This project won an award in Technion's Innovation Contest 2006
Changes in Heart Cells

by Dani (Yevgeny) Pinkovich and Alexander Berkovich
Supervised by Tomer Michaeli


At the connection points between grown heart cells lays a protein, called connexin 43, which is used as a conductor of the electrical signal in the heart. An important question is whether or not an internalization of the connexin 43 happens during a phase of hypoxia that the heart cells are put through.

In addition, we wish to examine the morphological changes that occur in young, circularly shaped heart cells as a result of a periodic stretch forced upon them.



In part 1 the following algorithm was used: 

1. Three points are chosen to describe the border between the two cells, assuming it can be modeled by a circle

.2. An area which encloses the border between the cells is defined using a polygon.

3. The program finds the area where the protein is considered to be on the egde of the cell. Protein outside this area will be considered as a protein inside the cell.

4. The program calculates the amount of protein on the border between the cells and inside the  cells  -  before and after  the hypoxia.

5.  Now the results of the previous calculation are compared. At this point it can be said whether there was indeed significant internalization of the protein as a result of the hypoxia.

In addition, we wish to examine the morphological changes that occur in young, circularly shaped heart cells as a result of a periodic stretch forced upon them.

In part 2 of the project the next steps are used:

1. A median filter is applied on the picture.

2. A threshold is used in order to detect the nuclei in the picture.

3. For each of the objects considered to be a nucleus some geometric characteristics are found:

  • Center
  • Area
  • erimeter
  • Minor and Major axes lengths
  • ngle of the major axis
  • ounding box

Using the data for each of the found objects, those that are not nuclei are filtered out using  their size or shape. Now only objects that are indeed nuclei with theircharacteristics are measured. Now the characteristics of the nuclei, before and after the stretch occurs, are compared.


In the first part, as can be seen from the pictures above, much more connexin 43 inside the cells after the stretch was implemented was detected. This can  be seen from the original pictures and the calculations confirm it.

In the second part, the algorithm detected a total of 1596 nuclei, out of which only 50 were deleted as mistakes (3% error).

This yields the following results:


In part 1 of the project, it can be seen without any doubt that there was indeed a process of internalization of the connexin 43 into the cell. The protein had clearly more presence inside the cell after the hypoxia stage, rather than prior to that stage of the experiment.

In part 2 it was found that the cells do stretch as a result of the force applied on them. In addition, the cells do tend to turn as a result of the stretch. They turn in the perpendicular direction to the direction of the stretch that the cells experiences, which contradicts the first assumption.


[3] Soille, P., Morphological Image Analysis: Principles and Applications, Springer-Verlag, 1999, pp. 173-174
[4] Madison S. Spach, J. Francis Heidlage, Paul C. Dolber, Roger C. Barr, Electrophysiological Effects of Remodeling Cardiac Gap Junctions and Cell Size
[5] Jeffrey E. Saffitz, James G. Laing, Kathryn A. Yamada, Connexin Expression and Turnover: Implications for Cardiac Excitability
[6] Stephan Rohr, Role of gap junctions in the propagation of the cardiac action potential
[7] P. M. Patel, B.SC., A. Plotnikov, M.D., PH.D.,P. Kanagaratnam, M.B., A. Shvilkin, M.D., PH.D.,C. T. Sheehan, B.SC., W. Xiong, M.D., P. Danilo JR, PH.D.,M. R. Rosen, M.D., and N. S. Peters, M.D., Altering Ventricular Activation Remodels Gap Junction Distribution in Canine Heart


We would like to thank our supervisor Tomer Michaeli  for his support and guidance throughout this project.
Also we would like to thank Prof. Ofer Binah and Asaf Danon who initiated the project and gave the biological point of view to the subject at hand.
Especially we would like to that Prof. Yonina Eldar who supervised the project and guided us through the different signal processing methods.
Finally many thanks to the Ollendorff Minerva Center that supported our project.

Related documentation