Značaj multifraktalne analize snimaka magnetne rezonance u proceni odgovora na hemioterapiju kod primarnih koštanih neoplazmi

Abstract

Osteosarcom je najučestaliji primarni maligni koštani tumor, Ewing sarkom je drugi po učestalosti u populaciji mlađoj od dvadeset godina. Iako se kombinovanom terapijom u njihovom lečenju značajno povećao procenat preživljavanja kod lokalizovanih formi oboljenja i dalje najveći problem predstavlja rezistencija na hemioterapiju primarnih koštanih tumora. Otuda bi što ranija procena terapijskog odgovora bila od neprocenjivog značaja u izboru terapijskog pristupa. Patohistološki grading sistem koji se definiše nakon radikalne hirurške intervencije posle sprovedene indukcione terapije i dalje je „zlatni standard“ u proceni terapijskog odgovora u odnosu na radiološke vizuelizacione tehnike. Uvođenje i primena metoda digitalne obrade slike u pre i postterapijskoj analizi MR snimaka primarnih koštanih tumora, koja se zasniva na matematičkim modelima, mogla bi da ubrza, objektivizuje i olakša svakodnevni dijagnostički postupak. Cilj: Cilj ove studije bila je procena parametara multifraktalne analize snimaka magnetne rezonance osteosarcoma i Ewing sarcoma u predikciji odgovora na terapiju kako inicijalno pre sprovedene hemioterapije, tako i nakon nje, kao i njihove razlike. Takođe i korelacija dobijenih parametara sa patohistološkom dijagnozom stepena tumorske nekroze po Huvos grading sistemu u cilju procene značaja multifraktalne analize u predikciji hemioterapijskog odgovora. Metodologija: Studijom preseka u petogodišnjem periodu (2010-2014) obuhvaćeno je 88 pacijenata sa osteosarkomom i Ewing sarkomom na dugim cevastim kostima kod kojih je bila moguća naknadna precizna obrada MRI snimaka pre i nakon sprovedene indukcione hemioterapije. MR snimci su prevedeni u digitalnu sliku rezolucije 1400 x 1054 pixela sive skale, a naknadnom upotrebom crop tool-a, tj isecanjem delova slike izdvajan je regions of interest (ROIs) u skladu sa granicama svakog pojedinačnog tumora...

Osteosarcoma is the most frequent primary malignant bone tumour while Ewing sarcoma is the second one judging by its incidence in the population below twenty years of age. Although the combined therapy in their treatment has significantly increased the percentage of survivals in localized forms of the diseases, the resistance of primary bone tumours to chemotherapy still poses the biggest problem. Hence as early assessment of a therapy response as possible would be of invaluable importance in the selection of the therapy approach. The pathohistological grading system that is defined after a radical surgery intervention after the employed induction therapy is still the ‘golden standard’ in the assessment of therapy response as compared to radiological visualization techniques. Introduction and application of the method of digital image processing in the pre- and post-therapy analysis of MRIs of primary bone tumours, which is based on mathematical models, could speed up, objectify, and facilitate the everyday diagnostic procedure. Objective: The goal of this study was assessment of parameters of multifractal analysis of magnetic resonance imaging of osteosarcoma and Ewing sarcoma in the prediction of responses to therapy both initially prior to employing chemotherapy, and after it, as well as of their differences. Also the correlation between the obtained parameters and the pathohistological diagnosis of the degree of necrosis in tumours according to the Huvos grading system for the purpose of assessment of the importance of the multifractal analysis in the prediction of the chemotherapy response. Methodology: The cross-sectional study in the five year period (2010-2014) covered 88 patients suffering from osteosarcoma and Ewing sarcoma on long tubular bones, in which cases it was possible to subsequently precisely processMRIs prior to and after employing the induction chemotherapy. MRIs were converted into digital images of the resolution of 1400 x 1054 grayscale pixels and, by subsequent use of a crop tool, i.e. by cutting out parts of an image, regions of interest (ROIs) were singled out in line with the borders of each individual tumour...