Characteristics of the DT models selected on the basis of both their performance with the training set and their complexity and results obtained on the validation set.

Classification performances are shown in terms of accuracy (the percentage of correctly classified tiles) and sensitivity (the percentage of blurred tiles correctly classified as blurred by the DT). Columns "B -> S" and "S -> B" show the number of false negative decisions (i.e., the blurred tiles classified as being sharp) and the number of false positive decisions (i.e., the sharp tiles classified as being blurred), respectively. Concerning DT training, the classification performances resulted from a nested cross-validation (5-fold x 5-fold) carried out either on the HE and IHC groups separately or on all of the training data (see column “Set”). Columns "DT algorithm" and "Features in DT model" describe the choices resulting from the model selection method (i.e., DT split criterion and selected features, respectively). All of the results leading to the presented selections are detailed in the supplementary data (Table S1). Concerning the quantitative validation step, the models selected during the training step (see column “DT algorithm”) were applied to independent sets of tiles (see column “Set”). The resulting classification performances are indicated as for training. The number of tiles in each set is specified in column “N tiles”. Rows “Pooled” indicate the results obtained by pooling the two previous rows (HE and IHC) to allow for comparisons with the results in rows “ALL” below.