tmp.bbox()
if not args.no_cleaning:
# clean by fitting ellipses enlarged of 10%
for z in xrange(labels.shape[0]):
edges = nd.morphological_gradient( labels[z] > 0,size=5 )
points = np.transpose(edges.nonzero())[:,::-1]
if len(points) == 0:
continue
points = np.array(map(lambda x:[x],points),dtype='int32')
ellipse = cv2.fitEllipse(points)
cv2.ellipse( labels[z], (ellipse[0],
(1.1*ellipse[1][0],1.1*ellipse[1][1]),
ellipse[2]) , 1, -1 )
irtk.imwrite(output_dir + "/seg_"+os.path.basename(file_img), labels )
irtk.imwrite(output_dir + "/res_"+os.path.basename(file_img), res )
# re-read the image in case we processed it
img = irtk.imread( file_img, dtype='float32' )
cropped_img = img[z_min:z_max+1,
y_min:y_max+1,
x_min:x_max+1]
cropped_img[labels==0] = -1
masked = cropped_img.bbox(crop=True)
irtk.imwrite(output_dir + "/masked_"+os.path.basename(file_img), masked )
# re-read the image in case we processed it
img = irtk.imread( file_img, dtype='float32' )
x0 = min_x_bare + (max_x_bare - min_x_bare) / 2
y0 = min_y_bare + (max_y_bare - min_y_bare) / 2
ofd = get_OFD(ga)/img.header['pixelSize'][0]
cropped_img = img[min_z_bare:max_z_bare+1,
max(0,int(round(y0-ofd/2))):min(img.shape[1],int(round(y0+ofd/2+1))),
max(0,int(round(x0-ofd/2))):min(img.shape[2],int(round(x0+ofd/2+1)))].copy()
irtk.imwrite(output_dir + "/very_large_"+os.path.basename(file_img),
cropped_img )
cropped_proba = res[min_z_bare:max_z_bare+1,
max(0,int(round(y0-ofd/2))):min(img.shape[1],int(round(y0+ofd/2+1))),
max(0,int(round(x0-ofd/2))):min(img.shape[2],int(round(x0+ofd/2+1)))].copy()
irtk.imwrite(output_dir + "/proba_"+os.path.basename(file_img),
cropped_proba )
def run_detection( filename, ga, output_folder ):
file_id = os.path.basename(filename).split('.')[0]
if '_' in os.path.basename(filename):
patient_id = file_id.split('_')[0]
else:
patient_id = file_id
print patient_id
# brain detection
vocabulary = "../brain-detector/trained_model/vocabulary_0.npy"
mser_detector = "../brain-detector/trained_model/mser_detector_0_LinearSVC"
mask_file = output_folder +"/" + file_id + "/brain_mask.nii.gz"
cmd = [ "python",
"../brain-detector/fetalMask_detection.py",
filename,
str(ga),
mask_file,
"--classifier", mser_detector,
"--vocabulary", vocabulary
]
print ' '.join(cmd)
proc = subprocess.Popen(cmd,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE )
(out, err) = proc.communicate()
print out
print err
# heart and body detection
## preprocessing (resampling+denoising)
img = irtk.imread( filename, dtype='float32', force_neurological=True )
img = irtk.Image(nd.median_filter(img.view(np.ndarray),(3,5,5)),img.get_header())
scale = get_CRL(ga)/get_CRL(30.0)
img = img.resample( 1.0*scale, interpolation='bspline' )
brain_center = img.WorldToImage( get_center_brain_detection(mask_file) )[::-1]
new_filename = output_folder + "/" + file_id + "/" + os.path.basename(filename)
irtk.imwrite(new_filename,img)
n_jobs = 5
output_folder1 = output_folder + "/" + file_id + "/prediction_1/"
output_folder2 = output_folder + "/" + file_id + "/prediction_2"
detector1 = "trained_model/stage1"
detector2 = "trained_model/stage2"
shape_model = "trained_model/stage1/shape_model.pk"
cmd = [ "python", "predict.py",
"--input", new_filename,
"--output", output_folder1,
"--output2", output_folder2,
"--detector", detector1,
"--detector2", detector2,
"--padding", str(10),
"--chunk_size", str(int(1e5)),
"--n_jobs", str(n_jobs),
"--brain_center"] + map(str,brain_center) + \
["--shape_model",shape_model,
"-l", str(0.5),
"--shape_optimisation",
"--narrow_band",
"--selective"]
print ' '.join(cmd)
proc = subprocess.Popen(cmd,
stdout=subprocess.PIPE,
stderr=subprocess.PIPE)
(out, err) = proc.communicate()
print out
print err
return
print args
start = time()
detector = heartdetector.HeartDetector( name=args.forest )
detector.load()
if not args.time:
print detector
if not os.path.exists("predictions/"+args.patient_id):
os.makedirs("predictions/"+args.patient_id)
all_frames = sorted(glob("denoised/"+args.patient_id+"_frame*.nii.gz"))
tmp = irtk.imread(all_frames[0])
mask = irtk.zeros(tmp.get_header(),dtype='float32')
for f in all_frames:
if "_seg" in f:
continue
mask += irtk.imread(f)
mask = (mask > 0).astype('uint8')
# if args.frame is not None:
# all_frames = [all_frames[args.frame]]
# elif not args.all:
# ED,ES = get_ED_ES(args.patient_id)
# all_frames = [all_frames[ED],all_frames[ES]]
for f in all_frames:
def preprocess_training_data(patient_id,
img_folder,
seg_folder,
resample,
offline=False,
online=True):
if offline or online:
if (offline and os.path.exists("offline_preprocessing/" + patient_id +
"_img.nii.gz")
and os.path.exists("offline_preprocessing/" + patient_id +
"_seg.nii.gz")):
return
img = irtk.imread(img_folder + "/" + patient_id + ".nii.gz",
dtype='float32')
seg = irtk.imread(seg_folder + "/" + patient_id + "_seg.nii.gz",
dtype="uint8")
wall = nd.binary_dilation(
seg,
morphology.ball(int(12.5 * 0.001 / seg.header['pixelSize'][0])))
wall = wall.astype('int')
points = np.transpose(np.nonzero(wall))[::4]
center, S, V = fit_ellipsoidPCA(points)
if V[0, 0] < 0:
V *= -1
points = np.transpose(np.nonzero(wall))
projections = np.dot(points - center, V[0])
# valves
index = projections > (projections.max() -
40.0 * 0.001 / seg.header['pixelSize'][0])
#print "VALVE size:",np.sum(index), projections.max(), 40.0*0.001/seg.header['pixelSize'][0]
wall[points[index, 0], points[index, 1], points[index, 2]] = 2
#print "VALVE1", wall.max()
wall = irtk.Image(wall, seg.get_header())
img = img.resample(pixelSize=resample,
interpolation='linear').rescale(0, 1000)
seg = seg.transform(target=img,
interpolation="nearest").astype('uint8')
wall = wall.transform(target=img,
interpolation="nearest").astype('uint8')
wall[seg > 0] = 0
seg[wall == 1] = 2
seg[wall == 2] = 3
#print "VALVE2", seg.max()
#irtk.imwrite("debug/"+patient_id+"_border.nii.gz",seg)
seg[img == 0] = 255
if offline:
irtk.imwrite("offline_preprocessing/" + patient_id + "_img.nii.gz",
img)
irtk.imwrite("offline_preprocessing/" + patient_id + "_seg.nii.gz",
seg)
return
if not online:
img = irtk.imread("offline_preprocessing/" + patient_id +
"_img.nii.gz")
seg = irtk.imread("offline_preprocessing/" + patient_id +
"_seg.nii.gz")
mask = irtk.ones(img.get_header(), dtype='uint8')
mask[img == 0] = 0
return {
'patient_id': patient_id,
'img': img,
'seg': seg,
'extra_layers': np.array([], dtype='float32'),
'metadata': None,
'mask': mask
}
print args
start = time()
detector = heartdetector.HeartDetector(name=args.forest)
detector.load()
if not args.time:
print detector
if not os.path.exists("predictions/" + args.patient_id):
os.makedirs("predictions/" + args.patient_id)
all_frames = sorted(glob("denoised/" + args.patient_id + "_frame*.nii.gz"))
tmp = irtk.imread(all_frames[0])
mask = irtk.zeros(tmp.get_header(), dtype='float32')
for f in all_frames:
if "_seg" in f:
continue
mask += irtk.imread(f)
mask = (mask > 0).astype('uint8')
# if args.frame is not None:
# all_frames = [all_frames[args.frame]]
# elif not args.all:
# ED,ES = get_ED_ES(args.patient_id)
# all_frames = [all_frames[ED],all_frames[ES]]
for f in all_frames:
sizes3 = np.load( args.detector + '/sizes3.npy' )
offsets4 = np.load( args.detector + '/offsets4.npy' )
sizes4 = np.load( args.detector + '/sizes4.npy' )
offsets5 = np.load( args.detector + '/offsets5.npy' )
offsets6 = np.load( args.detector + '/offsets6.npy' )
clf_heart = joblib.load( args.detector + '/clf_heart' )
reg_heart = joblib.load( args.detector + '/reg_heart' )
clf_heart.set_params(n_jobs=args.n_jobs)
reg_heart.set_params(n_jobs=args.n_jobs)
print "done loading detectors"
print "preprocessing..."
img = irtk.imread( args.input, dtype='float32', force_neurological=True )
grad = irtk.Image(nd.gaussian_gradient_magnitude( img, 0.5 ),
img.get_header())
sat = integral_image(img)
sat_grad = integral_image(grad)
blurred_img = nd.gaussian_filter(img,0.5)
gradZ = nd.sobel( blurred_img, axis=0 ).astype('float32')
gradY = nd.sobel( blurred_img, axis=1 ).astype('float32')
gradX = nd.sobel( blurred_img, axis=2 ).astype('float32')
irtk.imwrite(args.output + "/img.nii.gz", img)
irtk.imwrite(args.output + "/grad.nii.gz", grad)
#!/usr/bin/python
import sys, os
import irtk
import numpy as np
def rand(scale):
return float(scale) * (np.random.random(1) - 0.5) * 2
img = irtk.imread(sys.argv[1])
seg = irtk.imread(sys.argv[2])
prefix = sys.argv[3]
tx, ty, tz = img.ImageToWorld([(img.shape[2] - 1) / 2, (img.shape[1] - 1) / 2,
(img.shape[0] - 1) / 2])
centering = irtk.RigidTransformation(tx=-tx, ty=-ty, tz=-tz)
t = irtk.RigidTransformation(tx=rand(img.header['pixelSize'][0] * 10),
ty=rand(img.header['pixelSize'][1] * 10),
tz=rand(img.header['pixelSize'][2] * 10),
rx=rand(30),
ry=rand(30),
rz=rand(30))
print t
t = centering.invert() * t * centering
new_img = img.transform(t, target=img.get_header(), interpolation='linear')
new_seg = seg.transform(t, target=img.get_header(), interpolation='nearest')
def preprocess_training_data( patient_id,
img_folder,
seg_folder,
resample,
offline=False,
online=True):
if offline or online:
if ( offline
and os.path.exists( "offline_preprocessing/"+patient_id+"_img.nii.gz" )
and os.path.exists( "offline_preprocessing/"+patient_id+"_seg.nii.gz" ) ):
return
img = irtk.imread( img_folder + "/" + patient_id + ".nii.gz",
dtype='float32' )
seg = irtk.imread( seg_folder +"/"+patient_id+"_seg.nii.gz",
dtype="uint8" )
wall = nd.binary_dilation( seg,
morphology.ball(int(12.5*0.001/seg.header['pixelSize'][0])) )
wall = wall.astype('int')
points = np.transpose(np.nonzero(wall))[::4]
center,S,V = fit_ellipsoidPCA( points )
if V[0,0] < 0:
V *= -1
points = np.transpose(np.nonzero(wall))
projections = np.dot(points-center,V[0])
# valves
index = projections > (projections.max() - 40.0*0.001/seg.header['pixelSize'][0])
#print "VALVE size:",np.sum(index), projections.max(), 40.0*0.001/seg.header['pixelSize'][0]
wall[points[index,0],
points[index,1],
points[index,2]] = 2
#print "VALVE1", wall.max()
wall = irtk.Image(wall,seg.get_header())
img = img.resample( pixelSize=resample, interpolation='linear' ).rescale(0,1000)
seg = seg.transform(target=img,interpolation="nearest").astype('uint8')
wall = wall.transform(target=img,interpolation="nearest").astype('uint8')
wall[seg>0] = 0
seg[wall==1] = 2
seg[wall==2] = 3
#print "VALVE2", seg.max()
#irtk.imwrite("debug/"+patient_id+"_border.nii.gz",seg)
seg[img==0] = 255
if offline:
irtk.imwrite( "offline_preprocessing/"+patient_id+"_img.nii.gz", img )
irtk.imwrite( "offline_preprocessing/"+patient_id+"_seg.nii.gz", seg )
return
if not online:
img = irtk.imread( "offline_preprocessing/"+patient_id+"_img.nii.gz" )
seg = irtk.imread( "offline_preprocessing/"+patient_id+"_seg.nii.gz" )
mask = irtk.ones( img.get_header(), dtype='uint8' )
mask[img==0] = 0
return { 'patient_id': patient_id,
'img' : img,
'seg' : seg,
'mask' : mask }
import scipy.ndimage as nd
from skimage import morphology
import argparse
parser = argparse.ArgumentParser(
description='' )
parser.add_argument( '--seg', type=str, required=True )
parser.add_argument( '--img', type=str, required=True )
parser.add_argument( '--output', type=str, required=True )
parser.add_argument( '--narrow_band', type=int, default=5 )
parser.add_argument( '--debug', action="store_true", default=False )
args = parser.parse_args()
seg = irtk.imread( args.seg, dtype='int32', force_neurological=True )
img = irtk.imread( args.img, dtype='float32', force_neurological=True ).rescale(0,1000)
res = irtk.zeros( seg.get_header(), dtype='uint8' )
ball = morphology.ball( args.narrow_band )
nb_labels = 5
# for i in range(1,5):
# tmp_seg = (seg==i).astype('int32')
# # crop
# x_min,y_min,z_min,x_max,y_max,z_max = (tmp_seg).bbox()
# mask = tmp_seg[max(0,z_min-2*args.narrow_band):min(seg.shape[0],z_max+2*args.narrow_band+1),
# max(0,y_min-2*args.narrow_band):min(seg.shape[1],y_max+2*args.narrow_band+1),
# max(0,x_min-2*args.narrow_band):min(seg.shape[2],x_max+2*args.narrow_band+1)]
#!/usr/bin/python
import sys
import numpy as np
import irtk
full_file = sys.argv[1]
cropped_file = sys.argv[2]
full_img = irtk.imread(full_file, dtype='float32')
cropped_img = irtk.imread(cropped_file, dtype='float32')
(z, y, x), score = irtk.match_template(full_img, cropped_img, pad_input=False)
print score
print ' '.join(
map(str,
[full_img.shape[0], full_img.shape[1], full_img.shape[2], z, y, x]))
def get_training_data( file_img, file_mask, r ):
# create mask
input_mask = irtk.imread( file_mask )
x_min, y_min, z_min, x_max, y_max, z_max = (input_mask == 0).bbox()
background = irtk.zeros( input_mask.get_header(), dtype='uint8' )
background[z_min:z_max+1,
y_min:y_max+1,
x_min:x_max+1] = 1
background = nd.morphological_gradient( background, size=7)
n = background[z_min+1:z_max,
y_min+1:y_max,
x_min+1:x_max].sum()
z = np.random.randint(low=0, high=input_mask.shape[0],size=1.25*n)
y = np.random.randint(low=0, high=input_mask.shape[1],size=1.25*n)
x = np.random.randint(low=0, high=input_mask.shape[2],size=1.25*n)
background[z,y,x] = 1
background[z_min+1:z_max,
y_min+1:y_max,
x_min+1:x_max] = 0
foreground = (input_mask == 1).astype('uint8')
new_mask = irtk.zeros( input_mask.get_header(), dtype='uint8' )
new_mask[foreground == 1] = 1
new_mask[background != 0] = 2
img = irtk.imread( file_img, dtype='float32' )
X = []
Y = []
for z in xrange(img.shape[0]):
YX = np.transpose( np.nonzero( foreground[z] ) )
if DEBUG:
YX = YX[::10]
else:
YX = YX[::2]
if YX.shape[0] == 0:
continue
patches = extract_patches2D( img[z], r, YX )
patches = np.reshape( patches, (patches.shape[0],patches.shape[1]*patches.shape[2]) )
print patches.shape, YX.shape
X.extend( patches )
Y.extend( [1]*len(YX) )
for z in xrange(img.shape[0]):
YX = np.transpose( np.nonzero( background[z] ) )
if DEBUG:
YX = YX[::10]
else:
YX = YX[::2]
if YX.shape[0] == 0:
continue
patches = extract_patches2D( img[z], r, YX )
patches = np.reshape( patches, (patches.shape[0],patches.shape[1]*patches.shape[2]) )
print patches.shape, YX.shape
X.extend( patches )
Y.extend( [0]*len(YX) )
return X, Y
#!/usr/bin/python import sys import numpy as np import irtk from pyhull.convex_hull import ConvexHull from irtk.vtk2irtk import voxellise seg = irtk.imread(sys.argv[1]) output = sys.argv[2] ZYX = np.transpose(np.nonzero(seg)) pts = seg.ImageToWorld(ZYX[:, ::-1]) hull = ConvexHull(pts) img = voxellise(hull.points, hull.vertices, header=seg.get_header()) irtk.imwrite(output, img)
return labels == best_label
def background_distance(img,metric='geodesic',includeEDT=True):
background = get_background(img)
if metric == "euclidean":
distanceMap = edt( img, background )
elif metric == "geodesic":
distanceMap = gdt( img, background, includeEDT )
else:
raise ValueError("Unknown metric: "+ metric)
return irtk.Image(distanceMap,img.get_header())
if __name__ == "__main__":
img = irtk.imread( sys.argv[1], dtype="float64" )
#filtered = nd.minimum_filter(img,5)
filtered = nd.gaussian_gradient_magnitude(img,0.5)
img = irtk.Image(filtered,img.get_header())
irtk.imwrite("test2.nii.gz",img)
exit(0)
img = world_align(img,pixelSize=[2,2,2,1])
irtk.imwrite("distanceEDT.nii.gz",background_distance(img,metric="euclidean"))
irtk.imwrite( "distanceGDT.nii.gz", background_distance(img,metric="geodesic"))
def detect_mser(raw_file,
ga,
vocabulary,
mser_detector,
NEW_SAMPLING,
output_folder="debug",
DEBUG=False,
return_image_regions=False):
OFD = get_OFD(ga) / NEW_SAMPLING
BPD = get_BPD(ga) / NEW_SAMPLING
max_e = 0.64
mser = cv2.MSER(_delta=5,
_min_area=60,
_max_area=14400,
_max_variation=0.15,
_min_diversity=.1,
_max_evolution=200,
_area_threshold=1.01,
_min_margin=0.003,
_edge_blur_size=5)
sift = cv2.SIFT(nfeatures=0,
nOctaveLayers=3,
contrastThreshold=0.04,
edgeThreshold=10,
sigma=0.8)
siftExtractor = cv2.DescriptorExtractor_create("SIFT")
voca = np.load(open(vocabulary, 'rb'))
classifier = neighbors.NearestNeighbors(1, algorithm='kd_tree')
N = voca.shape[0]
classifier.fit(voca)
#flann = pyflann.FLANN()
#flann.build_index( voca.astype('float32') )
svc = joblib.load(mser_detector)
img = irtk.imread(raw_file, dtype='float32')
img = img.resample2D(NEW_SAMPLING).saturate().rescale().astype('uint8')
detections = []
image_regions = []
for z in range(img.shape[0]):
detections.append([])
image_regions.append([])
# Extract MSER
#print "extracting mser"
contours = mser.detect(img[z, :, :])
#print "mser done"
if DEBUG:
img_color = cv2.cvtColor(img[z, :, :], cv2.cv.CV_GRAY2RGB)
for c in contours:
ellipse = cv2.fitEllipse(
np.array(map(lambda x: [x], c), dtype='int32'))
cv2.ellipse(img_color,
(ellipse[0],
(ellipse[1][0], ellipse[1][1]), ellipse[2]),
(0, 0, 255))
cv2.imwrite(output_folder + "/" + str(z) + "_all_mser_.png",
img_color)
img_color_mser = cv2.cvtColor(img[z, :, :], cv2.cv.CV_GRAY2RGB)
# Filter MSER
selected_mser = []
mask = np.zeros((img.shape[1], img.shape[2]), dtype='uint8')
#print "fitting ellipses"
for c in contours:
ellipse = cv2.fitEllipse(
np.reshape(c, (c.shape[0], 1, 2)).astype('int32'))
# filter by size
if (ellipse[1][0] > OFD or ellipse[1][1] > OFD
or ellipse[1][0] < 0.5 * OFD or ellipse[1][1] < 0.5 * OFD):
continue
# filter by eccentricity
if math.sqrt(1 -
(np.min(ellipse[1]) / np.max(ellipse[1]))**2) > max_e:
continue
cv2.ellipse(mask, ellipse, 255, -1)
selected_mser.append((c, ellipse))
#print "ellipses done"
if len(selected_mser) == 0:
continue
# Extract SIFT
#print "extracting SIFT"
keypoints = sift.detect(img[z, :, :], mask=mask)
#print "SIFT done"
if keypoints is None or len(keypoints) == 0:
continue
(keypoints,
descriptors) = siftExtractor.compute(img[z, :, :], keypoints)
# words = np.zeros(len(keypoints),dtype="int")
# for i,d in enumerate(descriptors):
# words[i] = classifier.kneighbors(d, return_distance=False)
words = classifier.kneighbors(descriptors, return_distance=False)
#words, dist = flann.nn_index( descriptors.astype('float32') )
for i, (c, ellipse) in enumerate(selected_mser):
# Compute histogram
hist = np.zeros(N, dtype='float')
for ki, k in enumerate(keypoints):
if is_in_ellipse(k.pt, ellipse):
hist[words[ki]] += 1
# Normalize histogram
norm = np.linalg.norm(hist)
if norm > 0:
hist /= norm
cl = svc.predict(hist).flatten()
if DEBUG:
if cl == 1:
opacity = 0.4
img_color = cv2.cvtColor(img[z, :, :], cv2.cv.CV_GRAY2RGB)
for p in c:
img_color[p[1], p[0], :] = (
(1 - opacity) * img_color[p[1], p[0], :] +
opacity * np.array([0, 255, 0]))
cv2.imwrite(
output_folder + "/" + str(z) + '_' + str(i) +
"_region.png", img_color)
img_color = cv2.cvtColor(img[z, :, :], cv2.cv.CV_GRAY2RGB)
cv2.ellipse(img_color,
(ellipse[0],
(ellipse[1][0], ellipse[1][1]), ellipse[2]),
(0, 0, 255))
for k_id, k in enumerate(keypoints):
if is_in_ellipse(k.pt, ellipse):
if cl == 1:
cv2.circle(img_color, (int(k.pt[0]), int(k.pt[1])),
2, (0, 255, 0), -1)
else:
cv2.circle(img_color, (int(k.pt[0]), int(k.pt[1])),
2, (0, 0, 255), -1)
cv2.imwrite(
output_folder + "/" + str(z) + '_' + str(i) + ".png",
img_color)
cv2.ellipse(img_color_mser,
(ellipse[0],
(ellipse[1][0], ellipse[1][1]), ellipse[2]),
(0, 255, 0) if cl == 1 else (0, 0, 255))
if cl == 1:
ellipse_center = [ellipse[0][0], ellipse[0][1], z]
print c.shape
if return_image_regions:
image_regions[-1].append((ellipse_center, c))
else:
region = np.hstack((c, [[z]] * c.shape[0]))
detections[-1].append((img.ImageToWorld(ellipse_center),
img.ImageToWorld(region)))
if DEBUG:
cv2.imwrite(output_folder + "/" + str(z) + "_color_mser.png",
img_color_mser)
if return_image_regions:
return image_regions
else:
return detections
import os
def get_ED_ES(patient_id):
f = "/vol/biomedic/users/kpk09/DATASETS/CETUS_data/Training/Images/" + patient_id + "/" + patient_id + "_ED_ES_time.txt"
f = open(f, "rb")
res = []
for line in f:
line = line.rstrip() # chomp
res.append(int(line.split(' ')[-1]))
return res
all_files = glob("nifti/*_seg.nii.gz")
m = 0.0
n = 0.0
for f in all_files:
patient_id = os.path.basename(f).split("_")[0]
frame_id = int(os.path.basename(f).split("_")[1][len("frame"):])
ED_ES = get_ED_ES(patient_id)
print patient_id, frame_id, ED_ES
if frame_id == ED_ES[1]:
mask = irtk.imread(f).resample(0.001, interpolation='nearest')
m += mask.sum()
n += 1
m /= n
print "mean heart volume:", m
parser.add_argument( '--brain_center', type=float, nargs=3 )
parser.add_argument( '--verbose', action="store_true", default=False )
parser.add_argument( '--debug', action="store_true", default=False )
parser.add_argument( '--n_jobs', type=int, default=20 )
parser.add_argument( '--chunk_size', type=int, default=int(3e6) )
parser.add_argument( '-l', type=float, default=0.5 )
parser.add_argument( '--theta', type=int, default=90 )
args = parser.parse_args()
print args
output_dir = os.path.dirname(args.output)
if not os.path.exists(output_dir):
os.makedirs(output_dir)
template = irtk.imread(args.template,force_neurological=True)
img = irtk.imread(args.input,force_neurological=True)
if args.ga > 0:
scale = get_CRL(args.ga)/get_CRL(30.0)
resized_img = img.resample( 1.0*scale, interpolation='bspline' ).rescale(0,1000)
resized_input = output_dir + "/resized_" + os.path.basename(args.input)
irtk.imwrite( resized_input, resized_img )
heart_center = resized_img.WorldToImage([0,0,0])[::-1]#np.array(resized_img.shape,dtype='float32')/2
if not args.brain_center:
brain_center = heart_center + np.array([0,0,100])
else:
brain_center = np.array(args.brain_center,dtype='float32')
else:
#!/usr/bin/python
import irtk
from glob import glob
import os
import sys
import numpy as np
if not os.path.exists("png"):
os.makedirs("png")
if len(sys.argv) == 1:
all_files = glob("predictions/*/iter4_*_hard.nii.gz")
else:
all_files = glob("predictions/" + sys.argv[1] + "/iter4_*_hard.nii.gz")
for f in all_files:
print f
name = os.path.basename(f)[len("iter4_"):-len("_hard.nii.gz")]
img = irtk.imread("denoised/" + name + ".nii.gz", dtype='int32')
mask = irtk.imread(f).transform(target=img.get_header(),
interpolation="nearest")
irtk.imshow(img, mask, filename="png/" + name + ".png", opacity=0.4)
def process_file( f,
ga,
coordinates,
size,
classifier,
N,
DEBUG=False ):
X = []
Y = []
scan_id = os.path.basename(f).split('.')[0]
max_e = 0.64
OFD = get_OFD(30.0,centile=50)
mser = cv2.MSER( _delta=5,
_min_area=60,
_max_area=14400,
_max_variation=0.15,
_min_diversity=.1,
_max_evolution=200,
_area_threshold=1.01,
_min_margin=0.003,
_edge_blur_size=5)
sift = cv2.SIFT( nfeatures=0,
nOctaveLayers=3,
contrastThreshold=0.04,
edgeThreshold=10,
sigma=0.8)
siftExtractor = cv2.DescriptorExtractor_create("SIFT")
box = np.array(map(float,coordinates.split(',')),dtype='float')
size = np.array(map(float,size.split(',')),dtype='float')
img = irtk.imread( f, dtype='float32', force_neurological=False )
old_img = img.copy()
## Resample
img = resampleOFD( img, ga )
# Adjust coordinates and size
box = img.WorldToImage( old_img.ImageToWorld(box[::-1]) )[::-1]
box_size = size * old_img.header['pixelSize'][:3][::-1]/img.header['pixelSize'][:3][::-1]
z0,y0,x0 = box.astype('int')
d0,h0,w0 = box_size.astype('int')
brain_center = (x0 + w0/2, y0 + h0/2)
## Contrast-stretch with saturation
img = img.saturate(1,99).rescale().astype('uint8')
for z in range(img.shape[0]):
contours = mser.detect(img[z])
keypoints = sift.detect(img[z])
if keypoints is None or len(keypoints) == 0:
continue
(keypoints, descriptors) = siftExtractor.compute(img[z],keypoints)
for i,c in enumerate(contours):
hist = np.zeros(N, dtype='float')
ellipse = cv2.fitEllipse(np.array(map(lambda x:[x],
c),dtype='int32'))
# filter by size
if ( ellipse[1][0] > OFD
or ellipse[1][1] > OFD
or ellipse[1][0] < 0.5*OFD
or ellipse[1][1] < 0.5*OFD ) :
continue
# filter by eccentricity
if math.sqrt(1-(np.min(ellipse[1])/np.max(ellipse[1]))**2) > max_e:
continue
distance = math.sqrt((ellipse[0][0]-brain_center[0])**2
+(ellipse[0][1]-brain_center[1])**2)
if max(w0,h0)/2 >= distance >= min(w0,h0)/8:
continue
for k,d in zip(keypoints,descriptors):
if is_in_ellipse(k.pt,ellipse):
c = classifier.kneighbors(d, return_distance=False)
hist[c] += 1
# Normalize histogram
norm = np.linalg.norm(hist)
if norm > 0:
hist /= norm
if distance > max(w0,h0)/4:
if DEBUG: print 0
X.append(hist)
Y.append(0)
else:
if distance < min(w0,h0)/8 and z0 + d0/8 <= z <= z0+7*d0/8:
if DEBUG: print 1
X.append(hist)
Y.append(1)
else:
continue
if DEBUG and Y[-1] == 1:
img_color = cv2.cvtColor( img[z], cv2.cv.CV_GRAY2RGB )
cv2.ellipse( img_color, (ellipse[0],
(ellipse[1][0],ellipse[1][1]),
ellipse[2]) , (0,0,255))
for k_id,k in enumerate(keypoints):
if is_in_ellipse(k.pt,ellipse):
if Y[-1] == 1:
cv2.circle( img_color,
(int(k.pt[0]),int(k.pt[1])),
2,
(0,255,0),
-1)
else:
cv2.circle( img_color,
(int(k.pt[0]),int(k.pt[1])),
2,
(0,0,255),
-1)
cv2.imwrite("debug/"+scan_id+'_'+str(z) + '_' +str(i) +'_'+str(k_id)+".png",img_color)
return X,Y
#!/usr/bin/python import sys import numpy as np import irtk full_file = sys.argv[1] cropped_file = sys.argv[2] full_img = irtk.imread(full_file, dtype="float32") cropped_img = irtk.imread(cropped_file, dtype="float32") (z, y, x), score = irtk.match_template(full_img, cropped_img, pad_input=False) print score print " ".join(map(str, [full_img.shape[0], full_img.shape[1], full_img.shape[2], z, y, x]))
if os.environ['USER'] == "kevin":
raw_folder = "/home/kevin/Imperial/PhD/DATASETS/Originals/"
vocabulary = "/home/kevin/Imperial/PhD/MyPHD/Detection/BOW/pipeline2/LEARNING/vocabulary_" + args.fold + ".npy"
mser_detector = "/home/kevin/Imperial/PhD/MyPHD/Detection/BOW/pipeline2/LEARNING/mser_detector_" + args.fold + "_linearSVM"
ga_file = "/home/kevin/Imperial/PhD/MyPHD/Detection/BOW/pipeline2/LEARNING/metadata/ga.csv"
else:
raw_folder = "/vol/biomedic/users/kpk09/DATASETS/Originals"
vocabulary = "/vol/biomedic/users/kpk09/pipeline2/LEARNING/vocabulary_" + args.fold + ".npy"
mser_detector = "/vol/biomedic/users/kpk09/pipeline2/LEARNING/mser_detector_" + args.fold + "_linearSVM"
ga_file = "/vol/biomedic/users/kpk09/pipeline2/LEARNING/metadata/ga.csv"
print "Detect MSER regions"
detections = []
NEW_SAMPLING = 0.8
img = irtk.imread(filename, dtype="float32").saturate().rescale()
image_regions = detect_mser(filename,
ga,
vocabulary,
mser_detector,
NEW_SAMPLING,
DEBUG=args.debug,
output_folder=output_dir,
return_image_regions=True)
# flatten list
# http://stackoverflow.com/questions/406121/flattening-a-shallow-list-in-python
import itertools
chain = itertools.chain(*image_regions)
image_regions = list(chain)
#!/usr/bin/python
import irtk
from glob import glob
import os
import sys
import numpy as np
if not os.path.exists("png"):
os.makedirs("png")
if len(sys.argv) == 1:
all_files = glob( "predictions/*/iter4_*_hard.nii.gz" )
else:
all_files = glob( "predictions/"+sys.argv[1]+"/iter4_*_hard.nii.gz" )
for f in all_files:
print f
name = os.path.basename(f)[len("iter4_"):-len("_hard.nii.gz")]
img = irtk.imread("denoised/"+name+".nii.gz",dtype='int32')
mask = irtk.imread(f).transform(target=img.get_header(),interpolation="nearest")
irtk.imshow(img,mask,filename="png/"+name+".png",opacity=0.4)
#!/usr/bin/python
import sys
import numpy as np
import irtk
# img1 = irtk.imread("reconstruction/t2.nii", dtype='float32')
# img2 = irtk.imread("reconstruction/t2seg.nii", dtype='float32')
# irtk.imwrite( "reconstruction/segfixed.nii", img2.transform(target=img1) )
shape = map(int, sys.argv[1:4])
z = int(sys.argv[4])
y = int(sys.argv[5])
x = int(sys.argv[6])
target = irtk.imread(sys.argv[7])
img = irtk.imread(sys.argv[8], dtype='float32')
new_data = np.zeros(shape, dtype='int32')
new_data[z:z + img.shape[0], y:y + img.shape[1], x:x + img.shape[2]] = img
new_img = irtk.Image(new_data, target.get_header())
irtk.imwrite(sys.argv[9], new_img)
