METHOD OF DIAGNOSTICS OF LUNG CANCER BY ANALYSIS OF EXHALED AIR BY PATIENT ON THE BASIS OF ANALYSIS OF BIOELECTRIC POTENTIALS OF THE RAT OLFACTORY ANALYZER
Rossijskaya Federatsiya; ot imeni kotoroj vystupaet FOND PERSPEKTIVNYKH ISSLEDOVANIJ
发明人:
Medvedev Dmitrij Sergeevich,Медведев Дмитрий Сергеевич,Kiroj Valerij Nikolaevich,Кирой Валерий Николаевич,Ilinykh Andrej Sergeevich,Ильиных Андрей Сергеевич,Shepelev Igor Evgenevich,Шепелев Игорь Евге
申请号：
RU2017116407
公开号：
RU2666873C1
申请日:
2017.05.11
申请国别(地区):
RU
年份:
2018
代理人:
摘要：
FIELD: medicine.SUBSTANCE: invention relates to medicine, in particular to the study and analysis of gaseous biological materials, and can be used to diagnose lung cancer in humans. Method is based on an analysis of exhaled air by patient by analyzing the bioelectric potentials of the rat olfactory analyzer. For this purpose, a microelectrode matrix with working electrodes and at least one reference electrode implanted into the bone is implanted into the upper surface of the olfactory bulb of a rat. Bioelectric signals of the olfactory bulb is recorded in a given frequency range of 1–250 Hz at the time of inspiration. In the analysis of digital series, six groups of primary signs are extracted: the cross-correlation coefficients of the amplitudes of the bioelectric signal between the leads, cross-correlation coefficients of frequency amplitudes between leads, binary vectors of signal amplitudes, numerical characteristics of distribution of signal amplitudes, numerical characteristics of distribution of Fourier coefficients and the numerical characteristics of the distribution of the Daubechies-4 wavelet coefficients for each lead. Each group of features of a separate multilayer neural network (MNN) is processed. Secondary signs are formed, expressed as probabilistic estimates of classes of air samples from the MNN processing of primary signs. Each MNN training is conducted to calculate the classification weight coefficients by the back propagation error algorithm on the additionally formed array of air sample pointers given to the rat a specified number of times from a given number of sources. Air sample class is identified by calculating six groups of primary signs of the bioelectric signal. With the six MNN, calculate preliminary probabilistic estimates of the attribution of the rats to the negative class associated with the absence of a biomarker of lung cancer. Calculate preliminary estimates of attribution to a positive class associated with the presence of a l
