IS = { zkontrolovano 12 Jan 2014 },
  UPDATE  = { 2014-01-06 },
  author =       {Borovec, Ji{\v r}{\'\i}},
  supervisor =   {Kybic, Jan},
  title =        {Segmentation and registration of multiple stained
                  histological sections -- {PhD} Thesis Proposal},
  institution =  {Center for Machine Perception, K13133 FEE Czech Technical
  address =      {Prague, Czech Republic},
  year =         {2013},
  month =        {September},
  type =         {Research Report},
  number =       {CTU--CMP--2013--23},
  issn =         {1213-2365},
  pages =        {34},
  figures =      {9},
  authorship =   {100},
  psurl =        {[Borovec-TR-2013-23.pdf]},
  project =      {SGS12/190/OHK3/3T/13, GACR P202/11/0111},
  annote =       {The analysis of protein-level multigene expression signature
                  maps computed from the fusion of differently stained
                  immunohistochemistry images is an emerging tool in cancer
                  investigation. Creating these maps requires registering sets
                  of histological images, a challenging task due to their
                  large size, the non-linear distortions existing between
                  consecutive sections and to the fact that the images
                  correspond to different histological stains and thus, may
                  have very different appearance. This thesis proposal
                  discusses the registration of differently stained
                  consecutive histological sections together with typical
                  image size of several thousands times several thousands
                  pixels. The main idea is to do the segmentation farst and
                  then run the registration on the segmented images which
                  should be faster (simpler similarity metric to evaluate or
                  e.g. kind of contour registration) and more robust. So far,
                  we have preferred the segmentation and registration of
                  stained histological sections independently. Later on we
                  would like to do both processes simultaneously. The thesis
                  proposal briefly summarises the state-of-the-art methods,
                  mainly focusing on registration. We present our histological
                  images and also synthetic datasets we have designed to
                  simulate the real images. We discuss our existing
                  unsupervised multi-class segmentation method and new
                  similarity metric measure for registration of segmented
                  images. In the end, we discuss future work and propose the
                  future research directions.},
  keywords =     {segmentation; registration; stained; histology},