oasisad_model: OASISAD model

In dty0606/oasisad: 'OASISAD' an Improved Automated White Matter Hyper-intensity Classficiation Tool

Description

The input should be OASISAD data list, the function will train the model with training and vailidaion data, then use the testing data to evaluatoin performance

Usage

oasisad_model(
  train_df,
  test_df,
  valid_df = NULL,
  M1 = FALSE,
  refine = FALSE,
  neighbor = FALSE,
  wm_label = NULL,
  re_value = NULL
)

Arguments

train_df	A data list from oasisad_df function which inlcudes training samples informatin. If neighbor refinement function will be used, the list should include segmentation and white matter probability map for each training subject
test_df	A data list from oasisad_df function which inlcudes testing samples informatin. If neighbor refinement function will be used, the list should include segmentation and white matter probability map for each training subject
valid_df	A data list from oasisad_df function which inlcudes validation samples informatin. If neighbor refinement function will be used, the list should include segmentation and white matter probability map for each training subject. If it is NULL, optimal threshold algorithm will be used to calculate threshold
M1	A boolean indicates using full model 'M1' or reduced model 'M2', default is reduced model
refine	A boolean incicates whether use OASISAD refinement function, to refine probability map from logistic regression model
neighbor	A boolean incicates whether use neighbor refinement function, to refine probability map from logistic regression model. If true, segmentation information and white matter probability of brain is needed
wm_label	White matter label in segmentation input
re_value	A numeric value will be used in neighor refinement functoin to refine a voxel's probability of being White matter hyperintensity

Value

OASISAD model results

dty0606/oasisad documentation built on March 8, 2020, 11:18 p.m.