# Segment image
Y_h, post, theta = UGM.segment(X)
# Set cluster assignments to correct tissue labels
Y_h = set_classes(Y_h, z)
# Compute error
err[1, n, r] = np.mean(Y_h[M] != Y[M])
dcc[1, n, r] = dice(Y_h[M], Y[M])
if vis:
fn_segs = fn + 'UGM_segs_p' + str(n + 1) + '_r' + str(r +
1) + '.png'
plot_clustering(X[:, :, 0], Y_h, mode='subpixel', savefn=fn_segs)
''' Semi-supervised Gaussian Mixture '''
# Initialize model
SGM = SemisupervisedGaussianMixture(num_components=K,
num_channels=D,
max_iter=max_iter,
tol=x_tol,
init_params='nn')
# Segment image
Y_h, post, theta = SGM.segment(X, y)
# Compute error
err[2, n, r] = np.mean(Y_h[M] != Y[M])
dcc[2, n, r] = dice(Y_h[M], Y[M])
# Set cluster assignments to correct tissue labels
Y_h = set_classes(Y_h, z)
# Compute error
err[0, n, r] = np.mean(Y_h[M] != Y[M])
dcc[0, n, r] = dice(Y_h[M], Y[M])
if vis:
fn_segs = fn + 'SCK_segs' + str(n + 1) + '_r' + str(r + 1) + '.png'
plot_segmentation(Y_h[30:-30, 30:-30], savefn=fn_segs)
fn_segs = fn + 'SCK_segl' + str(n + 1) + '_r' + str(r + 1) + '.png'
plot_clustering(X[30:-30, 30:-30, 0],
Y_h[30:-30, 30:-30],
mode='subpixel',
savefn=fn_segs)
''' Unsupervised Gaussian Mixture '''
# Initialize model
UGM = UnsupervisedGaussianMixture(num_components=K,
num_channels=D,
max_iter=max_iter,
tol=x_tol,
init_params='kmeans')
# Segment image
Y_h, post, theta = UGM.segment(X)
# Set cluster assignments to correct tissue labels
Y_h = set_classes(Y_h, z)