Biological cells segmentation

    Cell segmentation is very important task in medical image processing.
The aim of the project was to develop the automatic solution for specific biological cells segmentation.
We shouldn’t talk about biological details of the challenge as this is not our research field and just focus on computer vision algorithms.
Our objective in this area was to develop robust and precise algorithm for automatic structural segmentation of β-catenin cells.
There some issues for segmentation of such cells. it’s a difficult to assess and determine cells borders visually. Some cells borders partially invisible also some cells are overlapped. Example of source image you can see below:

roi b-katenin2

Fig. 1. Source cell image

In order to develop high-precision algorithm of cell segmentation, the following scheme was applied:

 

cell segmentation diagram v2

First and foremost, we improve visual quality of source cell image using Adaptive histogram equalization algorithm. Also, we use some additional information to supplement some missing and invisible cell borders on source image.  In addition, very simple image filtration was applied to remove unnecessary cell candidates.

Main steps of image filtration next:

  • Gaussian smoothing to remove some noise;
  • CLAHE to highlight cell border;
  • A series of morphology operations, such as opening, closing, fill holes and other;
  • Cells clustering and filtration using geometric features.

You can see noticeable changes of cells processing below.

cell segmentation diagram2

Fig. 2. Result of cells improving and filtration

To improve segmentation algorithm and localize cells more accurately, were used dapi fluorescent images.

roi dapi colored

Fig. 3. Dapi fluorescent image

Watershed segmentation was applied to dapi image to obtain markers (labels) for -catenin cells segmentation.

Finally, dapi markers were used for final cells segmentation with improved watershed algorithm.

Below you can see a simple animated demo illustrating how the biological cells segmentation works.

Cell segm animation

Fig. 4. Сell segmentation demo

Thus, such fusion of processing from different input images allows to retrieve the required cells structure automatically.

Biological cells segmentation