ABSTRACT
M. Novikov. Positron-emission tomography combined with computed tomography in quantitative metabolic assessment of epithelial malignant tumors. – Manuscript.
Dissertation for the degree of candidate of medical sciences in specialty 14.01.23 – “Diagnostic radiology and radiotherapy”. Shupyk National Medical Academy of Postgraduate Education, Kyiv, 2020.
Current thesis is devoted to the problem of improving the non-invasive determination of the grade of cellular differentiation of epithelial malignant neoplasms of the head and neck, lungs, cervix using metabolic parameters of PET/CT with FDG with the development of multiparametric models.
In the dissertation, data from 97 patients with verified epithelial malignant neoplasms of the head and neck, lungs, cervix, who did not receive preliminary special oncological treatment were used. The work utilised a set of methods, which include positron emission tomography combined with computed tomography with fluorodeoxyglucose labeled with 18F, post-processing of DICOM data, morphological studies, laboratory diagnostic methods, statistical methods of data processing, in particular, the construction of multiparametric models.
Segmentations of metabolic volumes of primary tumors were conducted with four different techniques, including: SUVmax 2.5 fixed threshold technique, 41%SUVmax fixed threshold technique, liver pool threshold technique, and active contour evolution algorithm (ITK-SNAP) technique. Array of semiquantitative metabolic parameters (SUVmean, MTV, TLG), quantitative textural indices from four matrices, shape indeces and histogram indices were extracted from all metabolic volumes segmented by different algorithms.
The results of the study deepen the existing knowledge about non-invasive radiological characterization of epithelial malignant neoplasms of the head and neck, lungs, and cervix. A number of parameters (three semi-quantitative, two shape indices, four histogram indices, six texture indices) have been determined using PET/CT scans necessary for an objective metabolic assessment of these neoplasms. The optimal technique for selecting the volume of interest or segmentation (ITK-SNAP) of epithelial malignant neoplasms of the head and neck, lungs, cervix during PET/CT with FDG was determined. The dependence of metabolic parameters of epithelial malignant neoplasms of the head and neck, lungs, cervix on the method of segmentation of metabolic volumes and localization of neoplasms was studied.
Based on the array of metabolic parameters of epithelial malignant neoplasms of the head and neck, lungs and cervix, for the first time, using the group method of data handling, 12 qualification models were created, which allowed for non-invasively distinguish the grades of cell differentiation with a sensitivity of 66.7-98.1%, a specificity of 66. 7-98.8% and accuracy 70.6-97.8%.
Key words: positron-emission tomography combined with computed tomography (PET/CT), malignant epithelial tumors, quantitative metabolic parameters, image textural analysis, multiparametric modeling.
