sp.similarity.from.bin: Volume-based spatial similarity metrics calculated from...

In espadon: Easy Study of Patient DICOM Data in Oncology

Volume-based spatial similarity metrics calculated from binary modality 3D volumes. \loadmathjax

Description

The sp.similarity.from.bin function computes volumetric Dice similarity coefficient, Dice-Jaccard coefficient and Dice surface similarity coefficient.

Usage

sp.similarity.from.bin(
  vol.A,
  vol.B,
  coeff = c("DSC", "DJC", "MDC", "under.MDC", "over.MDC")
)

Arguments

vol.A, vol.B	"volume" class objects, of "binary" modality. vol.B is the reference for MDC calculation.
coeff	Vector indicating the requested metrics from among 'DSC' (Dice similarity coefficient),'DJC' (Dice-Jaccard coefficient), and 'MDC' (mean distance to conformity). Equal to NULL if not requested.

Value

returns a dataframe containing (if requested):

• volumetric Dice similarity coefficient DSC defined by : \mjdeqnDSC = 2 \fracV_A ~\cap~ V_BV_A + V_Bascii

• Dice-Jaccard coefficient DJC defined by : \mjdeqnDJC = \fracV_A ~\cap~ V_BV_A ~\cup~ V_Bascii

• mean distance to conformity MDC, over-contouring mean distance over.MDC and under-contouring mean distance under.MDC, defined by Jena et al [1]

References

[1] \insertRefJENA201044espadon

See Also

sp.similarity.from.mesh

Examples

# creation of to volume" class objects, of "binary" modality
vol.A <- vol.create (pt000 = c(-25,-25,0), dxyz = c (1 , 1, 1),
                    n.ijk = c(50, 50, 1), default.value = FALSE,
                    ref.pseudo = "ref1",
                    alias = "vol.A", modality = "binary",
                    description = "") 
vol.B <- vol.copy (vol.A,alias = "vol.B")   
vol.A$vol3D.data [as.matrix(expand.grid(15:35,20:35,1))] <- TRUE
vol.A$max.pixel <- TRUE
vol.B$vol3D.data [as.matrix(expand.grid(16:36,18:37,1))] <- TRUE
vol.B$max.pixel <- TRUE
display.plane (vol.A, vol.B, interpolate = FALSE, 
               main = "vol.A & vol.B @ z = 0 mm") 
               
sp.similarity.from.bin (vol.A, vol.B)

espadon documentation built on April 11, 2025, 5:57 p.m.