In ajharry/msiCompare:

Introduction

Purpose this is a demonstration of functionality *overview of features and availability

library(msiCompare)

Integration with Cardinal

library(Cardinal)

Simulating MS images

Model-based simulation

conditions <- ifelse((expand.grid(x=1:30, y=1:30)$x %in% (1+floor(30/5)):(30-floor(30/5)) & expand.grid(x=1:30, y=1:30)$y %in% (1+floor(30/5)):(30-floor(30/5))), 2, 1)

 s <- simSingle(
      reps = 3,
      diff = log2(1.5),
      tau2 = 0.1,
      sig2 = 0.1,
      seed = 8372,
      size1 = 30,
      center.pattern = T,
      pattern = conditions
      )

summary(s)

image(s$simSet, feature = 1)

 s_multi <- simMulti(
      sampleVar = 0.1,
      reps = 1,
      diff = log2(1.5),
      tau2 = 0.1,
      sig2 = 0.1,
      seed = 8372,
      size1 = 30,
      size2 = 30^2,
      numHealthy = 3,
      numDisease = 3)

image(s_multi$simSet, feature = 1, layout = c(3,2))

Examples of class comparison

Hiearchical Bayesian Spatial Model

show compareMSI

fitHBSM <- compareMSI(msset = s_multi$simSet,
           conditionOfInterest = s_multi$simSet$diagnosis,
           techRep = s_multi$simSet$sample,
           feature = 1,
           nsim = 5000, burnin=2500,
           trace = F, dropZeros = T)

fitHBSM$Feature1$gamma

note hiearchical centering bfdr adjustment

Location-wise ANOVA

summary(lm(c(spectra(s_multi$simSet)[1,]) ~ s_multi$simSet$diagnosis))

p_averaging(s_multi$simSet)

Cassese et al

fit_spautolm <- cass(msset = s$simSet, roiFactor = factor(s$simSet$diagnosis),
     logscale = F, thresholds = 1:5)

fit_spautolm$results$CAR_AIC_min_pvalues