Publishing house Radiotekhnika

"Publishing house Radiotekhnika":
scientific and technical literature.
Books and journals of publishing houses: IPRZHR, RS-PRESS, SCIENCE-PRESS

Тел.: +7 (495) 625-9241


Experience of biological image processing


S.D. Solnushkin - Senior Research Scientist, Pavlov Institute of Physiology, Russian Academy of Sciences (St.-Petersburg)
V.N. Chikhman - Ph.D. (Eng.), Senior Research Scientist, Head of Laboratory, Pavlov Institute of Physiology, Russian Academy of Sciences (St.-Petersburg)

Knowledge of the surrounding world is formed in man largely due to visual information. Therefore, software for image processing [1-3], in particular, computerized systems for biomedical images processing [4-7], are of great importance in scientific research. At the same time, because of the complexity of biomedical images and their variability, it is difficult to completely automate the process of their analysis.
An important place in the methodology of experimental physiological studies is occupied by an objective numerical analysis of photo images of prepared biological objects, in particular, organs of the gastrointestinal tract of laboratory animals (rats). When studying the role of protective hormonal mechanisms it becomes necessary to quantify the extent of mucosal involvement by calculating the total area of multiple erosions in different parts of the prepared organs. Examples of initial photo images of the gastrointestinal tract of the rats (gum and intestine) with the available erosions are shown in Fig. 1.
The use of automatic segmentation programs [7] to identify areas with erosion for the purpose of measuring their area later does not always yield satisfactory results. It seems that in our case it is most efficient to isolate areas of interest, as well as the necessary objects in the image for quantitative evaluation, only with the participation of a researcher with expert knowledge.
We have proposed an approach and developed software designed for quantitative analysis in an interactive mode of digital color images of the prepared organs of the gastrointestinal tract of laboratory animals (rats), as well as for the accumulation and maintenance of an appropriate database.
The program allows to solve the following tasks:
perform calibration and enter a scale factor;
edit images;
filter images;
convert the color image to binary (black and white);
allocate areas of "interest" in an interactive mode;
perform automatic image analysis (calculation of erosion areas);
maintaining the database (with the classification of images).
Selection of the object of research from the background is carried out by the experimenter with the help of the calculated and displayed brightness histogram on the screen. In the case of a color image using an additive RGB color model, there are three histograms of the components R, G, B in our location.
For the further possible classification of images by the degree of hyperemia (saturation of tissues with hemoglobin), we proposed to convert the matrix of the pixel values corresponding to the original image into the "red excess" matrix. Each element of this matrix is determined by the empirically found formula:
RR = R–(G+B)/2
R, G, B – the brightness values of the red, green, and blue components of the original image, respectively.
When constructing a new "excess red" histogram, only the positive RR values are taken into account. The received "excess red" histogram is displayed in the second window of the program's graphic interface under the image reconstructed with this histogram. The researcher uses the slider to select the RX value on the "excess red" histogram, resulting in a new color image, into which only the pixels whose RR value exceeds RX are transferred from the original image. The value of RX by the researcher is selected in an interactive mode by the method of successive approximation. Each time, assessing the visually obtained image, the researcher achieves the absence of artifacts (Fig. 2). The resulting value of RX is used later to classify the original images by the degree of hyperemia.
To quantify the destructive lesions from the filtered image, a halftone image is formed and a histogram of gray is plotted. Next, using the slider (specifying the gray scale threshold on the obtained gray histogram), the researcher in the interactive mode by the method of successive approximation (visual evaluation and comparison with the original image) achieves a satisfactory segmentation of the desired erosions displayed in black and white binary mode. Based on the segmentation results, a visually controlled binary image is constructed to perform the calculations. With the help of a moving graphical element (rectangle), the researcher sets the binary image of fragments of interest in which the erosion areas are counted automatically. The processed areas are numbered and marked with a marker (Fig. 3). The program generates and displays in a separate graphics window a table in which the number of erosions and their area in given units (pixel, millimeter) are displayed.
The measurement results are logged together with the image in the database. Element of the database with the source image and the results of processing is shown in Fig.4. If further processing is necessary, the measurement results can be exported, for example, using OLE tools in Excel. Using empirically determined coefficients obtained from the values of the color components of each pixel of the study object, the images are classified, for example, by the degree of hyperemia (reddening) of the tissue. These coefficients are the key parameters for sorting, selecting images in the database.
The developed means are used to process experimental material at the I.P. Pavlov Institute of Physiology of the Russian Academy of Sciences [9].

  1. Gonsales R., Vuds R. Cifrovaya obrabotka izobrazhenij. M.: Texnosfera. 2005. 1072 s.
  2. Potapov A.A., Paxomov A.A., Nikitin S.A., Gulyaev Ju.V. Novejshie metody' obrabotki izobrazhenij. M.: Fizmatlit. 2008. 496 s.
  3. Gonsales R., Vuds R., E'ddins S. Cifrovaya obrabotka izobrazhenij v srede MATLAB. M.: Texnosfera 2006. 616 s.
  4. Ablamejko S.V., Nedz'ved' A.M. Obrabotka opticheskix izobrazhenij kletochny'x struktur v mediczine. Minsk: OIPI NAN Belarusi. 2005. 156 s.
  5. Kosy'x N.E'., Smagin S.I., Gostyushkin V.V., Savin S.Z., Litvinov K.A. Sistema avtomatizirovannogo komp'yuternogo analiza mediczinskix izobrazhenij // Informaczionny'e texnologii i vy'chislitel'ny'e sistemy'. 2011. № 3. S. 51–56.
  6. Collins T.J. Image J for microscopy // BioTechniques. 2007. V. 43. P. 25–30.
  7. Privalov O.O., Butenko L.N. Avtomaticheskaya segmentacziya czifrovy'x izobrazhenij mediko-biologicheskix preparatov metodom klasternogo analiza // Sovremenny'e naukoemkie texnologii. 2007. № 10. S. 79–80.
  8. Jarushkina N.I., Bagaeva T.R., Filaretova L.P. Somaticheskaya bolevaya chuvstvitel'nost' kry's v usloviyax povrezhdayushhego dejstviya indometaczina // Rossijskij fiziologicheskij zhurnal im. I.M. Sechenova. 2014. T. 100. № 1. S. 73–85.
  9. Morozova O.Ju., Bagaeva T.R., Filaretova L.P. Mexanizmy' proul'czerogennogo dejstviya gidrokortizona, vvedennogo v farmakologicheskoj doze, na slizistuyu obolochku zheludka // Rossijskij fiziologicheskij zhurnal im. I.M. Sechenova. 2016. T. 102. № 3. S. 330–339.
June 24, 2020
May 29, 2020

© Издательство «РАДИОТЕХНИКА», 2004-2017            Тел.: (495) 625-9241                   Designed by [SWAP]Studio