The task of blood cells’ analysis is relevant under the diagnostics of many diseases. The classification of blood cells under hematologic diseases is the most difficult one, when along with typical white blood cells atypical cells and immature forms of cells appear (blast cells, premyelocytes, myelocytes and others) in peripheral blood.

High-technology hematology analyzers are wide-spread in clinical diagnostic laboratories. Modern analysers of 5Diff kind are able to measure 32 blood characteristics, including differential calculation of 5 main types of blood: lymphocytes, monocytes, neutrophilеs, basophilеs, eosinophilеs. But this is not enough in case of tumours of the blood system (leukemia, lymphocytoma and others). In such cases microscopic research of whole blood smear is made in which percentage distribution of blood cells of different stages of maturity is made.

Conducting of such analysis is labor-consuming and requires high qualification of the doctor, and the result estimation is subjected to the influence of subjective factors. Scientists of MEPhI departments “Computer medical systems”, “Advanced Mathematics” together with colleagues from N.N. Blokhin Russian Cancer Research Center and Institute of improvement of qualification of Federal Medical-Biological Agency of Russia have worked out a new method of diagnostics of acute leukemia on the basis of wavelet analysis (a wavelet is a mathematical function which lets analyse different frequency components of data) for classification of chromatine structure because of the necessity of improvement of the analysis of quality assessment and lowering of its labour consumption.

The specific peculiarity of the offered method is using of informative characteristics of the structure of cells’ nucleus’ chromatine which are typical for acute leukemia.

Digital microscopic depictions of blood smears of different patients with different pathology have been used as baseline data. Images were formed using microscopes with 100-times objective magnification. The photographed bloodcells were shown to experts for estimation of chromatine roughness.

The received pictures were processed digitally: on the first stage the pictures were segmented, as the result the nucleus area of the cells was evident.

The preliminary research shows that using wavelet characteristics can raise the efficiency of classification of atypical and immature forms of cells.