Figure 4From: Novel image markers for non-small cell lung cancer classification and survival prediction(Left) The plot of mean partial log likelihood against the number of boosting steps using six-fold cross validation and (Right) the plot of coefficient paths, i. e. the parameter estimation plotted against the boosting steps until the optimal steps were reached. M = 21 is used since it minimizes the mean partial log likelihood. Since we use an offset-based update mechanism, the same image feature might be reselected and its coefficient would be updated. We use branch-out for illustration purpose when one specific image feature is selected.Back to article page