def plot_histogram_tissue_radii(self, **kwargs):
style_scheme = styles.StyleSchemeCOSH()
x_range = (0, 40)
bin_width = 2
bar_fraction = 0.7
bar_width = 0.5 * bin_width * bar_fraction
bins = np.array(
np.linspace(x_range[0], x_range[1],
(x_range[1] - x_range[0]) / bin_width + 1))
ax = plt.gca()
func_radii = 1e6 * np.array(self.results.functional_tissue_radii())
topol_radii = 1e6 * np.array(self.results.topological_tissue_radii())
functional_count, _ = np.histogram(func_radii, bins=bins)
topological_count, _ = np.histogram(topol_radii, bins=bins)
x1 = 0.5 * (bins[:-1] + bins[1:]) - 0.5 * bar_width
x2 = 0.5 * (bins[:-1] + bins[1:]) + 0.5 * bar_width
plt.bar(x1,
functional_count,
bar_width,
edgecolor=style_scheme['int_func_radii_edge_color'],
facecolor=style_scheme['int_func_radii_face_color'],
linewidth=0.4,
label='functional')
plt.bar(x2,
topological_count,
bar_width,
edgecolor=style_scheme['int_topol_radii_edge_color'],
facecolor=style_scheme['int_topol_radii_face_color'],
linewidth=0.4,
label='geometric')
# plot mean and standard deviation
y_max = np.max(np.hstack((functional_count, topological_count)))
ax.errorbar(np.mean(func_radii),
1.1 * y_max,
xerr=np.std(func_radii),
fmt='o',
markersize=2,
linewidth=1,
elinewidth=0.5,
capsize=2,
capthick=0.5,
color=style_scheme['int_func_radii_face_color'])
ax.errorbar(np.mean(topol_radii),
1.05 * y_max,
xerr=np.std(topol_radii),
fmt='o',
markersize=2,
linewidth=1,
elinewidth=0.5,
capsize=2,
capthick=0.5,
color=style_scheme['int_topol_radii_edge_color'])
ax.set_xlim(x_range)
majorFormatter = FormatStrFormatter('%d')
ax.yaxis.set_major_formatter(majorFormatter)
styles.create_COSH_legend(ax, bbox_to_anchor=(1, 0.90))
setXLabel('\mathrm{tissue\;radius}\;r_t', 'um')
setYLabel('\mathrm{frequency}', '')
def plotStdHbProfiles(plotter):
plotter.plotSimStdHbProfile()
plotter.plotODEStdHbProfile()
plotter.plotODELinearizedStdHbProfile()
plotter.plotNoModelStdHbProfile()
ax = plt.gca()
create_COSH_legend(ax, bbox_to_anchor=(0.99, 0.82))
ax.yaxis.set_major_locator(MultipleWithMaxNLocator(0.01, 6))
ax.set_xlim(ax.xaxis.get_data_interval())
def plot_histogram_tissue_radii_multipanel(self, functional_colors,
functional_labels, **kwargs):
f, axarr = plt.subplots(len(self.results_list) + 2, sharex=True)
panel_annotation = 'ABCD'
style_scheme = styles.StyleSchemeCOSH()
x_range = (0, 40)
bin_width = 2
bar_fraction = 0.7
bar_width = bin_width * bar_fraction
bins = np.array(
np.linspace(x_range[0], x_range[1],
(x_range[1] - x_range[0]) / bin_width + 1))
bin_center = 0.5 * (bins[:-1] + bins[1:])
y_max = 0.0
hist_axes = axarr[0:-1]
# boxplot
data = [
1e6 * np.array(self.results_list[0].topological_tissue_radii()),
1e6 * np.array(self.results_list[1].functional_tissue_radii()),
1e6 * np.array(self.results_list[0].functional_tissue_radii())
]
labels = 'CBA'
medianprops = {'color': 'w'}
flierprops = {'marker': '.', 'markersize': 2}
bp = axarr[-1].boxplot(data,
whis=[5, 95],
vert=False,
labels=labels,
patch_artist=True,
medianprops=medianprops,
showfliers=False,
flierprops=flierprops)
fill_colors = [
style_scheme['int_topol_radii_face_color'], functional_colors[1],
functional_colors[0]
]
for patch, color in zip(bp['boxes'], fill_colors):
patch.set_facecolor(color)
# histograms
for i, (results, ax) in enumerate(zip(self.results_list, hist_axes)):
func_radii = 1e6 * np.array(results.functional_tissue_radii())
functional_count, _ = np.histogram(func_radii, bins=bins)
y_max = max(y_max, np.max(functional_count))
x_func = bin_center
ax.bar(x_func,
functional_count,
bar_width,
edgecolor='k',
facecolor=functional_colors[i],
linewidth=0.4,
label=functional_labels[i])
x_topol = bin_center
topol_radii = 1e6 * np.array(
self.results_list[0].topological_tissue_radii())
topological_count, _ = np.histogram(topol_radii, bins=bins)
axarr[-2].bar(x_topol,
topological_count,
bar_width,
edgecolor=style_scheme['int_topol_radii_edge_color'],
facecolor=style_scheme['int_topol_radii_face_color'],
linewidth=0.4,
label='geometric')
y_max = max(y_max, np.max(topological_count))
# plot mean and standard deviation
y_error_bar = 1.04 * y_max
for i, (results, ax) in enumerate(zip(self.results_list, hist_axes)):
func_radii = 1e6 * np.array(results.functional_tissue_radii())
ax.errorbar(np.mean(func_radii),
y_error_bar,
xerr=np.std(func_radii),
fmt='o',
markersize=2,
linewidth=1,
elinewidth=0.5,
capsize=2,
capthick=0.5,
color=functional_colors[i])
axarr[-2].errorbar(np.mean(topol_radii),
y_error_bar,
xerr=np.std(topol_radii),
fmt='o',
markersize=2,
linewidth=1,
elinewidth=0.5,
capsize=2,
capthick=0.5,
color=style_scheme['int_topol_radii_edge_color'])
axarr[0].set_xlim(x_range[0] - 1, x_range[1] + 1)
y_lim_max = 1.1 * y_max
setXLabel('\mathrm{tissue\;radius}\;r_t', 'um')
for ax in hist_axes:
majorLocator = MultipleLocator(10)
ax.yaxis.set_major_locator(majorLocator)
ax.set_ylim((0, y_lim_max))
styles.create_COSH_legend(ax,
loc='upper left',
bbox_to_anchor=(0.01, 0.92))
setYLabel('\mathrm{frequency}', '', ax=ax)
for ax, annotation in zip(axarr, panel_annotation):
annotate_axis_corner(ax, annotation, xy=(0.94, 0.82))
def plot_histogram_tissue_radii(self, functional_colors, functional_labels,
**kwargs):
ax = plt.gca()
style_scheme = styles.StyleSchemeCOSH()
n_bar_types = 1 + len(self.results_list)
x_range = (0, 40)
bin_width = 2
bar_fraction = 0.7
bar_width = bin_width * bar_fraction / n_bar_types
bins = np.array(
np.linspace(x_range[0], x_range[1],
(x_range[1] - x_range[0]) / bin_width + 1))
bin_center = 0.5 * (bins[:-1] + bins[1:])
y_max = 0.0
for i, results in enumerate(self.results_list):
func_radii = 1e6 * np.array(results.functional_tissue_radii())
functional_count, _ = np.histogram(func_radii, bins=bins)
y_max = max(y_max, np.max(functional_count))
offset = (-0.5 +
(i + 0.5) / n_bar_types) * bin_width * bar_fraction
x_func = bin_center + offset
plt.bar(x_func,
functional_count,
bar_width,
edgecolor='k',
facecolor=functional_colors[i],
linewidth=0.4,
label=functional_labels[i])
offset = (0.5 - 0.5 / n_bar_types) * bin_width * bar_fraction
x_topol = bin_center + offset
topol_radii = 1e6 * np.array(
self.results_list[0].topological_tissue_radii())
topological_count, _ = np.histogram(topol_radii, bins=bins)
plt.bar(x_topol,
topological_count,
bar_width,
edgecolor=style_scheme['int_topol_radii_edge_color'],
facecolor=style_scheme['int_topol_radii_face_color'],
linewidth=0.4,
label='geometric')
# plot mean and standard deviation
y_max = max(y_max, np.max(topological_count))
for i, results in enumerate(self.results_list):
func_radii = 1e6 * np.array(results.functional_tissue_radii())
y_error_bar = (1.1 - i / (len(self.results_list)) * 0.07) * y_max
ax.errorbar(np.mean(func_radii),
y_error_bar,
xerr=np.std(func_radii),
fmt='o',
markersize=2,
linewidth=1,
elinewidth=0.5,
capsize=2,
capthick=0.5,
color=functional_colors[i])
y_error_bar = 1.03 * y_max
ax.errorbar(np.mean(topol_radii),
y_error_bar,
xerr=np.std(topol_radii),
fmt='o',
markersize=2,
linewidth=1,
elinewidth=0.5,
capsize=2,
capthick=0.5,
color=style_scheme['int_topol_radii_edge_color'])
ax.set_xlim(x_range)
styles.create_COSH_legend(ax,
bbox_to_anchor=(0.48, 0.85),
handlelength=2.3,
handleheight=0.4)
setXLabel('\mathrm{tissue\;radius}\;r_t', 'um')
setYLabel('\mathrm{frequency}', '')
def plot_compared_hb_distal_distribution_multipanel(self, bin_width=0.02,
panel_annotation=('A', 'B', 'C')):
"""
Plot a comparison between the simulated and the integrated hemoglobin distribution.
Args:
bin_width (float): bin width for histogram
panel_annotation (sequence): panel annotations
"""
f, axarr = plt.subplots(4, sharex=True)
hist_axes = axarr[0:3]
bp_ax = axarr[-1]
sim_hb = self.postprocessor.distal_hb()
# compute integrated distal hemoglobin distributions
self.postprocessor.integrator.use_topological_radii = True
self.postprocessor.integrator.compute()
int_topol_hb, topol_weights = self.postprocessor.integrator.distal_hb_distribution()
int_topol_hb_repeat = self.postprocessor.integrator.distal_hb_distribution_with_repeat()
int_topol_mean = self.postprocessor.integrator.average_distal_hb()
int_topol_std = self.postprocessor.integrator.std_distal_hb()
self.postprocessor.integrator.use_topological_radii = False
self.postprocessor.integrator.compute()
int_func_hb, func_weights = self.postprocessor.integrator.distal_hb_distribution()
int_func_hb_repeat = self.postprocessor.integrator.distal_hb_distribution_with_repeat()
int_func_mean = self.postprocessor.integrator.average_distal_hb()
int_func_std = self.postprocessor.integrator.std_distal_hb()
# plot histograms
bounds = rounded_bounds(np.hstack((sim_hb, int_topol_hb, int_func_hb)), bin_width)
bins = np.arange(bounds[0], bounds[1]+0.5*bin_width, bin_width)
simul_count, _ = np.histogram(sim_hb, bins=bins, normed=True)
int_topol_count, _ = np.histogram(int_topol_hb, weights=topol_weights,
bins=bins, normed=True)
int_func_count, _ = np.histogram(int_func_hb, weights=func_weights,
bins=bins, normed=True)
bar_fraction = 0.8
bar_width = bin_width*bar_fraction
x = 0.5*(bins[:-1] + bins[1:])
hist_axes[0].bar(x, simul_count, bar_width,
edgecolor=style_scheme['simul_radii_edge_color'],
facecolor=style_scheme['simul_radii_face_color'],
linewidth=0.3,
label='moving RBCs')
hist_axes[1].bar(x, int_func_count, bar_width,
edgecolor=style_scheme['int_func_radii_edge_color'],
facecolor=style_scheme['int_func_radii_face_color'],
linewidth=0.3,
label='ODE (functional)')
hist_axes[2].bar(x, int_topol_count, bar_width,
edgecolor=style_scheme['int_topol_radii_edge_color'],
facecolor=style_scheme['int_topol_radii_face_color'],
linewidth=0.3,
label='ODE (geometric)')
xlim = max(0.0, round_to_base(axarr[0].get_xlim()[0], 0.1, func=np.floor)), \
axarr[0].get_xlim()[1]
axarr[0].set_xlim(xlim)
# plot mean and standard deviation
y_max = np.max(np.hstack((simul_count, int_topol_count, int_func_count)))
sim_mean = np.mean(self.postprocessor.distal_hb())
sim_std = np.std(self.postprocessor.distal_hb())
y_error_bar = 1.05*y_max
hist_axes[0].errorbar(sim_mean, y_error_bar, xerr=sim_std,
fmt='o', markersize=2, linewidth=1,
elinewidth=0.5,
capsize=2, capthick=0.5,
color=style_scheme['simul_radii_face_color'])
hist_axes[1].errorbar(int_func_mean, y_error_bar, xerr=int_func_std,
fmt='o', markersize=2, linewidth=1,
elinewidth=0.5,
capsize=2, capthick=0.5,
color=style_scheme['int_func_radii_face_color'])
hist_axes[2].errorbar(int_topol_mean, y_error_bar, xerr=int_topol_std,
fmt='o', markersize=2, linewidth=1,
elinewidth=0.5,
capsize=2, capthick=0.5,
color=style_scheme['int_topol_radii_edge_color'])
setXLabel('\mathrm{outflow\;saturation}\;S_v', '')
axarr[0].xaxis.set_major_locator(MultipleLocator(0.1))
for ax, annotation in zip(hist_axes, panel_annotation):
styles.create_COSH_legend(ax, loc='upper left')
ax.yaxis.set_major_locator(MultipleLocator(5))
# ax.yaxis.set_major_locator(MultipleLocator(10))
setYLabel('f_{S_v}', '', ax=ax)
for ax, annotation in zip(axarr, panel_annotation):
annotate_axis_corner(ax, annotation, xy=(0.956, 0.79))
# if annotation != 'C':
# annotate_axis_corner(ax, annotation, xy=(0.956, 0.79))
# else:
# annotate_axis_corner(ax, annotation, xy=(0.956, 0.72))
# boxplot
data = [int_topol_hb_repeat, int_func_hb_repeat, sim_hb]
offset = len(panel_annotation) - len(data)
labels = panel_annotation[-1-offset::-1]
medianprops = {'color': 'w'}
flierprops = {'marker': '.', 'markersize': 2}
bp = bp_ax.boxplot(data, whis=[5, 95], vert=False, labels=labels,
patch_artist=True, medianprops=medianprops, flierprops=flierprops,
showfliers=False)
fill_colors = [style_scheme['int_topol_radii_edge_color'],
style_scheme['int_func_radii_face_color'],
style_scheme['simul_radii_face_color']]
for patch, color in zip(bp['boxes'], fill_colors):
patch.set_facecolor(color)
print "5th percentile of topological: ", np.percentile(int_topol_hb_repeat, 5)
print "First quartile of topological: ", np.percentile(int_topol_hb_repeat, 25)
print "Median of topological: ", np.percentile(int_topol_hb_repeat, 50)
print "Third quartile of topological: ", np.percentile(int_topol_hb_repeat, 75)
print "95th percentile of topological: ", np.percentile(int_topol_hb_repeat, 95)
# statistical testing
F = int_topol_std**2/sim_std**2
pvalue = scipy.stats.f.sf(F, np.size(int_topol_hb) - 1, np.size(sim_hb) - 1)
print "p-value for ODE model (topological radii) vs. moving RBC: {:g}".format(pvalue)
F = int_func_std**2/sim_std**2
pvalue = scipy.stats.f.sf(F, np.size(int_func_hb) - 1, np.size(sim_hb) - 1)
print "p-value for ODE model (functional radii) vs. moving RBC: {:g}".format(pvalue)
F = int_topol_std**2/int_func_std**2
pvalue = scipy.stats.f.sf(F, np.size(int_topol_hb) - 1, np.size(int_func_hb) - 1)
print "p-value for ODE model, topological vs. functional radii: {:g}".format(pvalue)
def plot_compared_hb_distal_distribution(self, bin_width=0.02):
"""
Plot a comparison between the simulated and the integrated hemoglobin distribution.
Args:
bin_width (float): bin width for histogram
"""
ax = plt.gca()
sim_hb = self.postprocessor.distal_hb()
# compute integrated distal hemoglobin distributions
self.postprocessor.integrator.use_topological_radii = True
self.postprocessor.integrator.compute()
int_topol_hb, topol_weights = self.postprocessor.integrator.distal_hb_distribution()
int_topol_mean = self.postprocessor.integrator.average_distal_hb()
int_topol_std = self.postprocessor.integrator.std_distal_hb()
self.postprocessor.integrator.use_topological_radii = False
self.postprocessor.integrator.compute()
int_func_hb, func_weights = self.postprocessor.integrator.distal_hb_distribution()
int_func_mean = self.postprocessor.integrator.average_distal_hb()
int_func_std = self.postprocessor.integrator.std_distal_hb()
# plot histograms
bounds = rounded_bounds(np.hstack((sim_hb, int_topol_hb, int_func_hb)), bin_width)
bins = np.arange(bounds[0], bounds[1]+0.5*bin_width, bin_width)
bar_fraction = 0.7
bar_width = 1./3.*bin_width*bar_fraction
simul_count, _ = np.histogram(sim_hb, bins=bins, normed=True)
int_topol_count, _ = np.histogram(int_topol_hb, weights=topol_weights,
bins=bins, normed=True)
int_func_count, _ = np.histogram(int_func_hb, weights=func_weights,
bins=bins, normed=True)
x1 = 0.5*(bins[:-1] + bins[1:]) - bar_width
x2 = 0.5*(bins[:-1] + bins[1:])
x3 = 0.5*(bins[:-1] + bins[1:]) + bar_width
ax.bar(x1, simul_count, bar_width,
edgecolor=style_scheme['simul_radii_edge_color'],
facecolor=style_scheme['simul_radii_face_color'],
linewidth=0.3,
label='moving RBCs')
ax.bar(x2, int_func_count, bar_width,
edgecolor=style_scheme['int_func_radii_edge_color'],
facecolor=style_scheme['int_func_radii_face_color'],
linewidth=0.3,
label='ODE (functional)')
ax.bar(x3, int_topol_count, bar_width,
edgecolor=style_scheme['int_topol_radii_edge_color'],
facecolor=style_scheme['int_topol_radii_face_color'],
linewidth=0.3,
label='ODE (geometric)')
xlim = max(0.0, round_to_base(ax.get_xlim()[0], 0.1, func=np.floor)), ax.get_xlim()[1]
ax.set_xlim(xlim)
# plot mean and standard deviation
y_max = np.max(np.hstack((simul_count, int_topol_count, int_func_count)))
sim_mean = np.mean(self.postprocessor.distal_hb())
sim_std = np.std(self.postprocessor.distal_hb())
ax.errorbar(sim_mean, 1.1*y_max, xerr=sim_std,
fmt='o', markersize=2, linewidth=1,
elinewidth=0.5,
capsize=2, capthick=0.5,
color=style_scheme['simul_radii_face_color'])
ax.errorbar(int_func_mean, 1.06*y_max, xerr=int_func_std,
fmt='o', markersize=2, linewidth=1,
elinewidth=0.5,
capsize=2, capthick=0.5,
color=style_scheme['int_func_radii_face_color'])
ax.errorbar(int_topol_mean, 1.02*y_max, xerr=int_topol_std,
fmt='o', markersize=2, linewidth=1,
elinewidth=0.5,
capsize=2, capthick=0.5,
color=style_scheme['int_topol_radii_edge_color'])
styles.create_COSH_legend(ax, bbox_to_anchor=(0.45, 0.95))
setXLabel('\mathrm{outflow\;saturation}\;S_v', '')
setYLabel('\mathrm{probability\;density}', '')
ax.xaxis.set_major_locator(MultipleLocator(0.1))
ax.yaxis.set_major_locator(MultipleLocator(5))
plt.clf()
else:
plotter = DiffusiveInteractionComparisonPlotter(postprocessor, model_names)
if args.multipanel:
panel_layout = (2, 1) if not args.panelLayout else args.panelLayout
fig, axs = plt.subplots(*panel_layout, sharex=True)
plt.sca(axs[0])
plotter.plotHbProfiles(plot_mean=plot_mean)
annotate_axis_corner(axs[0], 'A')
if panel_layout[1] == 1:
axs[0].set_xlabel('')
plt.sca(axs[1])
plotter.plotHbDiffProfiles(exponential_fit=exponential_fit,
linearized_ode=linearized_ode)
# legend = create_COSH_legend(axs[1], bbox_to_anchor=(1, 0.86))
legend = create_COSH_legend(axs[1], bbox_to_anchor=(0.5, 0.86))
annotate_axis_corner(axs[1], 'B')
figOptions.saveFig('plotDiffusiveInteractionMultipanel' +
plot_name_suffix)
else:
plotter.plotHbProfiles()
plotter.plotHbDiffProfiles(exponential_fit=exponential_fit)
figOptions.saveFig('plotDiffusiveInteraction' + plot_name_suffix)
plt.clf()
for model_name, modelPlotter in plotter.modelPlotters.iteritems():
modelPlotter.plotOutfluxes()
figOptions.saveFig('plotOutfluxes')
plt.clf()
for model_name, modelPlotter in plotter.modelPlotters.iteritems():
modelPlotter.plotTissueRadii()
style.update(style_scheme['MVN1'])
style.update({'label': 'MVN 1'})
elif '20150715' in case_path or 'MVN2' in case_path:
style.update(style_scheme['MVN2'])
style.update({'label': 'MVN 2'})
plt.clf()
combined_plotter.plot_histogram_tissue_radii()
fig_options.saveFig('plotHistogramTissueRadii')
plt.clf()
for plotter, style in zip(plotters, styles):
plotter.plot_functional_vs_topological_tissue_radii(**style)
cosh_styles.create_COSH_legend(plt.gca(),
loc='upper left',
markerscale=0.7,
handletextpad=0,
handlelength=2.5)
fig_options.saveFig('plotFunctionalVsTopologicalTissueRadii')
plt.clf()
for plotter, style in zip(plotters, styles):
plotter.plot_tissue_radii_vs_hb_mean(**style)
cosh_styles.create_COSH_legend(plt.gca(),
loc='upper left',
markerscale=0.8,
handletextpad=0,
handlelength=2.5)
fig_options.saveFig('plotTissueRadiiVsHbMean')
plt.clf()