Open AccessPossibilities of differential diagnostics of histologicalforms of primary lung cancer with multi-spiral computed tomography based on artificial intelligence
Author(s) -
О. В. Борисенко,
В. К. Коновалов,
А. Ф. Лазарев,
Sergey Leonov
Publication year - 2020
Publication title -
rossijskij onkologičeskij žurnal
Language(s) - English
Resource type - Journals
eISSN - 2412-9119
pISSN - 1028-9984
DOI - 10.18821/1028-9984-2019-24-3-6-102-104
Subject(s) - lung cancer , medicine , adenocarcinoma , spiral computed tomography , differential diagnosis , radiology , lung , spiral (railway) , cancer , population , computed tomography , pathology , mathematics , environmental health , mathematical analysis
The problem of lung cancer, visualized including spherical formation of the lung, is becoming increasingly important every year. In the structure of the oncological morbidity of the Russian population among men in 2018, this pathology occupied the leading position 16.9% (in women 4.0%). When analyzing the distribution of patients with lung cancer of various age groups depending on the histotype of the tumor, it was found that in most cases it is adenocarcinoma and squamous lung cancer 85%. MSCT was performed in 342 patients with spherical formation of the lung aged 45 to 80 years using computed tomographs Aquillion 64 and Asteion 4 (Toshiba Medical Systems). Digital analysis of scans was performed using the X-ray + program (Russia, Barnaul), which allows direct sampling of average pixel densities in a tabular form in selected areas of interest from DICOM files for subsequent analysis and statistical processing. The obtained densitometric indicators were received at the inputs of an artificial neural network. The effectiveness of differential diagnosis of histological forms: sensitivity 35.7 + 2.6%, specificity 40.6 + 2.6%, accuracy 76.3 + 2.3%.