imaging.ROC: A function to create ROC plots from a list of prediction...

In JMMaronge/jmm.funcs: Functions for personal use by Jacob M. Maronge

Description

This function creates different types of ROC plots from prediction images for segmentation. It also calculates AUC and returns the data.frame used for the plot.

Usage

imaging.ROC(preds.list, mask.list, y.list, vec.id, type, group.list, max.fpr)

Arguments

preds.list	A list of prediction IMAGES (will change later) as NIfTI objects for different subjects (can be of length 1)
mask.list	A list of PATHS to brain masks in the same order as preds.list
y.list	A list of PATHS to the gold standard segmentations in the same order as preds.list.
vec.id	A vector of ID's (as characters) corresponding to preds.list
type	There are 3 options: "subject", "total", or "group". "subject" creates a seperate ROC for each subject, "total" creates one overall ROC, and by group makes a seperate ROC for each group as determined by the group.list argument.
group.list	A list of sub-lists. This is only necessary if type="group". This should be a list of sublists where each sublist is the IDs of the subjects wanted in each group. For example, list(list("1","2"),list("3","4")) would create two ROCs, one with subjects "1" and "2" and the other with subjects "3" and "4".
max.fpr	A scalar between (0,1). Use this to create partial ROCs instead of full ROCs. The ROC will be cut-off at the max false positve rate (max.fpr) specified.

Examples

imaging.ROC()

JMMaronge/jmm.funcs documentation built on May 7, 2019, 10:12 a.m.