Automatic assessment of leishmania infection indexes on in vitro macrophage cell cultures

  • Authors:

  • Pedro Leal;Luís Ferro;Marco Marques;Susana Romão;Tânia Cruz;Ana M. Tomá;Helena Castro;Pedro Quelhas

  • Affiliations:

  • Faculdade de Engenharia, Departamento de Engenharia Química, Universidade do Porto, Portugal;Faculdade de Engenharia, Departamento de Engenharia Química, Universidade do Porto, Portugal;Faculdade de Engenharia, Departamento de Engenharia Química, Universidade do Porto, Portugal;IBMC-Instituto de Biologia Molecular e Celular, Universidade do Porto, Portugal;IBMC-Instituto de Biologia Molecular e Celular, Universidade do Porto, Portugal;IBMC-Instituto de Biologia Molecular e Celular, Universidade do Porto, Portugal;IBMC-Instituto de Biologia Molecular e Celular, Universidade do Porto, Portugal;INEB- Instituto de Engenharia Biomédica, Portugal, Faculdade de Engenharia, Departamento de Engenharia Electrotécnica e Computadores, Universidade do Porto, Portugal

  • Venue:
  • ICIAR'12 Proceedings of the 9th international conference on Image Analysis and Recognition - Volume Part II
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

Evaluation of parasite infection indexes on in vitro cell cultures is a practice commonly employed by biomedical researchers to address biological questions or to test the efficacy of novel anti-parasitic compounds. In the particular case of Leishmania infantum, a unicellular parasite that parasitizes macrophages, infection indexes are usually determined either by visual inspection of cells directly under the microscope or by counting digital images using appropriate software. In either case assessment of infection indexes is time consuming, thus motivating the creation of automatic image analysis approaches that allow large scale studies of Leishmania-infected macrophage cultures. We propose a fully automated method for automatic evaluation of parasite infection indexes through the segmentation of individual macrophages nucleus and cytoplasm, as well as the segmentation and co-localization of the parasites in the image. To perform such analysis with robustness and increased performance we propose the use of local image filters tuned to the specific size of the objects to detect, in conjunction with image segmentation approaches. The objects size estimation is then improved through a learning feedback loop. Cytoplasm is detected by seeded watershed segmentation. Our approach obtains, for 86 images from 4 experiments, an average parasite infection index evaluation error of 2.3%.