P.S. Kudryavtsev - Post-graduate Student, Department of Biomedical Engineering, Southwestern State University (SWSU), Kursk, Russia E-mail: 79Pavel_97@mail.ru A.A. Kuzmin - Associate Professor, Department of Biomedical Engineering, Southwestern State University, Kursk, Russia E-mail: ku3bmin@gmail.com S.A. Filist - Dr.Sc. (Eng.), Professor, Department of Biomedical Engineering, Southwestern State University, Kursk, Russia E-mail: SFilist@gmail.com

For statement of the diagnosis of diseases of lungs clinical and radiological symptoms and signs, and also laboratory diagnostics and additional methods of research are used. The modified method of boosting is proposed to use for classification of the photofluorographies. Built on the basis of the Viola-Jones method classifier is used as the first \"weak\" classifier, and as a second classifier is used another discriminator based on using descriptors-primitives. These primitives are approximate brightness histogram of the photofluorography in the analysis window. Aggregator of these two classifiers implements a linear combination of weak classifiers of the boosting model. The analysis of histograms of various X-rays showed that in the absence of pathological changes in the analysis window the his-tograms have a multimodal form. For pneumonia model was used low-pass filtering by means of a two-dimensional Fourier trans-form. In this case modality of brightness histogram disappears and histogram assumes a shape similar to a triangle. The analysis of the examined images enables to conclude that the real pathological formations of pneumonia have luminance histogram with the triangular form. Therefore, the second classifier handles shaped like primitives that approximate the luminance histogram in the analysis window. Aggregator classifiers makes the final decision about the correlation of fragment in the analysis window with the test class. The formation of a binary image is made as you move the image analysis window through the all picture. As an example the intellectual system intended for diagnostics of pneumonia is considered. Results of control quality check of classification on control selection showed practical coincidence to results of expert estimation.

 

1. Djudin M.V., ZHilin V.V., Kudrjavcev P.S. i dr. Sposob vydelenija kontura izobrazhenija legkikh na rentgenovskom snimke grudnojj kletki // Izvestija JUgo-Zapad­nogo gosudarstvennogo universiteta. Ser. Upravlenie, vychislitelnaja tekhnika, informatika. Medicinskoe priborostroenie. 2014. № 4. S. 107-114.
2. Akimov A.V., Sirota A.A. Razrabotka i issledovanie algoritmov raspoznavanija izobrazhenijj na osnove metoda Violy-Dzhonsa s ispolzovaniem tekhnologii vychislenijj na graficheskikh processorakh Cuda // Vestnik VGU. Ser. Sistemnyjj analiz i informa­cionnye tekhnologii. 2014. № 3. S. 105-108.
3. Djudin M.V. Tomakova R.A., Tomakov M.V. Nejjrosetevye modeli prinjatija reshenijj dlja diagnostiki zabolevanijj legkikh na osnove analiza fljuorogramm grudnojj kletki // Biomedicinskaja radioehlektronika. 2014. № 9. S. 12-16.
4. Djudin M.V., Filist S.A., Kudrjavcev P.S. Sposob vydelenija i klassifikacii konturov legkikh na izobrazhenijakh fljuorogramm grudnojj kletki // Naukoemkie tekhnologii. 2014. T. 15. № 12. S. 25-30.
5. Tomakova R.A. Proektirovanie gibridnojj nejjronnojj seti dlja analiza slozhnostrukturirovannykh medicinskikh izobrazhenijj // Sistemnyjj analiz i upravlenie v biomedicinskikh sistemakh. 2011. T. 10. № 4. S. 916-923.
6. Tomakova R.A., Maslak A.A., SHvecova N.A. Nechetkie nejjrosetevye tekhnologii dlja vydelenija segmentov s patologicheskimi obrazovanijami i morfologicheskimi strukturami na medicinskikh izobrazhenijakh // Biomedicinskaja radioehlektronika. 2012. № 4. S. 56-62.
7. Filist S.A. Djudin M.V., Zuev I.V. i dr. Avtomaticheskie klassifikatory slozhnostrukturiruemykh izobrazhenijj na osnove multimetodnykh tekhnologijj mnogokriterialnogo vybora // Voprosy radioehlektroniki. Ser. Sistemy i sredstva otobrazhenija informacii i upravlenija spectekhnikojj (SOIU). 2015. Vyp. 1. S. 130-140.
8. Filist S.A. Tomakova R.A., JAa Zar Do. Universalnye setevye modeli dlja  zadach klassifikacii biomedicinskikh dannykh // Izvestija JUgo-Zapadnogo gosudarstvennogo universiteta. 2012. № 4 (43). CH. 2. S. 44-50.