def mean_shift_selection( votes,
n_points=5000,
bandwidth=10,
n=10,
cutoff=None,
debug=False,
debug_output="./" ):
points = np.argwhere(votes)
probas = votes[points[:,0],
points[:,1],
points[:,2]]
points, probas = random_pick(points,probas,n_points)
ms = MeanShift(bandwidth=bandwidth)
ms.fit(points)
weights = np.zeros( ms.cluster_centers_.shape[0], dtype='float' )
for i,c in enumerate(ms.cluster_centers_):
weights[i] = np.sum(probas[ms.labels_==i])
detection = ms.cluster_centers_[np.argsort(weights)[::-1]]
#points = points[np.argsort(weights)[::-1]]
#ms.labels_ = ms.labels_[np.argsort(weights)[::-1]]
weights = np.sort(weights)[::-1]
weights /= weights.max()
if debug:
print weights.tolist()
res = irtk.zeros(votes.get_header(),dtype='int')
points = np.array(detection,dtype='int')
if points.shape[0] > 30:
points = points[:30]
print np.arange(1,int(points.shape[0]+1))[::-1]
res[points[:,0],
points[:,1],
points[:,2]] = np.arange(1,int(points.shape[0]+1))#[::-1]
irtk.imwrite(debug_output+"/clusters.nii.gz",res)
irtk.imwrite(debug_output+"/clusters_centers.nii.gz",irtk.landmarks_to_spheres(res,r=5))
if cutoff is not None:
selected = np.sum( weights >= cutoff )
detection = detection[:selected]
weights = weights[:selected]
else:
if len(detection) > n:
detection = detection[:n]
weights = weights[:n]
return detection, weights
shutil.rmtree(args.output2)
os.renames(tmp_output+"/"+str(best_candidate), args.output2)
if os.path.exists(tmp_output):
shutil.rmtree(tmp_output)
else:
p = np.unravel_index(np.argmax(votes_heart), votes_heart.shape )
score_candidate( p-args.padding,
args.brain_center,
output=args.output2,
verbose=True )
landmark = irtk.zeros(hough_header,dtype='uint8')
landmark[p[0],p[1],p[2]] = 1
landmark = irtk.landmarks_to_spheres(landmark, r=10)
irtk.imwrite(args.output + "/detected_heart.nii.gz",landmark)
# for each landmark, run predict2, keep best result
if args.rw:
img_filename = args.output2 + "/img.nii.gz"
seg_filename = args.output2 + "/final_seg.nii.gz"
rw_filename = args.output2 + "/rw.nii.gz"
cmd = [ "python", os.path.dirname(__file__)+"/segment_random_walker.py",
"--img", img_filename,
"--seg", seg_filename,
# p_heart = np.unravel_index( np.argmax(votes_heart), img.shape )
landmarks = irtk.zeros(hough_header,dtype='uint8')
landmarks[p_left_lung[0],p_left_lung[1],p_left_lung[2]] = 1
landmarks[p_right_lung[0],p_right_lung[1],p_right_lung[2]] = 2
landmarks[p_heart[0],p_heart[1],p_heart[2]] = 3
if ( 0 <= p_liver[0] < landmarks.shape[0] and
0 <= p_liver[1] < landmarks.shape[1] and
0 <= p_liver[2] < landmarks.shape[2] ):
landmarks[p_liver[0],p_liver[1],p_liver[2]] = 4
if ( 0 <= p_brain[0] < landmarks.shape[0] and
0 <= p_brain[1] < landmarks.shape[1] and
0 <= p_brain[2] < landmarks.shape[2] ):
landmarks[p_brain[0],p_brain[1],p_brain[2]] = 5
landmarks = irtk.landmarks_to_spheres(landmarks, r=10)
irtk.imwrite(args.output + "/landmarks.nii.gz",landmarks)
irtk.imwrite(args.output + "/votes_blurred_left_lung.nii.gz",
irtk.Image(votes_left_lung,hough_header))
irtk.imwrite(args.output + "/votes_blurred_right_lung.nii.gz",
irtk.Image(votes_right_lung,hough_header))
irtk.imwrite(args.output + "/votes_blurred_heart.nii.gz",
irtk.Image(votes_heart,hough_header))
irtk.imwrite(args.output + "/votes_blurred_liver.nii.gz",
irtk.Image(votes_liver,hough_header))
if args.back_proj:
back_proj_left_lung = hough_votes_backprojection(np.ascontiguousarray(offsets_left_lung.astype('int32')),
votes_left_lung,
padding=args.padding)
