Spectral Band Subset Selection for Discrimination of Healthy Skin and Cutaneous Leishmanial Ulcers

Ricardo Franco-Ceballos, Maria C. Torres-Madronero, July Galeano-Zea, Javier Murillo, Artur Zarzycki, Johnson Garzon, Sara M. Robledo

    Resultado de la investigación: Capítulo del libro/informe/acta de congresoContribución a la conferenciarevisión exhaustiva

    Resumen

    Leishmaniasis is a parasitic disease, transmitted by the bite of an insect that has previously fed on an infected host. One of its clinical forms is Cutaneous Leishmaniasis - CL and due to its increasing incidence, it is necessary to create effective and easy-use diagnostic methods. In this paper, we assess two unsupervised band-selection algorithms that allow the dimensional reduction of hyperspectral data taken from CL ulcers, maintaining a high classification accuracy. This is an important task for the development of an non-invasive system based on multispectral imaging, that support the diagnosis and treatment follow-up of cutaneous ulcer caused by Leishmaniasis. Spectral data was obtained in golden hamsters subjected to varying conditions of infection. Two algorithms, one based on similarity and the other based on singular values decomposition, are implemented using MATLAB functions and are applied to the spectral data. The selected subsets of bands are used to classify the spectra into healthy skin, border and ulcer centers using support vector machines - SVM and neural networks - NN. The obtained results are represented in precision tables and allow to observe that both methods achieve an appropriate dimensional reduction of multispectral data without losing key information for their subsequent classification. At the end, we show that it is possible to obtain a subset of spectral bands to discriminate between healthy skin and cutaneous ulcers caused by Leishmaniasis.

    Idioma originalInglés
    Título de la publicación alojadaPattern Recognition and Image Analysis - 9th Iberian Conference, IbPRIA 2019, Proceedings
    EditoresAythami Morales, Julian Fierrez, José Salvador Sánchez, Bernardete Ribeiro
    EditorialSpringer
    Páginas398-408
    Número de páginas11
    ISBN (versión impresa)9783030313319
    DOI
    EstadoPublicada - 2019
    Evento9th Iberian Conference on Pattern Recognition and Image Analysis, IbPRIA 2019 - Madrid, Espana
    Duración: 1 jul. 20194 jul. 2019

    Serie de la publicación

    NombreLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
    Volumen11867 LNCS
    ISSN (versión impresa)0302-9743
    ISSN (versión digital)1611-3349

    Conferencia

    Conferencia9th Iberian Conference on Pattern Recognition and Image Analysis, IbPRIA 2019
    País/TerritorioEspana
    CiudadMadrid
    Período1/07/194/07/19

    Nota bibliográfica

    Publisher Copyright:
    © 2019, Springer Nature Switzerland AG.

    Huella

    Profundice en los temas de investigación de 'Spectral Band Subset Selection for Discrimination of Healthy Skin and Cutaneous Leishmanial Ulcers'. En conjunto forman una huella única.

    Citar esto