CASPIAN JOURNAL
MANAGEMENT AND HIGH TECHNOLOGIES
Methods and algorithms of analyzing chest radiographs using local windows for pathology detection
 |
Read | Malyutina I.A., Kuzmin A.A., Shatalova O.V. Methods and algorithms of analyzing chest radiographs using local windows for pathology detection // Caspian journal : management and high technologies. — 2017. — №3. — pp. 131-138. |
/131-138.jpg)
Malyutina I.A. - postgraduate, Southwest State University, 19B Chelyuskintsev St., Kursk, 305004, Russian Federation, Irina92_2010@mail.ru
Kuzmin A.A. - Cand. Sci. (Engineering), Associate Professor, Southwest State University, 19B Chelyuskintsev St., Kursk, 305004, Russian Federation, ku3bmin@gmail.com
Shatalova O.V. - Cand. Sci. (Engineering), Associate Professor, Southwest State University, 19B Chelyuskintsev St., Kursk, 305004, Russian Federation, shatРѕlg@mail.ru
Methods and algorithms for classification of images on chest radiographs based on the analysis of local image windows are proposed. At the first hierarchical level of the analysis, “weak” classifiers are built in local windows based on two approaches to the data analysis. The approach to construct a “weak” classifier by the first method is based on the analysis of amplitude Fourier spectra in a sliding window. The X-ray image is successively scanned by windows of different scale. In each window, the amplitude Fourier spectrum is determined, on the basis of which a “weak” classifier is constructed. It refers a fragment of an image in a sliding window to a certain class. The second method of constructing a “weak” classifier is based on the use of descriptors obtained as a result of approximating the brightness histograms in the analysis window. There are as much “weak” classifiers, based on two methods of analysis, as the chosen scales of the analysis windows. At the second hierarchical level of analysis, solutions of “weak” classifiers are combined within each method of the first hierarchical level analysis by building a strong classifier. The final decision “is taken” by the final classifier that aggregates the decisions of two strong classifiers of the second hierarchical level. The advantage of this approach is combination of the advantages of methods based on the analysis of the energy and structural properties of local windows.
Key words: рентгенограмма грудной клетки, классификатор, окно анализа, гистограмма яркости, двумерный спектр Фурье, агрегатор решений, алгоритмы, программное обеспечение, chest radiograph, classifier, analysis window, brightness histogram, two-dimensional Fourier sp