Peculiarities of dynamic evaluation of globular formation outlines of the lungs with multislice computed tomography

Abstract

Background. Visualization of infiltration in lung tissue surrounding the globular formation of the lungs (GFL) determined by X-ray is one of the important points in the differential diagnosis of primary lung cancer, specific and non-specific inflammatory processes. At CT gauge body phantoms test facilities are widely used for evaluating the performance of scanners that allow the evaluation of scanner characteristics : noise, contrast sensitivity, positioning accuracy, stiffness of the radiation beam, the layer thickness, spatial resolution, etc. Aim. To develop a methodology for assessing the GFL outlines of the dynamics of multislice computed tomography (MSCT) by selecting the optimal image processing algorithms. Materials and methods. The visual analysis of two- component physical model images of the electronic window level (WL) and electronic window width (WW) was installed on the basis of the best conditions for studying a specific group of tissues. In the case of indistinct, poorly defined outlines of globular formations, visual assessment is operator-dependent and requires development and application of quantitative methods of analysis. For a quantitative description of the outlines of the image of the GFL model, a vector in a polar coordinate system coming from the center of the figure mass bounded by the outline was used. The following outline complexity measures were adopted: modified Shannon information entropy H(S(k)) for k harmonics of the normalized spectral power density S(k) of the length of oscillation of loop radius vector R(n); the number of local maxima L of signature radius vector R(n); the maximum value of the normalized power spectral density S(k); product (multiplicity) of the entropy H(S) and the number of local maxima L. Results. “Multiplicity”, “the number of local maxima” of the outline depend on the GFL geometric dimensions and cannot be used for diagnosis without first normalizing for GFL outline length. The parameters, such as “entropy” and “maximum value of the normalized power spectral density” are invariant under GFL geometric sizes and can be used for differential diagnosis at any phase of the disease.