Melanoma Segmentation and Classification Employing Imaging and Graph Encodings

Histopathology imaging is crucial for the diagnosis and the decision on treating associated tissue diseases. For this reason, Computer-assisted approaches have gained popularity and shown promise in tasks such as segmentation and classification. However, collecting a significant amount of data and high-quality annotations remains a challenge. This work describes a pipeline for toponomy profiling on the population of cells that combines melanoma multidimensional fluorescent imaging, a supervised segmentation of the region of interest, and graph unsupervised embedded representation.The Multi-Epitope-Ligand Cartography (MELC) provides a range of data shown to be optimal for solving supervised and unsuper-vised tasks. This imaging technique and the methodology, such as the one presented here, are promising for further improvement in the study of diagnostic techniques and therapies for skin diseases, such as melanoma.