·Table of Contents
Digital Autoradiography: possibilities and applications
Institute of Nuclear Physics of Uzbek Academy of Sciences
Ulugbek, Tashkent, 702132, Uzbekistan
Application of the fast-acting digital computers and the technical means of signals treatment allowed to spread the possibilities and the resolution of activation autoradiography. Analysis of autoradiographic features is connected with the elucidation of the optical distribution of the exposed and treated photoemulsion used as of detector of the secondary beta irradiation.
In this report the possibilities of the digital image processing by the autoradoigraphic investigations of impurity and component distributions in the garnet crystal are shown. Identification of radionuclides and the qualitative evaluation of their contents were conducted by gamma-spectrum of specimens irradiated by neutrons.
The treatment of autoradiograms was conducted with the help of the dialogue system having matrix in 512x512 elements. For the interpretation of the experimental data clustering analysis methodology was used. Classification of zones on the minimum of the square mistake was conducted according to the data of histograms of the optical densities of the studying autoradiograms. It was proposed algorithm of digital treatment for reconstruction of autoradiographic features. At a minimal contrast the resolution of the method has been enhanced on degree by adaptation of methods of digital image processing (DIP) to suppress background activity.
Key words: activation autoradiography, digital image processing, background activity, elemental shifts, the impurity distribution, the resolution, radionuclide, minimal contrast, reconstruction.
As known the nuclear physical methods are characterized as analytically highly sensitive ones. One of the methods, widely spread for investigations of the spatial distribution of chemical elements in the different objects (crystals, biology, geology et al) is activation autoradiography based on the secondary beta-irradiation registration[2-5]. Usually specimens are irradiated by nuclear particles (neutrons, protons and other) and after they are placed on a photoemulsion. The photoemulsion is used as a detector of the secondary beta-irradiation. Impurities are identified by energy lines of the gamma spectra obtained and by half-life period.The impurity content in a specimens is evaluated by a relative method based on comparison of the radionuclidy activity of the element to identify with that of the reference. Gamma spectra of irradiated specimens are measured by means of a Ge(Li)-detector and a multi-channel pulse analyzer.
An analysis of autoradiography features is connected with the elucidation of optical density distribution of photoemulsion by means of photometry. But photometrization makes it difficult to study the unhomogeneous chemical elements with the little gradient of concentration ( near 20%,) '. It doesn't allow to carry out the local analysis of microarea, etc. The fine structure of chemical element distribution is interested for different technological and nature materials to have elemental shifts. The suppression of the background and betterment of linear s-Tvability are a main problem ofautoradiographic analysis.
Limitation is the possibility of allowing separate two close located grain of a film grain. Two point of identical brightness are minimum permitted, if first zero of the function the Airy for the image of one point precisely coincides a central maxim of the function the Airy for the image of the second point. Under such condition intensity in an average point of allocation of intensity of the image on 26,5 % it is less, than in each maxim'. Another criterion is differently determined:
two point sources are minimum allowed, if the flexion of density allocation of the image will be converted in zero in an average point between Gausse images of sources '.
The super resolution is possible under condition of, if the sizes of solved details in the initial image much exceed the size of one pixel. In the digital computing system used by us the sizes of a pixel have made 10000MKM2, quadrate with the side 100 micron, and sizes of a film grain 196 MKM2. The sizes of discontinuities on the autoradiographic images were not less 1 MM2.
Mechanism of creation of autoradigraphic features is based at following.
The obtained an autoradiographic image is a projection of passing a beta particles through film. I.e., autoradiographic feature represents a two-dimensional projection of three-dimensional allocation of film grains under the effect of beta particles of different energy.
The film grains represent small-sized crystals of halloid silver. If film is subjected to operation flying by through it particles, in some of grains defined changes happen which reduce in a fallout of particles of metal silver, which seem black, as they consist from shallow crystalline aggregates.
The changes of film grains by affect of radiation, are usually invisible, and all appearances are described as derivation of the hidden image.
As known film grain can contain two local defects: own imperfections of a lattice and impurity. Secondary processes, carrying on to derivation purely of the hidden image reduce in localization in an aggregate of some atoms silver near discontinuities of crystal depending on energy of particles. One aggregate, which appear effective as a unit of the hidden image, contains not more three or four atoms of silver. This circumstance explains rise of a resolution of autoradiographic analysis at small values of a radio-activity of a sample.
The main mechanisms of origin of a noise at autoradiographic analysis are reduced to following processes:
- Fluctuation of number of beta-particles, registered by unit of an area of film surface;
- Fluctuations, stipulated by irregular allocation of film grains and their characteristics;
- Presence background of a film;
- Contribution of beta-particles from hindering radionuclides.
The external manifestations of a noise on the image will depend on a spatially - frequency characteristic. At small number high-frequency component a sharpness of the image is decreased.
Thus, autoradiographic feature represents a projection of allocation shallow crystalline aggregates of silver.
At DIP the continuous analogue initial image will be transformed to the equivalent digital array. Autoradiographic image is represented as the function of two variables, specifying value of optical density. The mathematical model of the image is represented by the function, depending on two variables, varying on autoradiogram image Values of the function in the given point (x,y) the planes express quantitatively physical properties of the simulated image - it brightness (optical density).
It present work the possibilities of application of the digital image processing (DIP) in the activation autoradiography are shown for example of studing of granat crystal.
2. EXPERIMENTAL EQUIPMENT
For experimental purposes we selected GdScGa granat crystals. The selected crystals were used to cut, then the latter were polished, chemically treated end exposed to the neutron flow of reactor.
Autoradiographic features of scandium distribution got by the secondary irradiation registration of the radionuclide ^Sc , formed by nuclear reaction ^Sc^, y). The half-life period of the formed radionuclide is 83,9 days.The irradiation time is 10 h., the exposure time continious for 24 h. The interpretation of scandium distribution autoradiograms was conducted by (DIP). During the uninterrupted analogous picture transformes into the equivalent digital tract, characterized the optical density in every point of the autoradiographic feature. With help of the row of the consecutive mathematical operations the starting image is reconstructed and the important for the further analysis of the characteristics image the most substantial details, their shape, mutual arrangement and so on.
For the DIP autoradiograms interactive complex for the image treatment was used.The system allows to get 256 gradations of the grey colour, the quantity of the points decomposition 512x512.
Programme security has modulus, which on their functional purpose are subdivided into the treatment and service picture.
Digital code or the image elements content is transformed in any colour with the help of 3 tables, which operate by 3 quick digital converters. Each of this table with the quick access contains 256 words with 8 bits. Every meaning of the element content (from 0 to 225) corresponds to the colour chosen from 224 (near 16 mm) of different colours.In particulai, the system allows to get 256 gradations of the gray colour.
The service units of the software determine interactive operational mode of the system, reading of the file of the images from the disk and output in pceudo-colours on the screen oi the display. For image processing algorithms of clusterization on a minimum of a square-mistake were used.
Processing the autoradiographic image included the following operations:
- Delineating the image (content of pixels, maximum, average and minimum value of optical density on the image, calculation of an integral from a content of square);
- Construction of the histograms (fig. I):
- Plotting curves of the horizontal or vertical profile, indicating changes of values of brightness along line in the usual and logarithmic scale;
- plotting of curves, indicating statistical allocation of values of brightness;
- Nonlinear conversion of brightness (selection of "windows", equillization, degree conversions);
- Suppression of high-frequency noise (integrated operator of the Laplace, median filtering);
- Suppression low-frequency component of the images by a nominal regress method;
- Edge enhancement (differential operator of the Laplace, the operator of Sobel).
Fig 1: Initial image of scandium distribution autoradiogram
Fig 2: The image after linearization
The putting of the autoradiographic picture is carried out with help oftelevizion camera, then the analogous picture transformed into digital form.
For elimination of nonlinearity in transmission of the image operation of linearization will be carried out. The linearization of the image consists of conversion it under so that the content of points was proportionally to light exposure. To decrease highfrequencies, connected to fluctuations of a hum noise of the gauge of the image (of the telecamera), operation of smoothing is applied. Each point (x,y) of the smoothed image varies depending on a content of eight points, enclosing it, and from the previous content under the following formula:
The smoothing represents an outcome of a convolution with a matrix 3x3 coefficients of which:
1/4 in center, 1/8 above, below, from the right, 1/16 in four comers (fig. 3 ).
Fig 3: Reconstruction by filtration with different coefficients:|
a)T=16 b)T=16 c)T=12
p=30 p=28 p=0
The fluctuations of a hum noise of the gauge of the image decrease after smoothing, filtering upper highfrequencies. For exception of selected points with very small value of brightness filtering the structural unit will be carried out. This operation consists of processing by a matrix 3x3, enclosing a processed point, and finding among nine points of this matrix of a point, value of which as much as possible, with the consequent record of this content in a central point.
The value, opposite, among nine points of this matrix discovers a point with minimum value. Its content is written in a central point. In an outcome of these processings the gray scale image becomes brighter in zones of transition.
The decrease of noise in the resulting image is possible to achieve by filtering - convolution of the image with a matrix of coefficients 3x3. A principle of this filtering: a content of each point and eight points is at first read out which it enclose. Nine read out values are multiplied on nine weight coefficients, marked from A up to J, which can be positive or negative.
The filtering by a Laplacian, being a flexon on space coordinates, underlines all transitions ((( fig-4)
Fig 4: Reconstruction by a laplacian coefficients:p=1 ,T=128
Fig 5: Reconstruction by "windows":|
a)Initial image b) After a treatment
Fig 6: Reconstructed image of scandium distribution autoradiogram
Fig 7: Initial image of copper distribtuin autoradiogram
Fig 8: The image after linearization
Filtering underline of transitions consists of toting of the initial image with the Laplacian with appropriate weight. The treated thus image gives effect of a contour. The strong underline is received at P =1 and T = 128 ( fig.4 ). After processing the initial array of numbers the circumscribed programs will derivate the new numerical arrays, used for image reconstruction ( fig.6).
For to determine of resolution of autoradiographic method with DIP the following experiment was spent. On the neutron generator were irradiated of a copper plate by irregular flow of nuclear particles. Nonuniformity optical density of autoradiographic feature was inspected under the histogram of change of optical density ( fig.7 ). Then through a middle of a copper plate a slice of width 2mm was cut out and is divided on 20 specimens by the sizes 2x2 mm. On measured the gamma spectra of each specimens was constructed a curve of change of a content of a radionuclide ^Cu. Autoradiographic feature of allocation radioactive ^Cu was treated by DIP ( fig. 7 - 9 ). After DIP of autoradiographic feature from the treated image along the same bar a curve of density change is also constructed.
The resolution of method was determined as product by the first derivative from the required function and bound of an error (dI/dx) DI.
The resolution of a autoradiography method at minimum contrast in case photometry is equalled 2,7 arb.units, at use DIP- 32 arb.units.
Fig 9: Reconstructed image of copper distribution autoradiogram
|| Fig 10: The Histrograms:
1. of radioactivity change of "Cu by measured the gamma radiation;
2. of optical density change by means of photometry;
3. of optical density change after reconstruction of autoradiographic image.
On the basis of common representations about methods of processing of the image problems of reconstruction of autoradiographic image connected with a various kind by fluctuations are considered: numbers of beta-particles, registered by unit of an area of film, fluctuations, stipulated by irregular allocation of film grains and their characteristics. Using DIP it provides to suppress the background and better of linear solvability of autoradiographic analysis.
- J. W. Goodman, Introduction to Fourier optics, MCGRAW-Hill Book Company, New York,1968.
- D. S. Gafitullina and M. Kh. Ashurov. "Pattern recognition in autoradiography of optical materials", Proceedings SPIE, XRays in Materials Analysis :Novel Applications and recent Developments, Vol. 1550, pp.50-54, Jule, 1991.
- L. I. Zhuk, D. S. Gafitullina, A. A. Kist and M. Kh. Ashurov. "Activation autoradiographic study of elemental distribution in human hair". Radiation Measurements, Vol.25, No s.1-4, pp. 394-395, 1995.
- D. S. Gafitullina, "Autoradiography for investigation of the impurity distributions in the cotton leaves", 17th International conference on Nuclear tracks in solids. Book of abstracts, p. 183, Dubna, August, 1994.
- D. S. Gafitullina, M.Kh.Ashurov and B.L.Oksengendler, "Alternative growth mecnanisms in natural diamonds", NISTIR 5692, Supplement to NIST Special publication 885, pp.41s-44s, 1995.
- R. H. T. Bates and M.J.MCDonnel, Image restoration and reconstruction. Clarendon Press, Oxford, 1986.