hetE: Runs heterogeneity analysis based on isotonic regression, as...
In ericwol/mia: Medical Imaging Analysis Tools

Runs heterogeneity analysis based on isotonic regression, as in Eary JF, O<e2><80><99>Sullivan F, O<e2><80><99>Sullivan J, Conrad EU. Spatial heterogeneity in sarcoma 18F-FDG uptake as a predictor of patient outcome. J Nucl Med. 2008; 49:1973<e2><80><93>1979 Arguments z: input ROI y=z[,5]; x=z[,2:4] #x is location and y is uptake eg: read ROID; output is column 5; location is col 2:4 nr=scan('ROID',n=1); z=matrix(scan('ROID',skip=1),nrow=nr,byrow=T) par0: initial parameters for nls uu1=par0[1]; uu2=par0[2]; uu3=par0[3]; a1=par0[4] ; a2=par0[5] ; a3=par0[6] ; a4=par0[7] ; a5=par0[8] Values v: output of iso.reg: - fitted monotonic regression line: yh=v[,2] - residuals: res = (v[,3]-v[,2]) het0: 100*mean(res^2)/mean(y^2) het1: 100*mean(res^2)/var(y) vary: var(y) hetvals: round(c(het0,het1,var(y)),4) [for retro-compatibility purposes] Updates: 12 Dec 2016: - restricted call to nls to case where sd>mv, where 'sd' = sqrt(var(yh0)) (yh0 = fitted isotonic values) 'mv' = .0001*abs(mean(yh0)) - more robust call to nls