OBTAINING THREE-DIMENSIONAL X-RAY IMAGES FROM SHADOW RADIOGRAPHS BY HOLOGRAPHIC METHODS

V. A. BERDONOSOV, V. I. GORBUNOV, A. A. POPOV, A. K. STOYANOV

For many applications of radiography in engineering and medicine, two-dimensional radiographs prove insufficient; there arises a need to obtain three-dimensional X-ray images. Existing methods and apparatus of stereoradiography are complex and have not become widely used in practice. Information of interest about spatial structure is usually obtained by methods of X-ray tomography, which gives focused images of some section. By means of layer-by-layer sections one can accumulate a sufficient number of images and form an idea of the volumetric structure of an object. Such a representation is not visual and is inconvenient also because, after recording tomograms, there is no possibility of obtaining an image of a section lying outside the fixed cuts.

Three-dimensional images of the internal structures of opaque objects can also be obtained by means of methods of non-optical holography. The absence of coherent X-ray sources and the inevitable changes in the dimensions of reconstructed images have so far made these methods of little suitability, or entirely unsuitable, for the purposes of radiation introscopy. A more suitable approach is a combination of the methods of holography and radiography. There are several ways of obtaining stereo X-ray images.

In work [1], a series of X-ray images of an object, obtained as the radiation source is displaced along an arc around the object, is successively exposed. Before each exposure the photographic plate on which the hologram is recorded, together with the reference beam, is rotated through an angle equal to the angular displacement of the source between two successive exposures. During reconstruction, such a hologram reconstructs a series of views of the object, creating the illusion of volume.

In work [2], a simpler method was proposed. At the first stage, a series of radiographs is taken, shifting the source between exposures along a straight line by a certain distance. Then, on adjacent areas of a photographic plate, holograms are recorded of the radiographs projected through a zone Fresnel plate onto a diffusing screen. During reconstruction, each area of such a hologram reconstructs an image corresponding to a definite view of the object. Superposition of these images creates a stereoscopic effect.

Work [3] considers a similar variant, differing in that, by selecting a special diffusing screen with a sufficiently broad scattering indicatrix, the need for a Fresnel plate is eliminated.

We investigated a scheme for obtaining holographic stereoroentgen images in which, in contrast to [3], the roentgenogram, when recorded on the photographic plate, was not projected onto a diffuse-scattering screen but was illuminated through it.

Such a scheme is more convenient for two reasons: first, no special requirements are imposed on the scattering characteristics of the screen; and second, when a roentgenogram is projected onto a screen, some blurring of the image occurs, and this blurring is the smaller the closer the roentgenogram is to the screen. In the variant under consideration there is no blurring.

The first stage in obtaining a stereoroentgen image consisted in obtaining a series of two-dimensional shadow roentgenograms. The radiation source was moved along a straight line with a step of 1 cm. After each exposure the reloaded cassette was placed back in its previous position. In this way, series of views were obtained of a metal spiral behind a plastic partition and of the opened hand of a human arm. At the second stage, a holographic matrix of the roentgenograms was created. The roentgenograms were recorded on a photographic plate in the same sequence in which they had been obtained. Before exposure, each of them, with the aid of adjustment devices, was set in a position common to all the images; a movable diaphragm in front of the photographic plate was shifted by a step equal to the displacement step of the x-ray source. The position of the reference beam remained unchanged; its wavefront was arbitrary.

Such a complex hologram reconstructs an x-ray image, upon viewing which a complete illusion of three-dimensionality arises. When the observation point is changed, the object seems to turn sharply, but the illusion does not disappear. By recording a larger number of roentgenograms obtained with small steps of the source displacement, it is possible to ensure that, when the observer moves his head, he will see a smooth change of views. The object appears as though it were being viewed from the side of the x-ray source.

Everything said above applies to the reconstructed virtual image. The reconstructed real image possesses properties that make it possible to carry out tomography of the object. As in ordinary holographic reconstruction, a projection screen is needed in order to observe the real image. On it, those parts of the object that fall into the plane of section by the screen are clearly delineated. Everything lying outside the section is blurred, as in x-ray tomography. An essential difference between such tomograms and ordinary ones is that the required depth of the section and its direction can be chosen at will after the roentgenograms have been obtained. The described properties of images reconstructed from a hologram matrix can be explained as follows.

If the object were transparent, the illusion of three-dimensionality would be created by the distribution of the phase of the light wave that had passed through it. On a roentgenogram, information about the phase is lost during recording. By obtaining a series of roentgenograms by displacing the source, we carry out a certain sampling from the phase distribution, the information about which will now be contained in the difference between the amplitude distributions on two-dimensional projections. The hologram simultaneously reconstructs all projections, i.e., the entire sampling from the true phase distribution. A sufficiently complete sampling makes it possible to observe a smooth parallax when the observation point is changed; a small sampling makes it jump-like. When sections of the real image are observed on a projection screen, their quality depends on the number of elements of the recorded roentgenograms of the matrix. We observed satisfactory sections already with eight elements. With a smaller number of elements, the images

the sections are washed away. The method presented for obtaining and studying stereo X-ray images requires more thorough experimental and theoretical analyses. Nevertheless, even the preliminary results, which show that such images can be obtained with sufficient simplicity and that they are convenient to observe, store, and study, allow one to hope that the method will prove useful in radiographic applications in technology and medicine.

REFERENCES

1. I. D. Redman, W. P. Wolton and E. Shuttleworth. “Nature,” 220, 58 (1968).
2. T. Kaisahara, V. Kimura, R. Hioki and S. Tanaka. “Japan. I. Appl. Phys,” 8, 124 (1969).
3. G. Groh, M. Cock. “Appl Opt,” 1970, 9, No. 3, 775–777.