||Image Analysis of Breast Cancer Tissue Microarrays
||Højrup, Helene Hvidegaard
||Larsen, Rasmus (Image Analysis and Computer Graphics, Department of Informatics and Mathematical Modeling, Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark)
||Technical University of Denmark, DTU, DK-2800 Kgs. Lyngby, Denmark
||Breast cancer is by far the most common cancer in women in Denmark.
Approximately 4000 new cases are diagnosed annually and about 1200 die
from the disease . When changes in the breast are observed a biopsy
is taken to determine whether or not the worrisome lumps are cancer or
not. If it is cancer, the tissue from the biopsy is further processed to
state the type of cancer. This diagnosis is based on manual countings of
the cell nuclei in the tissue hence this is a slow and tedious process. For
this reason automatic segmentation methods are wanted.
The goal of this study was to develop an efficient and accurate algorithm for detecting and segmenting cell nuclei in 2D histological images
followed by an automated counting of the cell nuclei. The algorithm developed constituted on, texture analysis, color segmentation and dynamic
programming. The results gave the impression of an algorithm that was
somehow able to reconstruct the manually counted results. However
these results did not seem to depend much on the segmentation of the
nuclei, rather on the distribution of blue and brown colors in the image.
Questions about the validity of the results were also raised, since
only thirteen cores from three different TMA blocks were used in the
analysis. More data was available, however the computation time for the
algorithm, was far too long for all this to be analysed. Of cause a great
buttleneck in the algorithm.
||Technical University of Denmark (DTU) : Kgs. Lyngby, Denmark
Creation date: 2011-02-03
Update date: 2011-02-03