Studio di fattibilità di features di delta radiomica estratte da immagini kVCT giornaliere in trattamenti tomoterapici

Radiomics is a technique that provides for the extraction of quantitative mathematical informations from medical images called radiomic features. These quantitative informations are obtained analyzing the intensities and spatial distributions of pixels or voxels in regions of interest (ROIs) of medical images. They are therefore supposed to give indications about tissues properties, such as shape, size or heterogeneity, providing complementary informations that could be used for personalized treatments and giving predictions about the clinical outcome. More specifically, delta radiomics analyzes variations in radiomic features values during time, especially before and after treament, in hopes of finding associations with treatment response or indications about prognosis. My master thesis concentrates on the analysis of radiomics features that are extracted from kVCT scans executed on an abdominal phantom. These images are obtained from the Radixact Treatment Delivery System machine that integrates a kVCT imaging system, called ClearRT. This CT machine operates at IRCCS Candiolo, a hospital specialized in cancer research and treatment that is located in Candiolo (TO), where I performed my thesis work over a period of approximately nine months. My work inspects the characterization of the Radixact CT scan and the analysis of tendency of radiomic features during time, to assess their stability and reliability. To do so, I used an abdominal phantom to perform two series of kVCT images on which I segmented three regions of interest: one in the bone, the most homogeneous tissue, one in the healthy liver, and one in tumoral tissue. From these ROIs I have extracted radiomic features through a Python tool, PyRadiomics. Analyzing radiomics features tendency through time I have been able to check the temporal stability of the machine and the robustness of the features. The objective of my master thesis project was therefore to obtain preliminary results about radiomic features reliability that can be set as a basis for future clinical analysis performed on patients treated with Radixact Tomotherapy Treatment Delivery System.