1

> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <



Image Processing inResearch of Saliva

T.Victorova

Abstract—A saliva crystal structure is an important information source for dental researches of a person healthstate. In this article questions of images processing automation forsaliva crystal structure after liquid substance drying are considered.

Index Terms— saliva, crystal structure, biomedical image processing

I.INTRODUCTION

S

alivaof the person and, more exactly, its crystal structure contains a lot of information used in research of a health state [3]. The system for saliva research consists of a glass transparent plate («subject glass»), anoptical microscope, digital photo camera Nikon COOLPIX 950 or similar for transformation of video image to digital form with the high resolution and computer connected tothe camera. Microscope NU-2E or similar with magnification ratio of 100-200 is used. The requirements to the computer areCeleron CPU at 1.8 GHz and 512 MbRAM.

II.The Saliva Structure

A.Review Stage

The saliva sample is put on glass plate, which after drying of a liquid substance is located under a microscope objective. The fragment of the real image is shownin Fig.1.

It is clearly visible the following. Linear extended regular enough forms prevail, the crystals width is approximately constant, and the background is homogeneous. Also there are appendices of regular form, located in approximately constant period relative to a basic crystal body. The angle under which appendices are formed as a rule is near 90 degrees.

In the Table 1 the resultant key parameters of crystals which were used in diagnostics are specified [3].

The task consists in creating of a method and software for the automated definition of parameters, specified in the TableI. However it is necessary to notice that final resolution of equipment not always allows identifying a parent crystal for certain appendix, especially, if some parent crystals are perceived as crossed. Also, it should be taken into account that an image or image fragments can be deformed by artifacts which are not concerning crystal structure. For example, there are two similar artifacts at the bottom part of image on fig. 1.

The similar difficulties are met in research of metallographic images [4].

B.Microcrystallization saliva index (IMK).

Degree of crystallization (microcrystallization) of a saliva is described by index IMK [1], which also known as “Crystallization Indicator” (CI) [2].Thus it is useful to take advantage of

.

Here is NС – quantity of pixels within a zone of crystals and N – totalnumber ofimage pixels. It is considered to be thatfigh e are mineralizationor high CI = 1…0.6 did not have any new dental lesion, i.e. the patient is healthy. For patients withmiddle CI = 0.6…0.4a new dental lesions are expected to appear within 12-18 months.If the CI belongs to 0.4...0 interval,it is possible to assert with a high probability that caries will be detected not later than 6 months after testing.

Apparently, borders between gradations of factor CIare defined with small accuracy. It allows adjustingaccuracy of definition of the crystal area or accordingly quantity of the pixels NCwhich have coincided with the image of 0.1 or 10%.

III.Procedure of Processing

The method consists of the following procedures.

  1. Filtration in frequency or time domain for the purpose of background elimination.For definitions of boundary frequencies of the filterit is necessary to execute preliminary a spectral estimation crystallographic signals, free origins, noise and interventions. The first step of filtration is noise elimination in the form of separate points, not connected with considerable formations.Isolated points can be detected by absence of junctions with other objects.

The second step consists in image contrast increasing by statistical alignment of distribution of instant values of a signal.Then a signal has to be strengthened and limited for the purpose of transformation to binary sequence of 1/0. An example of image fragments of a crystal after a filtration is presented in Fig.2.Now the image is presented in the form of a two-dimensional file which each element contains 1 or 0.

The following step is exclusion of areas with artifacts which occur as figures essentially different from crystal structure. It is obvious that excluded areas with artifacts should be excluded from total number of pixels Nof the image also.

  1. Segmentation [5]. This stage includes computer image processing for definition of the closed crystals borders, formation of a crystals list and definition of their parameters, specified in Table I.
  2. Measurement of the crystals parameters from Table 1. To solve such not a difficult problem, it is necessary to construct the central axis of a crystal equally spaced from borders with all appendices. Then definition of parameters from the table is performed.

Calculation of the area of a crystal is made by summation of quantity of pixels with the zero value, located between two boundary pixels with the individual values located within one line.This sum collects at the analysis of all lines of the image.

The wide range of gradation CIand low accuracy of area definition allowsusing the simplified algorithm of calculations.For this purpose it is necessary to make a tentative estimation of the area of crystals without branchesof high order.That allowsconsiderably reducetime expenses.If the received result is appeared in the field of one of borders of gradation in limits ±10 % it is useful to specify result by calculation of the areas of branches. It is necessary to mean that accidental clutter in binary form transformation can lead to rupture of a continuous curve and to the big errors in crystals area definition.

IV.Conclusion

At present definition of the crystal area is performed by the operator manually by means of visualobserving of an image either in a microscope eyepiece, or a photo copy.Analysis carried out by means of transparent sheet with the scale grid. This process is very labor-intensive, the analysis occupies a long time and consequently such way is used only for scientific researches. The described technique is based on the automated definition of crystals area and their key parameters, and it can be widely introduced into daily practice in medical institutions, school dental rooms etc. for the purpose of preventive maintenance and the timely prevention of patients of possible development of caries.

As a result of work it is possible to draw the following conclusion.The method for definition of crystals parameters from the table allows estimating only roughly a chemical compound and concentration of salts as a part of saliva. So, the described method should be expanded for specification of a chemical compound by comparison of crystals forms with basic records from a databank, and also procedure of an estimation of concentration.

This system and method has been created at Institute of the Space and Information technology of Siberian Federal University and allows expediting the analysis of crystal structure of a saliva.

References

[1]G.Pancu and al. Evaluation of caries activity using the microcrystalli-zation saliva index (IMK). Rev Med Chir Soc Med Nat Iasi110(1):206-11 (2006) PMID 19292107

[2]Changes of potential for a mineralization of an oral liquid depending on a food kind. zubik.com.ua …zubov-u…potenciala20070402423.html(In Russian).

[3]A.B.Denisov. The saliva and salivary glands (In Russian). Published of the Russian academy of medical sciences, 2009.

[4]D.A.Perfil’ev, Yu.A. Maglinets and G.M, Tsibyl’sky. Family of Models for Describing One Class of Metallografic Images. Pattern Recogn. Image Annal Adv. Mrth. Theory Appl. 19 (2), 334-341 (2009).

[5]V.I. Borisenko, А.А. Zlatopol’ski, I.B. Muchnik. Image Segmentation (The State of Art). Avtomat.Telemekhan. 1987, №7.

.

Tatyiana Victorova is with the Institute of the Space and Information technology of Siberian Federal University (e-mail: )