BP_FitMLRadialCompactness: Estimation of the likelihood of a bone section

In BoneProfileR: Tools to Study Bone Compactness

Estimation of the likelihood of a bone section

Description

Estimation of the compactness of a bone section using radial model.
If the fitted.parameters and fixed.parameters are NULL and the analysis includes a BP_FitMLCompactness() result, the values of this result is used as a reference for fitted.parameters and fixed.parameters.
If no BP_FitMLCompactness() result is available, it will use:
fitted.parameters=c(P=0.5, S=0.05, Min=-2, Max=5); fixed.parameters=c(K1=1, K2=1).
The reference for radial estimation of compactness is the trigonometric circle for rotation.angle=0 in BP_EstimateCompactness():
- The top of the section is located at -pi/2.
- The left of the section is located at -pi and +pi.
- The bottom of the section is located at pi/2.
- The right of the section is 0.
If rotation.angle is different from 0, the value of rotation.angle is added to the angle modulo 2.pi.
The two-steps analysis performs first a quasi-Newton method, then a Bayesian MCMC and finally again a quasi-Newton method. It generally ensures that global minimum is found. On the other hand, it doubles the time to complete for each angle.

Usage

BP_FitMLRadialCompactness(
  bone,
  fitted.parameters = NULL,
  priors = NULL,
  fixed.parameters = NULL,
  analysis = 1,
  silent = FALSE,
  twosteps = TRUE
)

Arguments

bone	The bone image to be used
fitted.parameters	Parameters of the model to be fitted
priors	If twosteps is TRUE, ell what prior should be used.
fixed.parameters	Fixed parameters of the model
analysis	Name or rank of analysis
silent	Should the function displays some information?
twosteps	Should a 2-steps analysis be performed?

Details

BP_FitMLRadialCompactness estimates likelihood of model of a bone section

Value

The -Ln L

Author(s)

Marc Girondot [e-mail]

See Also

Other BoneProfileR: BP_AutoFit(), BP_ChooseBackground(), BP_ChooseCenter(), BP_ChooseForeground(), BP_DetectBackground(), BP_DetectCenters(), BP_DetectForeground(), BP_DuplicateAnalysis(), BP_EstimateCompactness(), BP_FitBayesianCompactness(), BP_FitMLCompactness(), BP_GetFittedParameters(), BP_ListAnalyses(), BP_LnLCompactness(), BP_OpenImage(), BP_Report(), Erinaceus_europaeus, plot.BoneProfileR(), summary.BoneProfileR()

Examples

## Not run: 
# Not run
library(BoneProfileR)
path_Hedgehog <- system.file("extdata", "Erinaceus_europaeus_fem_2-1_small.png", 
                             package = "BoneProfileR")
 bone <- BP_OpenImage(file=path_Hedgehog)
 # or 
 bone <- BP_OpenImage(ijtiff=TRUE)
 bone <- BP_DetectBackground(bone=bone, analysis="logistic")
 bone <- BP_DetectForeground(bone=bone, analysis="logistic")
 bone <- BP_DetectCenters(bone=bone, analysis="logistic")
 bone <- BP_EstimateCompactness(bone, analysis="logistic")
 bone <- BP_EstimateCompactness(bone, analysis="logistic", cut.angle=30)
 bone <- BP_FitMLCompactness(bone, analysis="logistic")
 plot(bone)
 plot(bone, type="observations")
 plot(bone, type="observations+model", analysis=1)
 fittedpar <- BP_GetFittedParameters(bone, analysis="logistic")
 bone <- BP_DuplicateAnalysis(bone, from="logistic", to="flexit")
 bone <- BP_FitMLCompactness(bone, 
                fitted.parameters=c(fittedpar, K1=1.01, K2=1.01), 
                fixed.parameters=NULL, analysis="flexit")
 bone <- BP_FitBayesianCompactness(bone, analysis="flexit")
 mcmc <- RM_get(bone, RMname = "flexit", value="mcmc")
 fittedpar <- as.parameters(mcmc)
 bone <- BP_FitMLCompactness(bone, 
                fitted.parameters=fittedpar, 
                fixed.parameters=NULL, analysis="flexit")
 compare_AIC(Logistic=BP_GetFittedParameters(bone, analysis="logistic", alloptim=TRUE), 
             Flexit=BP_GetFittedParameters(bone, analysis="flexit", alloptim=TRUE))
 out4p <- plot(bone, type="observations+model", analysis="logistic")
 out6p <- plot(bone, type="observations+model", analysis="flexit")
 # The twosteps fit is more acurate but is around 100 times slower
 bone <- BP_FitMLRadialCompactness(bone, analysis="logistic", twosteps=TRUE)
 bone <- BP_FitMLRadialCompactness(bone, analysis="logistic", twosteps=FALSE)
 plot(bone, type="observations", angle=0)
 plot(bone, type="model", analysis="logistic", angle=0)
 plot(bone, type="observations+model", angle=0)
 plot(bone, type="observations+model", angle=pi)
 plot(bone, type="radial", radial.variable=c("P", "S"), analysis="logistic")
 plot(bone, type="radial", radial.variable=c("P", "S", "Min", "Max"), analysis="logistic")
 plot(bone, type="radial", radial.variable=c("TRC"), analysis="logistic")
 # Test using the change of orientation using default.angle from BP_EstimateCompactness():
 bone <- BP_DuplicateAnalysis(bone, from="logistic", to="logistic_rotation_pi")
 # With a pi rotation, the top moves to the bottom and the left moves to the right
 bone <- BP_EstimateCompactness(bone, rotation.angle=pi, analysis="logistic_rotation_pi")
 bone <- BP_FitMLRadialCompactness(bone, analysis="logistic_rotation_pi")
 plot(bone, type="radial", radial.variable=c("P", "S"), analysis="logistic")
 plot(bone, type="radial", radial.variable=c("P", "S"), analysis="logistic_rotation_pi")
 BP_Report(bone=bone, 
           analysis=1,
           docx=NULL, 
           pdf=NULL, 
           xlsx=file.path(getwd(), "report.xlsx"), 
           author="Marc Girondot", 
           title=attributes(bone)$name)

## End(Not run)

BoneProfileR documentation built on Sept. 7, 2022, 1:06 a.m.