Figure 2.

Flow chart of the main principle of the tumour viability assessment. The support vector machine (SVM) model is trained and optimised with the training set of single-tissue entity images (STEI), representing the different tissue categories of interest. The discrimination of the model is evaluated in parallel in test set of STEIs and in whole-slide images (WSIs). On the test STEI set, the agreement to classify a test image into viable or non-viable tumour category is evaluated by comparing result to manual labelling. Similarly on the WSI test set, the agreement in tissue segmentation and finally tumour viability assessment are evaluated by comparing obtained results with expert annotations.