Densitometrical volume diagnostics of spherical formations in lungs with the use of artificial neuron

Cancer, specific, nonspecific inflammatory and other pathological processes in lungs have similar lines in their visualization by multislice computed tomography methods. In order to improve the accuracy of the differential diagnosis of spherical formations in lungs by quantifying their internal structure with multislice computed tomography a newly developed method comprising sighting volume densitometry was used. 328 patients with primary cancer of a lung, pneumonia and infiltrative tuberculosis are surveyed. The set of densitometry parameters for differential diagnostics of diseases is defined. For paired division of diseases the nonlinear neuron was used. Sensitivity at division of the «primary cancer - infiltrative tuberculosis» made 91,1%, specificity - 88,6%, accuracy - 90,2%. Sensitivity at division of the «primary cancer - pneumonia» made 90,4%, specificity - 74,5%, accuracy - 84,2%. Sensitivity at division of the «infiltrative tuberculosis - pneumonia» made 79,5%, specificity - 80,8%, accuracy - 80,2%.

spherical formations of lungs, multispiral computed tomography, densitometry, structure, artificial neuron, differential diagnostics