Softwarové prostředí pro modelování a kvantifikaci distribuce radiofarmaka pro potřeby nukleární medicíny

Abstract

The diploma thesis summarizes the fundamentals from the field of nuclear medicine and radiation protection, and deals with the design of a model for automatic segmentation of the spleen from PET/CT images. The aim of this work is to extract brightness intensities from individual PET frames and, based on statistical analysis, to quantify the degree of distribution of radiopharmaceuticals in the spleen. A possible solution for accurate extraction of the spleen from PET images is to create a CT binary mask of a segmented object of interest and transfer it to original PET frames. The segmentation method based on active contours is tested in the MATLAB environment on fifty probands. To ensure a high-quality segmentation process, it is necessary to convert input data to the same matrix size, reduce the whole dataset to the region of interest only, and perform multimodal registration. Subsequently, it is possible to apply a method based on active contours to such modified images, which runs collaterally in all sections of both modalities at once. The quantification of the radioindicator manifestation is based on the Radiomics study and is carried out through statistical analysis including the calculation of the average value of the pixel intensity in individual frames, the determination of the maximum and minimum intensities, and the standard deviation.