Towards automated segmentation of cells and cell nuclei in nonlinear optical microscopy

Authors

  • Anna Medyukhina,

    1. Institute of Photonic Technology (IPHT) Jena e.V., Albert-Einstein-Str. 9, 07745 Jena, Germany
    Search for more papers by this author
  • Tobias Meyer,

    1. Institute of Photonic Technology (IPHT) Jena e.V., Albert-Einstein-Str. 9, 07745 Jena, Germany
    Search for more papers by this author
  • Michael Schmitt,

    1. Institute of Physical Chemistry and Abbe Center of Photonics, Friedrich-Schiller University Jena, Helmholtzweg 4, 07743 Jena, Germany
    Search for more papers by this author
  • Bernd F. M. Romeike,

    1. Institute of Pathology, Neuropathology, Jena University Hospital, Erlanger Allee 101, 07747 Jena, Germany
    Search for more papers by this author
  • Benjamin Dietzek,

    1. Institute of Photonic Technology (IPHT) Jena e.V., Albert-Einstein-Str. 9, 07745 Jena, Germany
    2. Institute of Physical Chemistry and Abbe Center of Photonics, Friedrich-Schiller University Jena, Helmholtzweg 4, 07743 Jena, Germany
    Search for more papers by this author
  • Jürgen Popp

    Corresponding author
    1. Institute of Photonic Technology (IPHT) Jena e.V., Albert-Einstein-Str. 9, 07745 Jena, Germany
    2. Institute of Physical Chemistry and Abbe Center of Photonics, Friedrich-Schiller University Jena, Helmholtzweg 4, 07743 Jena, Germany
    • Institute of Photonic Technology (IPHT) Jena e.V., Albert-Einstein-Str. 9, 07745 Jena, Germany
    Search for more papers by this author

Abstract

Nonlinear optical (NLO) imaging techniques based e.g. on coherent anti-Stokes Raman scattering (CARS) or two photon excited fluorescence (TPEF) show great potential for biomedical imaging. In order to facilitate the diagnostic process based on NLO imaging, there is need for an automated calculation of quantitative values such as cell density, nucleus-to-cytoplasm ratio, average nuclear size. Extraction of these parameters is helpful for the histological assessment in general and specifically e.g. for the determination of tumor grades. This requires an accurate image segmentation and detection of locations and boundaries of cells and nuclei. Here we present an image processing approach for the detection of nuclei and cells in co-registered TPEF and CARS images. The algorithm developed utilizes the gray-scale information for the detection of the nuclei locations and the gradient information for the delineation of the nuclear and cellular boundaries. The approach reported is capable for an automated segmentation of cells and nuclei in multimodal TPEF-CARS images of human brain tumor samples. The results are important for the development of NLO microscopy into a clinically relevant diagnostic tool. (© 2012 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim)

Ancillary