Validation of confocal microscope through noise measurements of homogeneous field

A thesis accepted by Tallinn University of Technology for the Degree of Bachelor of Science

Ürgo Saaliste

Tallinn University of Technology, Estonia

Pearu Peterson, PhD

Institute of Cybernetics at Tallinn Technical University, Estonia

David Schryer, MSc

Institute of Cybernetics at Tallinn Technical University, Estonia

Jaan Kalda, PhD.

Institute of Cybernetics at Tallinn Technical University, Estonia


The aim of this work was to study the noise characteristics of a confocal microscope to provide an adequate basis for the use of deconvolution algorithms. Depending on the nature of the noise one must select an appropriate deconvolution algorithm. Measurements were carried out on three different homogeneous fields which were total darkness, distilled water and fluorescence in solution. Data processing was later carried out using a universal scripting language called Python. The total darkness means that the laser was turned off and the microscope was set on eye- piece so that no light from the objective goes to the confocal apparatus. Measurements were carried out using the following pixel acquisition times: 10 μs and 1000 μs. Florescent field was achieved by using 1 μM and a 2.5 μM dextran Alexa 647 dye solutions. Distilled water was used as solvent. Measurements were carried out using the following pixel acquisition times: 2 μs, 4 μs, 8 μs, 16 μs and 32 μs. Distilled water measurements were also carried out with the same pixel acquisition times mentioned last. Multiple pictures were acquired for each pixel acquisition time. The first part of the thesis consists of 8 pages of literature review which describes the development and the main principle of the confocal microscope. The second part of the thesis consists of 5 pages and focuses on the different analysis techniques used in this work. In the third part consisting of 6 pages, results are shown and discussed after which summary is given. At the end of the thesis 37 references are given. There are also 37 appendixes. 13 figures and graphs in total are being used to illustrate main ideas of the thesis. Five different analysis techniques were used in the work. First of all autocorrelation was used to prove the independence of each pixel of the image, which was followed by creation of normalized histograms from image data and comparing it to the Poisson mass distribution, based on the mean of the histogram. From that it was concluded that the distribution was very close to the theoretical distribution. To assess whether the changes were statistical in nature or they were generated by an unknown source of noise, numerical arrays of different length, which followed the Poisson distribution, were generated based on the means of the histograms. Afterwards histograms were made and compared to the Poisson mass distribution. The results were grouped together on different graphs. From the graphs it is easy to see that the generated numbers always follow the Poisson distribution more accurately than the results from the experiments, thus confirming the existence of unknown secondary noise. Next photon registering gradient was measured. To measure the gradient, pictures from different manifestations of the same experiment were summed together so that the pixels that had the same indexes were added together. After that a 400 pixel diameter circle was cut out of the middle of the picture. The circle was then cut in half horizontally and the ratio of the upper and lower part was calculated. This was repeated with different angels up to 180 degrees after which the ratios were presented as a graph. It can be seen from the graphs that there exists a gradient field under the direction of 20 degrees. Finally, on average of all experiments were found and the means of average columns as well as average rows were depicted on graphs. The graphs showed that less photons were registered in the beginning of scanning lines and in the lines scanned earlier. An important observation was made during experimentation: the dark count of the detector fluctuated in time quite a lot. Suggestions were made to further analyze the dark count of the detector within longer periods of time. In conclusion it was stated that the topic needs further investigation because this work did not give answers to origin of the unknown noise.


The presentation of the thesis took place on June, 2010, at Tallinn University of Technology, Tallinn, Estonia.


Thesis is written in Estonian.

Softcopy: (1.9MB).