Reproduced with permission from Blood Cancer Discovery.
Password-protected, demo can be provided for interested
Unsupervised UMAP vizualisation of visual features extracted with convolutional neural networks from H&E-stained bone marrow biopsies. Digital images were split into small tiles and projeced based on their dissimilarity. Colors reflect clusters, which have beeb regrouped into morphologic subcategories (dashed circles). The slingshot analysis demonstrate morphologic trajectories (superposed lines). Reproduced with permission from Blood Cancer Discovery.
Reproduced with permission from Blood Advances.
Lymphocyte density in various regions of renal cell carcinoma tissue sections defined with image analysis.
Reproduced with permission from Blood Cancer Discovery.