def uniform_weighting(self, n_regions=5, perc=95):
from numpy import union1d as union
from numpy import intersect1d as intersect
u, s = self.u, self.s
u_b = np.linspace(0, np.percentile(u, perc), n_regions)
s_b = np.linspace(0, np.percentile(s, perc), n_regions)
regions, weights = {}, np.ones(len(u))
for i in range(n_regions):
if i == 0:
region = intersect(np.where(u < u_b[i + 1]),
np.where(s < s_b[i + 1]))
elif i < n_regions - 1:
lower_cut = union(np.where(u > u_b[i]), np.where(s > s_b[i]))
upper_cut = intersect(np.where(u < u_b[i + 1]),
np.where(s < s_b[i + 1]))
region = intersect(lower_cut, upper_cut)
else:
region = union(
np.where(u > u_b[i]),
np.where(s > s_b[i])) # lower_cut for last region
regions[i] = region
if len(region) > 0:
weights[region] = n_regions / len(region)
# set weights accordingly such that each region has an equal overall contribution.
self.weights = weights * len(u) / np.sum(weights)
self.u_b, self.s_b = u_b, s_b
def compute_dice_coef(seg1, seg2, labelset=None):
if labelset is None:
lbset = np.union(np.unique(seg1), np.unique(seg2))
else:
lbset = np.asarray(labelset, dtype=int)
seg1.flat[~np.in1d(seg1, lbset)] = -1
seg2.flat[~np.in1d(seg2, lbset)] = -1
dicecoef = {}
for label in lbset:
l1 = (seg1==label)
l2 = (seg2==label)
d = 2*np.sum(l1&l2)/(1e-9 + np.sum(l1) + np.sum(l2))
dicecoef[label] = d
return dicecoef
def compute_dice_coef(seg1, seg2, labelset=None):
if labelset is None:
lbset = np.union(np.unique(seg1), np.unique(seg2))
else:
lbset = np.asarray(labelset, dtype=int)
seg1.flat[~np.in1d(seg1, lbset)] = -1
seg2.flat[~np.in1d(seg2, lbset)] = -1
dicecoef = {}
for label in lbset:
l1 = (seg1 == label)
l2 = (seg2 == label)
d = 2 * np.sum(l1 & l2) / (1e-9 + np.sum(l1) + np.sum(l2))
dicecoef[label] = d
return dicecoef
def aprioriGen(freqSets, k):
# generate candidate 2-itemsets
if k == 2:
Ck = np.array(
[list(comb) for comb in itertools.combinations(freqSets, 2)])
else:
# generate candidate k-itemsets (k > 2)
Ck = []
for i in range(0, np.shape(freqSets)[1]):
for j in range(i + 1, np.shape(freqSets)[1]):
# Merge a pair of (k-1) frequent itemsets if k-2 items are identical using Fk-1 x Fk-1 method
L1 = np.sort(freqSets[i, 0:k - 2])
L2 = np.sort(freqSets[j, 0:k - 2])
if np.isequal(L1, L2):
Ck = np.concatenate(
(Ck, np.union(freqSets[i, :], freqSets[j, :])), axis=0)
return (Ck)
def compute_dice_per_slice(seg1, seg2, labelset=None, axis=0):
if labelset is None:
lbset = np.union(np.unique(seg1), np.unique(seg2))
else:
lbset = np.asarray(labelset, dtype=int)
seg1.flat[~np.in1d(seg1, lbset)] = -1
seg2.flat[~np.in1d(seg2, lbset)] = -2
nslice = seg1.shape[axis]
s = [slice(None) for d in range(seg1.ndim)]
dcoefs = {}
for i in range(nslice):
s[axis] = i
s1 = seg1[s]
s2 = seg2[s]
n1 = np.sum(s1 >= 0)
n2 = np.sum(s2 >= 0)
if n1 == 0 and n2 == 0:
continue
d = 2. * np.sum(s1 == s2) / float(n1 + n2)
dcoefs[i] = d
return dcoefs
def compute_dice_per_slice(seg1, seg2, labelset=None, axis=0):
if labelset is None:
lbset = np.union(np.unique(seg1), np.unique(seg2))
else:
lbset = np.asarray(labelset, dtype=int)
seg1.flat[~np.in1d(seg1, lbset)] = -1
seg2.flat[~np.in1d(seg2, lbset)] = -2
nslice = seg1.shape[axis]
s = [slice(None) for d in range(seg1.ndim)]
dcoefs = {}
for i in range(nslice):
s[axis] = i
s1 = seg1[s]
s2 = seg2[s]
n1 = np.sum(s1>=0)
n2 = np.sum(s2>=0)
if n1==0 and n2==0:
continue
d = 2.*np.sum(s1==s2)/float(n1 + n2)
dcoefs[i] = d
return dcoefs
