A.A. Pakhomov, A.A. Potapov

Increase of quality of human organism internal structures images is necessary to obtain more perfect diagnostic information. Usually, when massive examining of patients, it is hard to adjust the diagnostic equipment and take into account the features of each particular person. In this case, obviously, multiple types of noise deformation appear and make it hard to diagnose. Often, deformation factors show up in the view of small defocusing and the noise of high level additive irregular background. Taking this into account, the authors tried to apply well-known digital post detection methods of processing of multidimensional signals from remote objects to the problem of medical images quality increase. From the mathematical point of view, even the original problem definition - the relation between the original and registered image, is a rather complicated problem and it depends on the particular device construction. In the general case, the connection of input and output images is described by Fredholm integral equation of first kind with unknown core. The authors did not go to the mathematical details of this problem but applied proven algorithms and methods directly to the obtained images. When processing, the algorithms of iterative removal of irregular background and algorithms based on frequency correction in the area of Fourier spectrum were sequentially used. The results of processing show safe increase of quality of processed medical images. Applied methods were verified on the real medical images. Both methods of single frame processing work in the real time mode (processing time 0,01 - 0,001 sec), and with special processor, this time can be decreased by a factor of ten. The combined method of its consecutive applying and the values of filtration coefficients were obtained experimentally by authors when processing a lot of real noisy images. The obtained data can be useful for building the complex information and telemedical systems