def __init__(self,ptcode,ctcode,basedir, seed_th=[600], show=True,
normalize=False, modelname='model'):
import os
import numpy as np
import visvis as vv
from visvis import ssdf
from stentseg.utils import PointSet, _utils_GUI
from stentseg.utils.datahandling import select_dir, loadvol, loadmodel
from stentseg.stentdirect.stentgraph import create_mesh
from stentseg.stentdirect import stentgraph, StentDirect, getDefaultParams
from stentseg.stentdirect import AnacondaDirect, EndurantDirect, NellixDirect
from stentseg.utils.visualization import show_ctvolume
from stentseg.utils.picker import pick3d, label2worldcoordinates, label2volindices
import scipy
from scipy import ndimage
import copy
# Select dataset to register
cropname = 'prox'
# phase = 10
#dataset = 'avgreg'
#what = str(phase) + dataset # avgreg
what = 'avgreg'
# Load volumes
s = loadvol(basedir, ptcode, ctcode, cropname, what)
# sampling was not properly stored after registration for all cases: reset sampling
vol_org = copy.deepcopy(s.vol)
s.vol.sampling = [vol_org.sampling[1], vol_org.sampling[1], vol_org.sampling[2]] # z,y,x
s.sampling = s.vol.sampling
vol = s.vol
## Initialize segmentation parameters
stentType = 'nellix' # 'zenith';'nellix' runs modified pruning algorithm in Step3
p = getDefaultParams(stentType)
p.seed_threshold = seed_th # step 1 [lower th] or [lower th, higher th]
# p.seedSampleRate = 7 # step 1, nellix
p.whatphase = what
## Perform segmentation
# Instantiate stentdirect segmenter object
if stentType == 'anacondaRing':
sd = AnacondaDirect(vol, p) # inherit _Step3_iter from AnacondaDirect class
#runtime warning using anacondadirect due to mesh creation, ignore
elif stentType == 'endurant':
sd = EndurantDirect(vol, p)
elif stentType == 'nellix':
sd = NellixDirect(vol, p)
else:
sd = StentDirect(vol, p)
## show histogram and normalize
# f = vv.figure(3)
# a = vv.gca()
# vv.hist(vol, drange=(300,vol.max()))
# normalize vol to certain limit
if normalize:
sd.Step0(3071)
vol = sd._vol
# b= vv.hist(vol, drange=(300,3071))
# b.color = (1,0,0)
## Perform step 1 for seeds
sd.Step1()
## Visualize
if show:
fig = vv.figure(2); vv.clf()
fig.position = 0.00, 22.00, 1920.00, 1018.00
clim = (0,2000)
# Show volume and model as graph
a1 = vv.subplot(121)
a1.daspect = 1,1,-1
# t = vv.volshow(vol, clim=clim)
t = show_ctvolume(vol, axis=a1, showVol='MIP', clim =clim, isoTh=250,
removeStent=False, climEditor=True)
label = pick3d(vv.gca(), vol)
sd._nodes1.Draw(mc='b', mw = 2) # draw seeded nodes
#sd._nodes2.Draw(mc='b', lc = 'g') # draw seeded and MCP connected nodes
vv.xlabel('x (mm)');vv.ylabel('y (mm)');vv.zlabel('z (mm)')
# Show volume and cleaned up graph
a2 = vv.subplot(122)
a2.daspect = 1,1,-1
sd._nodes1.Draw(mc='b', mw = 2) # draw seeded nodes
# t = vv.volshow(vol, clim=clim)
# label = pick3d(vv.gca(), vol)
# sd._nodes2.Draw(mc='b', lc='g')
# sd._nodes3.Draw(mc='b', lc='g')
vv.xlabel('x (mm)');vv.ylabel('y (mm)');vv.zlabel('z (mm)')
# # Show the mesh
#===============================================================================
# a3 = vv.subplot(133)
# a3.daspect = 1,1,-1
# t = vv.volshow(vol, clim=clim)
# pick3d(vv.gca(), vol)
# #sd._nodes3.Draw(mc='b', lc='g')
# m = vv.mesh(bm)
# m.faceColor = 'g'
# # _utils_GUI.vis_spared_edges(sd._nodes3)
# vv.xlabel('x (mm)');vv.ylabel('y (mm)');vv.zlabel('z (mm)')
#===============================================================================
# Use same camera
a1.camera = a2.camera #= a3.camera
switch = True
a1.axis.visible = switch
a2.axis.visible = switch
#a3.axis.visible = switch
## Store segmentation to disk
# Get graph model
model = sd._nodes1
# Build struct
s2 = vv.ssdf.new()
s2.sampling = s.sampling
s2.origin = s.origin
s2.stenttype = s.stenttype
s2.croprange = s.croprange
for key in dir(s):
if key.startswith('meta'):
suffix = key[4:]
s2['meta'+suffix] = s['meta'+suffix]
s2.what = what
s2.params = p
s2.stentType = stentType
# Store model (not also volume)
s2.model = model.pack()
# Save
filename = '%s_%s_%s_%s.ssdf' % (ptcode, ctcode, cropname, modelname+ what)
ssdf.save(os.path.join(basedir, ptcode, filename), s2)
print('saved to disk as {}.'.format(filename) )
## Make model dynamic (and store/overwrite to disk)
#===============================================================================
#
# import pirt
# from stentsegf.motion.dynamic import incorporate_motion_nodes, incorporate_motion_edges
#
# # Load deforms
# s = loadvol(basedir, ptcode, ctcode, cropname, '10deforms')
# deformkeys = []
# for key in dir(s):
# if key.startswith('deform'):
# deformkeys.append(key)
# deforms = [s[key] for key in deformkeys]
# deforms = [pirt.DeformationFieldBackward(*fields) for fields in deforms]
# paramsreg = s.params
#
# # Load model
# s = loadmodel(basedir, ptcode, ctcode, cropname, 'model'+what)
# model = s.model
#
# # Combine ...
# incorporate_motion_nodes(model, deforms, s.origin)
# incorporate_motion_edges(model, deforms, s.origin)
#
# # Save back
# filename = '%s_%s_%s_%s.ssdf' % (ptcode, ctcode, cropname, 'model'+what)
# s.model = model.pack()
# s.paramsreg = paramsreg
# ssdf.save(os.path.join(basedir, ptcode, filename), s)
# print('saved to disk as {}.'.format(filename) )
#===============================================================================
from visvis import ssdf
from stentseg.stentdirect import StentDirect, StentDirect_old, getDefaultParams, stentgraph
from stentseg.stentdirect.stentgraph import create_mesh
# Load volume data, use Aarray class for anisotropic volumes
vol = vv.volread('stent')
vol = vv.Aarray(vol,(1,1,1))
#stentvol = vv.ssdf.load(r'C:\Users\Maaike\Dropbox\UT MA3\Research Aortic Stent Grafts\Data_nonECG-gated\lspeas\lspeas_003.ssdf')
#vol = vv.Aarray(stentvol.vol,stentvol.sampling)
#stentvol = vv.ssdf.load('/home/almar/data/lspeas/LSPEAS_002/LSPEAS_002_1month_ring_avg3090.ssdf')
#vol = vv.Aarray(stentvol.vol,stentvol.sampling)
# Get parameters. Different scanners/protocols/stent material might need
# different parameters.
p = getDefaultParams()
p.graph_weakThreshold = 10 # step 3, stentgraph.prune_very_weak:
p.mcp_maxCoverageFronts = 0.03 # step 2, create MCP object
p.graph_expectedNumberOfEdges = 2 # 2 for zig-zag, 4 for diamond shaped
# # step 3, in stentgraph.prune_weak
#p.graph_trimLength = # step 3, stentgraph.prune_tails
#p.graph_strongThreshold # step 3, stentgraph.prune_weak and stentgraph.prune_redundant
p.seed_threshold = 800 # step 1
#p.graph_minimumClusterSize # step 3, stentgraph.prune_clusters
#p.mcp_speedFactor # step 2, speed image
# Instantiate stentdirect segmenter object
#sd = StentDirect_old(vol, p)
sd = StentDirect(vol, p)
ry = int(vol_crop.origin[1] / vol_crop.sampling[1] + 0.5)
ry = ry, ry + vol_crop.shape[1]
rx = int(vol_crop.origin[2] / vol_crop.sampling[2] + 0.5)
rx = rx, rx + vol_crop.shape[2]
s.croprange = rz, ry, rx # in world coordinates
vol2 = vol_crop
else:
vol2 = vol
## Initialize segmentation parameters
stentType = 'anacondaRing'
# 'anacondaRing' runs AnacondaDirect with modified Step3 and _Step3_iter
# 'branch' or 'nellix' runs Nellixdirect with modified seeding
p = getDefaultParams(stentType)
p.seed_threshold = [700] # step 1 [lower th] or [lower th, higher th]
p.mcp_speedFactor = 750 # step 2, costToCtValue;
# lower-> longer paths (costs low) -- higher-> short paths (costs high)
p.mcp_maxCoverageFronts = 0.008 # step 2, base.py; replaces mcp_evolutionThreshold
p.graph_weakThreshold = 700 # step 3, stentgraph.prune_very_weak
p.graph_expectedNumberOfEdges = 3 # step 3, stentgraph.prune_weak
p.graph_trimLength = 0 # step 3, stentgraph.prune_tails
p.graph_minimumClusterSize = 10 # step 3, stentgraph.prune_clusters
p.graph_strongThreshold = 1300 # step 3, stentgraph.prune_weak and stentgraph.prune_redundant
p.graph_min_strutlength = 5 # step 3, stent_anaconda prune_redundant
p.graph_max_strutlength = 13 # step 3, stent_anaconda prune_redundant
# p.graph_angleVector = 5 # step 3, corner detect
# p.graph_angleTh = 180 # step 3, corner detect
## Perform segmentation