def compute(self):
for nodeLabel in self.scoreProcedure.m_scoreDefinition.keys():
try:
proxLabel = self.scoreProcedure.m_scoreDefinition[nodeLabel][
"proximal"]
distLabel = self.scoreProcedure.m_scoreDefinition[nodeLabel][
"distal"]
proxNode = self.m_model.getSegment(proxLabel).getReferential(
'TF').getNodeTrajectory(nodeLabel)
distNode = self.m_model.getSegment(distLabel).getReferential(
'TF').getNodeTrajectory(nodeLabel)
score = numeric.rms((proxNode - distNode), axis=1)
scoreValues = np.array([
score,
np.zeros(self.acq.GetPointFrameNumber()),
np.zeros(self.acq.GetPointFrameNumber())
]).T
btkTools.smartAppendPoint(self.acq,
str(nodeLabel + "_Score"),
scoreValues,
PointType=btk.btkPoint.Scalar,
desc="Score")
except:
LOGGER.logger.error(
"[pyCGM2] Score residual for node (%s) not computed" %
(nodeLabel))
def compute(self):
for nodeLabel in self.scoreProcedure.m_scoreDefinition.keys():
proxLabel = self.scoreProcedure.m_scoreDefinition[nodeLabel][
"proximal"]
distLabel = self.scoreProcedure.m_scoreDefinition[nodeLabel][
"distal"]
proxNode = self.m_model.getSegment(
proxLabel).anatomicalFrame.getNodeTrajectory(nodeLabel)
distNode = self.m_model.getSegment(
distLabel).anatomicalFrame.getNodeTrajectory(nodeLabel)
score = numeric.rms((proxNode - distNode), axis=1)
scoreValues = np.array([
score,
np.zeros(self.acq.GetPointFrameNumber()),
np.zeros(self.acq.GetPointFrameNumber())
]).T
btkTools.smartAppendPoint(self.acq,
str(nodeLabel + "_Score"),
scoreValues,
PointType=btk.btkPoint.Scalar,
desc="Score")
def test_point_rms(acq, RefLabel, LabelToTest, threshold, init=-1, end=-1):
init = 0 if init == -1 else init - acq.GetFirstFrame()
end = acq.GetLastFrame() - acq.GetFirstFrame(
) if end == -1 else end - acq.GetFirstFrame()
np.testing.assert_array_less(
numeric.rms(acq.GetPoint(RefLabel).GetValues()[init:end, :] -
acq.GetPoint(LabelToTest).GetValues()[init:end, :],
axis=0), threshold)
def comparisonOpensimVsCGM(motFilename, acqIK, pointLabelSuffix):
# note : assert_almost_equal tests fail. Prefer rmsd computation
values = osimProcessing.mot2pointValues(
motFilename, ["hip_flexion_r", "hip_adduction_r", "hip_rotation_r"],
orientation=[1.0, 1.0, 1.0])
myValues = acqIK.GetPoint(str("RHipAngles" + "_" +
pointLabelSuffix)).GetValues()
#np.testing.assert_almost_equal( values,myValues, decimal =3)
np.testing.assert_array_less(numeric.rms((values - myValues), axis=0),
0.05)
values = osimProcessing.mot2pointValues(
motFilename, ["knee_flexion_r", "knee_adduction_r", "knee_rotation_r"],
orientation=[-1.0, 1.0, 1.0])
myValues = acqIK.GetPoint(str("RKneeAngles" + "_" +
pointLabelSuffix)).GetValues()
#np.testing.assert_almost_equal( values,myValues, decimal =3)
np.testing.assert_array_less(numeric.rms((values - myValues), axis=0),
0.05)
values = osimProcessing.mot2pointValues(
motFilename,
["ankle_flexion_r", "ankle_rotation_r", "ankle_adduction_r"],
orientation=[1.0, -1.0, 1.0])
myValues = acqIK.GetPoint(str("RAnkleAngles" + "_" +
pointLabelSuffix)).GetValues()
#np.testing.assert_almost_equal( values,myValues, decimal =3)
np.testing.assert_array_less(numeric.rms((values - myValues), axis=0),
0.05)
values = osimProcessing.mot2pointValues(
motFilename,
["forefoot_flexion_r", "forefoot_adduction_r", "forefoot_rotation_r"],
orientation=[1.0, 1.0, -1.0])
myValues = acqIK.GetPoint(str("RForeFootAngles" + "_" +
pointLabelSuffix)).GetValues()
#np.testing.assert_almost_equal( values,myValues, decimal =3)
np.testing.assert_array_less(numeric.rms((values - myValues), axis=0),
0.05)
def compareKinetics(acqGait, init, end, forceThreshold, momentThreshold,
powerThreshold):
forceArrayThreshold = np.array(
[forceThreshold, forceThreshold, forceThreshold])
np.testing.assert_array_less(
numeric.rms(
(acqGait.GetPoint("LHipForce").GetValues()[init:end, :] -
acqGait.GetPoint("LHipForce_cgm1_6dof").GetValues()[init:end, :]),
axis=0), forceArrayThreshold)
np.testing.assert_array_less(
numeric.rms((
acqGait.GetPoint("LKneeForce").GetValues()[init:end, :] -
acqGait.GetPoint("LKneeForce_cgm1_6dof").GetValues()[init:end, :]),
axis=0), forceArrayThreshold)
np.testing.assert_array_less(
numeric.rms(
(acqGait.GetPoint("LAnkleForce").GetValues()[init:end, :] -
acqGait.GetPoint("LAnkleForce_cgm1_6dof").GetValues()[init:end, :]
),
axis=0), forceArrayThreshold)
np.testing.assert_array_less(
numeric.rms(
(acqGait.GetPoint("RHipForce").GetValues()[init:end, :] -
acqGait.GetPoint("RHipForce_cgm1_6dof").GetValues()[init:end, :]),
axis=0), forceArrayThreshold)
np.testing.assert_array_less(
numeric.rms((
acqGait.GetPoint("RKneeForce").GetValues()[init:end, :] -
acqGait.GetPoint("RKneeForce_cgm1_6dof").GetValues()[init:end, :]),
axis=0), forceArrayThreshold)
np.testing.assert_array_less(
numeric.rms(
(acqGait.GetPoint("RAnkleForce").GetValues()[init:end, :] -
acqGait.GetPoint("RAnkleForce_cgm1_6dof").GetValues()[init:end, :]
),
axis=0), forceArrayThreshold)
momentArrayThreshold = np.array(
[momentThreshold, momentThreshold, momentThreshold])
np.testing.assert_array_less(
numeric.rms((
acqGait.GetPoint("RHipMoment").GetValues()[init:end, :] -
acqGait.GetPoint("RHipMoment_cgm1_6dof").GetValues()[init:end, :]),
axis=0), momentArrayThreshold)
np.testing.assert_array_less(
numeric.rms(
(acqGait.GetPoint("RKneeMoment").GetValues()[init:end, :] -
acqGait.GetPoint("RKneeMoment_cgm1_6dof").GetValues()[init:end, :]
),
axis=0), momentArrayThreshold)
np.testing.assert_array_less(
numeric.rms((
acqGait.GetPoint("RAnkleMoment").GetValues()[init:end, :] -
acqGait.GetPoint("RAnkleMoment_cgm1_6dof").GetValues()[init:end, :]
),
axis=0), momentArrayThreshold)
np.testing.assert_array_less(
numeric.rms((
acqGait.GetPoint("LHipMoment").GetValues()[init:end, :] -
acqGait.GetPoint("LHipMoment_cgm1_6dof").GetValues()[init:end, :]),
axis=0), momentArrayThreshold)
np.testing.assert_array_less(
numeric.rms(
(acqGait.GetPoint("LKneeMoment").GetValues()[init:end, :] -
acqGait.GetPoint("LKneeMoment_cgm1_6dof").GetValues()[init:end, :]
),
axis=0), momentArrayThreshold)
np.testing.assert_array_less(
numeric.rms((
acqGait.GetPoint("LAnkleMoment").GetValues()[init:end, :] -
acqGait.GetPoint("LAnkleMoment_cgm1_6dof").GetValues()[init:end, :]
),
axis=0), momentArrayThreshold)
powerThreshold = powerThreshold
np.testing.assert_array_less(
numeric.rms(
(acqGait.GetPoint("LHipPower").GetValues()[init:end, :] -
acqGait.GetPoint("LHipPower_cgm1_6dof").GetValues()[init:end, :]),
axis=0)[2], powerThreshold)
np.testing.assert_array_less(
numeric.rms((
acqGait.GetPoint("LKneePower").GetValues()[init:end, :] -
acqGait.GetPoint("LKneePower_cgm1_6dof").GetValues()[init:end, :]),
axis=0)[2], powerThreshold)
np.testing.assert_array_less(
numeric.rms(
(acqGait.GetPoint("LAnklePower").GetValues()[init:end, :] -
acqGait.GetPoint("LAnklePower_cgm1_6dof").GetValues()[init:end, :]
),
axis=0)[2], powerThreshold)
np.testing.assert_array_less(
numeric.rms(
(acqGait.GetPoint("RHipPower").GetValues()[init:end, :] -
acqGait.GetPoint("RHipPower_cgm1_6dof").GetValues()[init:end, :]),
axis=0)[2], powerThreshold)
np.testing.assert_array_less(
numeric.rms((
acqGait.GetPoint("RKneePower").GetValues()[init:end, :] -
acqGait.GetPoint("RKneePower_cgm1_6dof").GetValues()[init:end, :]),
axis=0)[2], powerThreshold)
np.testing.assert_array_less(
numeric.rms(
(acqGait.GetPoint("RAnklePower").GetValues()[init:end, :] -
acqGait.GetPoint("RAnklePower_cgm1_6dof").GetValues()[init:end, :]
),
axis=0)[2], powerThreshold)
def compute(self,analysis,normativeData):
gvs = dict()
nLeftCycles,nRightCycles = analysis.getKinematicCycleNumbers()
# --- MAP ---
# left cycles
for label,context in self.matchingNormativeDataLabel.keys():
matchingNormativeDataLabel = self.matchingNormativeDataLabel[label,context]
if context == "Left":
left_rms_local = np.zeros((nLeftCycles,3))
for i in range(0,nLeftCycles):
values = analysis.kinematicStats.data[label, context]["values"][i]
valuesNorm = normativeData[matchingNormativeDataLabel]["mean"]
if valuesNorm.shape[0] == 51:
rms = numeric.rms(values[0:101:2]-valuesNorm,axis=0)
else:
rms = numeric.rms(values-valuesNorm,axis=0)
left_rms_local[i,:] = rms
gvs[label,context]=left_rms_local
# right cycles
for label,context in self.matchingNormativeDataLabel.keys():
matchingNormativeDataLabel = self.matchingNormativeDataLabel[label,context]
if context == "Right":
right_rms_local = np.zeros((nRightCycles,3))
for i in range(0,nRightCycles):
values = analysis.kinematicStats.data[label, context]["values"][i]
valuesNorm = normativeData[matchingNormativeDataLabel]["mean"]
if valuesNorm.shape[0] == 51:
rms = numeric.rms(values[0:101:2]-valuesNorm,axis=0)
else:
rms = numeric.rms(values-valuesNorm,axis=0)
right_rms_local[i,:] = rms
gvs[label,context]=right_rms_local
# --- GPS ---
# number of axis. 15 according articles ( left )
n_axis =0
for axis in self.axes:
n_axis = n_axis + len(self.axes[axis])
# computation of rms for left cycles
left_rms_global = np.zeros((nLeftCycles,n_axis))
for i in range(0,nLeftCycles):
cumAxis =0
for label,context in gvs.keys():
if context == "Left":
axisIndex = self.axes[self.matchingNormativeDataLabel[label,context]] # use of the tip here
for axis in axisIndex:
left_rms_global[i,cumAxis] = gvs[label,context][i,axis]
cumAxis+=1
# computation of rms for right cycles
right_rms_global = np.zeros((nRightCycles,n_axis))
for i in range(0,nRightCycles):
cumAxis =0
for label,context in gvs.keys():
if context == "Right":
axisIndex = self.axes[self.matchingNormativeDataLabel[label,context]]
for axis in axisIndex:
right_rms_global[i,cumAxis] = gvs[label,context][i,axis]
cumAxis+=1
# output dictionnary
outDict_gvs = dict()
outDict_gps_context = dict()
outDict_gps = dict()
for label,context in self.matchingNormativeDataLabel.keys():
outDict_gvs[label,context]={'mean':np.mean(gvs[label, context],axis=0),
'std':np.std(gvs[label, context],axis=0),
'median': np.median(gvs[label, context],axis=0),
'values':gvs[label,context]}
for context in ["Left","Right"]:
gpsValues = left_rms_global if context == "Left" else right_rms_global
# construction of the global dictionnary outputs
outDict_gps_context[context]={'mean':np.array([np.mean(gpsValues.mean(axis=1))]),
'std':np.array([np.std(gpsValues.mean(axis=1))]),
'median': np.array([np.median(gpsValues.mean(axis=1))]),
'values': gpsValues.mean(axis=1)}
overall_gps_values = np.concatenate((outDict_gps_context["Right"]["values"],outDict_gps_context["Left"]["values"]))
outDict_gps={'mean':np.array([np.mean(overall_gps_values)]),
'std':np.array([np.std(overall_gps_values)]),
'median': np.array([np.median(overall_gps_values)]),
'values': overall_gps_values}
return outDict_gvs, outDict_gps_context,outDict_gps
def test_point_rms(acq, RefLabel, LabelToTest, threshold):
np.testing.assert_array_less(
numeric.rms(
(acq.GetPoint(utils.str(RefLabel)).GetValues() -
acq.GetPoint(utils.str(LabelToTest)).GetValues()[init:end, :]),
axis=0), threshold)
