A Discriminant Multi-Scale Histopathology Descriptor Using Dictionary Learning

    David Romo, Juan D. García‐Arteaga, Pablo Arbeláez, Eduardo Romero
    TLDR The new image descriptor helps identify skin cancer structures with good accuracy.
    The study proposed a novel multi-scale image descriptor using dictionary learning to enhance the identification of histological structures and concepts in histopathology. By employing sparse coding algorithms, separate dictionaries were created for each histological concept, allowing for the discrimination of visual elements associated with specific pathologies. The descriptor's effectiveness was evaluated through a naive classification strategy, achieving average recall and precision measures of 0.81 and 0.86, respectively, in classifying epidermis, eccrine glands, hair follicles, and nodular carcinoma in basal skin carcinoma images.
    Discuss this study in the Community →

    Related Community Posts Join

    6 / 6 results

    Similar Research

    5 / 1000+ results