def evaluateQuality():
data_dir = path.join(path.dirname(path.realpath(__file__)),
"../example_data/")
df_path = path.join(data_dir, "f_reg_quality.csv")
registration_qc(
df_path,
value={"similarity": "Similarity"},
group={"subject": "Subject"},
repeat={"session": "Session"},
extra={"acquisition": "Acquisition"},
save_as="registration_qc.png",
show=False,
)
df = pd.read_csv(df_path)
#model="similarity ~ C(acquisition) + C(session) + C(subject)"
model = "similarity ~ C(session) + C(subject)"
regression_model = smf.ols(model, data=df).fit()
anova_summary = sm.stats.anova_lm(regression_model, typ=2)
print(inline_anova(anova_summary, "C(subject)", style="python"))
print(inline_anova(anova_summary, "C(session)", style="python", max_len=2))
#print(inline_anova(anova_summary,"C(acquisition)",style="python"))
print(inline_anova(anova_summary, "C(subject)", style="tex"))
print(inline_anova(anova_summary, "C(session)", style="tex", max_len=2))
#print(inline_anova(anova_summary,"C(acquisition)",style="tex"))
# plotting variance results for different metrics
# TODO: warum MI = 0?
metrics = ['CC', 'MI', 'GC']
means = []
stds = []
for metric in metrics:
df_path = path.join('/media/nexus/storage/coding/github/SAMRI/',
"f_reg_quality_" + metric + ".csv")
df = pd.read_csv(df_path)
variances = df.groupby('subject').agg('var')
std = sqrt(variances['similarity'].sum())
mean = df.groupby('subject').agg('mean')['similarity'].sum()
means.append(mean)
stds.append(std)
print(metrics)
print(means)
print(stds)
return metrics, means, stds
def reg_gc():
from samri.plotting.aggregate import registration_qc
from samri.report.registration import iter_measure_sim
from samri.typesetting import inline_anova
df = iter_measure_sim(
"~/ni_data/ofM.dr/preprocessing/composite",
"~/.samri_files/templates/mouse/DSURQE/DSURQEc_200micron_average.nii",
#modality="anat",
metric="GC",
radius_or_number_of_bins=0,
sampling_strategy="Regular",
sampling_percentage=0.5,
save_as=False,
)
anova_summary = registration_qc(
df,
value={"similarity": "Similarity"},
group={"sub": "Subject"},
repeat={"ses": "Session"},
show=False,
save_as="/tmp/f_reg_gc.png",
)
subject_effect = inline_anova(anova_summary, "C(Subject)", style="python")
print("Subject Main Effect: {}".format(subject_effect))
session_effect = inline_anova(anova_summary, "C(Session)", style="python")
print("Session Main Effect: {}".format(session_effect))
def reg_gc():
from samri.plotting.aggregate import registration_qc
from samri.report.registration import iter_measure_sim
from samri.typesetting import inline_anova
df = iter_measure_sim(
"~/ni_data/ofM.dr/preprocessing/composite",
"~/.samri_files/templates/mouse/DSURQE/DSURQEc_200micron_average.nii",
#modality="anat",
metric="GC",
radius_or_number_of_bins = 0,
sampling_strategy = "Regular",
sampling_percentage=0.5,
save_as=False,
)
anova_summary = registration_qc(df,
value={"similarity":"Similarity"},
group={"sub":"Subject"},
repeat={"ses":"Session"},
show=False,
save_as='/tmp/f_reg_gc.png',
)
subject_effect = inline_anova(anova_summary,"C(Subject)",style="python")
print("Subject Main Effect: {}".format(subject_effect))
session_effect = inline_anova(anova_summary,"C(Session)",style="python")
print("Session Main Effect: {}".format(session_effect))
def test_autofind():
"""We may be able to turn this into a CI function, if we put together a data fetching script for dummy (empty) BIDS-formatted data.
"""
from samri.utilities import bids_autofind
from samri.plotting.aggregate import registration_qc
from samri.report.registration import get_scores
from samri.typesetting import inline_anova
from samri.utilities import bids_substitution_iterator
path_template, substitutions = bids_autofind("~/ni_data/ofM.dr/preprocessing/composite","func")
df = get_scores(path_template, substitutions,
"~/ni_data/templates/DSURQEc_200micron_average.nii",
metric="CC",
radius_or_number_of_bins=10,
sampling_strategy="Regular",
sampling_percentage=0.33,
save_as=False,
)
anova_summary = registration_qc(df,
value={"similarity":"Similarity"},
group={"sub":"Subject"},
repeat={"ses":"Session"},
show=False,
save_as="/tmp/f_reg_cc10.png",
)
subject_effect = inline_anova(anova_summary,"C(Subject)",style="python")
print("Subject Main Effect: {}".format(subject_effect))
session_effect = inline_anova(anova_summary,"C(Session)",style="python")
print("Session Main Effect: {}".format(session_effect))
def reg_cc(
radius=5,
autofind=True,
):
from samri.utilities import bids_autofind
from samri.plotting.aggregate import registration_qc
from samri.report.registration import get_scores
from samri.typesetting import inline_anova
from samri.utilities import bids_substitution_iterator
if autofind:
path_template, substitutions = bids_autofind(
"~/ni_data/ofM.dr/preprocessing/composite", "func")
else:
substitutions = bids_substitution_iterator(
["ofM", "ofM_aF", "ofM_cF1", "ofM_cF2", "ofM_pF"],
[
"4001", "4007", "4008", "4011", "5692", "5694", "5699", "5700",
"5704", "6255", "6262"
],
["EPI_CBV_chr_longSOA", "EPI_CBV_jb_long"],
"~/ni_data/ofM.dr/",
"composite",
)
path_template = "~/ni_data/ofM.dr/preprocessing/{preprocessing_dir}/sub-{subject}/ses-{session}/func/sub-{subject}_ses-{session}_trial-{trial}.nii.gz"
df = get_scores(
path_template,
substitutions,
"~/ni_data/templates/DSURQEc_200micron_average.nii",
metric="CC",
radius_or_number_of_bins=radius,
sampling_strategy="Regular",
sampling_percentage=0.33,
save_as="f_reg_quality.csv",
)
anova_summary = registration_qc(
df,
value={"similarity": "Similarity"},
group={"sub": "Subject"},
repeat={"ses": "Session"},
extra={"trial": "Type"},
model="{value} ~ C({extra}) + C({repeat}) + C({group}) -1",
save_as="registration_qc.png",
print_model=True,
print_anova=True,
show=False,
)
subject_effect = inline_anova(anova_summary, "C(Subject)", style="python")
print("Subject Main Effect: {}".format(subject_effect))
session_effect = inline_anova(anova_summary,
"C(Session)",
style="python",
max_len=2)
print("Session Main Effect: {}".format(session_effect))
type_effect = inline_anova(anova_summary, "C(Type)", style="python")
print("Scan Type Main Effect: {}".format(type_effect))
from os import path
import matplotlib.pyplot as plt
from samri.plotting.aggregate import registration_qc
from samri.typesetting import inline_anova
this_path = path.dirname(path.realpath(__file__))
data_dir = path.join(this_path, "../../example_data/")
df_path = path.join(data_dir, "f_reg_quality.csv")
plt.style.use(path.join(this_path, "dark.conf"))
registration_qc(
df_path,
value={"similarity": "Similarity"},
group={"subject": "Subject"},
repeat={"session": "Session"},
extra={"acquisition": "Type"},
save_as="2017-09_aic-pom.pdf",
show=False,
samri_style=False,
extra_cmap=["#FFFFFF", "#000000"],
cmap=["#EE1111", "#11DD11", "#1111FF", "#CCCC22", "#AA11AA"],
values_rename={
"sub": "Subject",
"EPI_CBV_chr_longSOA": "longer",
"EPI_CBV_jb_long": "shorter",
},
)
import pandas as pd
import statsmodels.api as sm
import statsmodels.formula.api as smf
from os import path
from samri.plotting.aggregate import registration_qc
from samri.typesetting import inline_anova
data_dir = path.join(path.dirname(path.realpath(__file__)),
"../../example_data/")
df_path = path.join(data_dir, "f_reg_quality.csv")
registration_qc(
df_path,
value={"similarity": "Similarity"},
group={"subject": "Subject"},
repeat={"session": "Session"},
extra={"acquisition": "Acquisition"},
save_as="registration_qc.png",
show=False,
)
df = pd.read_csv(df_path)
#model="similarity ~ C(acquisition) + C(session) + C(subject)"
model = "similarity ~ C(session) + C(subject)"
regression_model = smf.ols(model, data=df).fit()
anova_summary = sm.stats.anova_lm(regression_model, typ=2)
print(inline_anova(anova_summary, "C(subject)", style="python"))
print(inline_anova(anova_summary, "C(session)", style="python", max_len=2))
#print(inline_anova(anova_summary,"C(acquisition)",style="python"))
print(inline_anova(anova_summary, "C(subject)", style="tex"))
from os import path
from samri.plotting.aggregate import registration_qc
from samri.typesetting import inline_anova
data_dir = path.join(path.dirname(path.realpath(__file__)),"../../example_data/")
df_path = path.join(data_dir,"f_reg_quality.csv")
anova_summary = registration_qc(df_path,
value={"similarity":"Similarity"},
group={"sub":"Subject"},
repeat={"ses":"Session"},
extra={"trial":"Type"},
model="{value} ~ C({extra}) + C({repeat}) + C({group}) -1",
save_as="registration_qc.png",
print_model=True,
print_anova=True,
show=False,
anova_type=2,
)
print(inline_anova(anova_summary,"C(Subject)",style="python"))
print(inline_anova(anova_summary,"C(Session)",style="python", max_len=2))
print(inline_anova(anova_summary,"C(Type)",style="python"))
print(inline_anova(anova_summary,"C(Subject)",style="tex"))
print(inline_anova(anova_summary,"C(Session)",style="tex", max_len=2))
print(inline_anova(anova_summary,"C(Type)",style="tex"))
data_dir = path.join(this_path, "../../example_data/")
df_path = path.join(data_dir, "f_reg_quality.csv")
plt.style.use(path.join(this_path, "dark.conf"))
anova_summary = registration_qc(
df_path,
value={"similarity": "Similarity"},
group={"sub": "Subject"},
repeat={"ses": "Session"},
extra={"trial": "Type"},
model="{value} ~ C({extra}) + C({repeat}) + C({group}) -1",
save_as="2017-09_aic-pom.pdf",
print_model=True,
print_anova=True,
show=False,
anova_type=3,
samri_style=False,
extra_cmap=["#FFFFFF", "#000000"],
cmap=["#EE1111", "#11DD11", "#1111FF", "#CCCC22", "#AA11AA"],
values_rename={
"sub": "Subject",
"EPI_CBV_chr_longSOA": "longer",
"EPI_CBV_jb_long": "shorter",
},
)
print(inline_anova(anova_summary, "C(Subject)", style="python"))
print(inline_anova(anova_summary, "C(Session)", style="python", max_len=2))
print(inline_anova(anova_summary, "C(Type)", style="python"))
print(inline_anova(anova_summary, "C(Subject)", style="tex"))
