Simultaneously checks one or more quality assurance constraints on one or more DVHs. Reports compliance with each constraint as well as observed difference between linearly interpolated DVH and the given constraints in terms of (relative) dose, (relative) volume, and (relative) minimal Euclidean distance.
checkConstraint(x, constr, byPat=TRUE, semSign=FALSE,
sortBy=c("none", "observed", "compliance", "structure",
"constraint", "patID", "deltaV", "deltaD",
"dstMin", "dstMinRel"),
interp=c("linear", "spline", "smooth"), ...)# S3 method for DVHs
checkConstraint(x, constr, byPat=TRUE, semSign=FALSE,
sortBy=c("none", "observed", "compliance", "structure",
"constraint", "patID", "deltaV", "deltaD",
"dstMin", "dstMinRel"),
interp=c("linear", "spline", "smooth"), ...)
# S3 method for DVHLst
checkConstraint(x, constr, byPat=TRUE, semSign=FALSE,
sortBy=c("none", "observed", "compliance", "structure",
"constraint", "patID", "deltaV", "deltaD",
"dstMin", "dstMinRel"),
interp=c("linear", "spline", "smooth"), ...)
# S3 method for DVHLstLst
checkConstraint(x, constr, byPat=TRUE, semSign=FALSE,
sortBy=c("none", "observed", "compliance", "structure",
"constraint", "patID", "deltaV", "deltaD",
"dstMin", "dstMinRel"),
interp=c("linear", "spline", "smooth"), ...)
A single DVH (object of class DVHs), multiple DVHs from one patient/structure (object of class DVHLst), or multiple DVHs from many patients/structures (object of class DVHLstLst). See readDVH.
One or more constraints - given as a character vector or as a data.frame. See Details.
logical. Relevant if multiple DVHs are given. If x has class DVHLst: byPat=TRUE means that the DVHs are for one patient with multiple structures. byPat=FALSE means that the DVHs are for one structure from multiple patients. If x has class DVHLstLst: byPat=TRUE means that the DVHs are for multiple patients (list components of x) with multiple structures. byPat=FALSE means that the DVHs are for multiple structures (list components of x) from multiple patients.
logical. Meaning of the sign of the observed dose/volume differences between DVHs and constraints. semSign=TRUE means that negative differences indicate constraint compliance, positive differences indicate constraint violations. With semSign=FALSE, the algebraic differences are returned as is.
character vector. Sorting criteria for the output data frame.
A data frame with details on constraint compliance / violation.
Patient ID
Structure
The checked constraint
The observed value for the metric given in the constraint
Does the DVH satisfy the constraint?
Volume difference between constraint and observed DVH (for the constraint dose) in measurement unit specified by constraint
Percent volume difference between constraint and observed DVH (for the constraint dose) relative to constraint volume
Dose difference between constraint and observed DVH (for the constraint volume) in measurement unit specified by constraint
Percent dose difference between constraint and observed DVH (for the constraint volume) relative to constraint dose
Minimal Euclidean distance between constraint and the cumulative DVH, using linear interpolation
Dose coordinate of closest point on cumulative DVH to constraint
Volume coordinate of closest point on cumulative DVH to constraint
A DVH constraint is a character string that consists of three parts: The DVH metric, the comparison operator (<, >, <=, >=), and the reference value together with the measurement unit. See getMetric for defining a DVH metric, as well as for possible measurement units for dose and volume. For constraints involving the relative dose, the DVH must contain the prescription dose.
Some example constraints are "V10Gy > 80%" (more than 80% of the structure should have received 10Gy), "V20% < 10CC" (less than 10cm^3 of the structure should have received 20% of the prescription dose), or "D10CC > 500cGy" (The "hottest" 10cm^3 of the structure should have received more than 500cGy).
For constraints on DEUD, DNTCP and DTCP (see getMetric), the reference measurement unit must be Gy, cGy, even though NTCP and TCP are probabilities. Example: "DNTCP < 0.5Gy".
A DVH constraint can apply to a specific patient or to all patients, and to a specific structure or to all structures.
If constraints apply to all patients/structures in x, constr can be a character vector with elements like the examples above.
If constraints apply only to some patients/structures, constr must be a data frame with variables constraint, patID and structure. Each row then defines one constraint and its scope: constraint must be a character string with one constraint definition as in the examples above. patID must be either a character string with a valid patient ID or "*" if the the constraint applies to all patients. structure must be either a character string with a valid structure or "*" if the the constraint applies to all structures. If variable patID is missing from the data frame, the constraints apply to all available patients. If variable structure is missing from the data frame, the constraints apply to all available structures. See readConstraint for reading appropriate constraint data.frames from external text files.
For calculating the minimal Euclidean distance between the constraint point and the DVH, the constraint point is orthogonally projected onto each DVH segment between (interpolated) DVH nodes. The relative Euclidean distance is the minimum of these distances divided by the distance of the constraint point to the closer one of both axes (dose and volume).
If volume or dose values outside the range of possible values for a structure are requested, metrics cannot be calculated, and the result will be NA with a warning.
getMetric,
getEUD,
getNTCP,
getTCP,
readConstraint,
saveConstraint,
showConstraint
# NOT RUN {
res <- checkConstraint(dataMZ, c("D10CC < 10Gy", "V20Gy < 20%"))
head(res)
# define constraints
constr <- data.frame(
patID=c("P123", "P234"),
structure=c("HEART", "*"),
constraint=c("D1CC < 20Gy", "V10% > 8CC"),
stringsAsFactors=FALSE) # this is important
checkConstraint(dataMZ, constr=constr)
# }
Run the code above in your browser using DataLab