Projekt 11
EVALUATION OF THE APPLICATION OF RADIOMICS METHODS USING THE EXAMPLE OF A STUDY WITH NUTRITIONAL INTERVENTION IN PATIENTS WITH HEAD AND NECK CANCER RECEIVING RADIATION THERAPY
Head and neck tumors are usually treated with surgical intervention, often combined with radiation therapy. The most modern imaging methods are used in therapy to assess the anatomical position and extent of the tumor. These procedures also enable further information to be obtained. A promising strategy for the extraction of this information is represented by so-called "radiomics" methods, which define quantitative image features. Radiomics analyses comprise several steps: data acquisition, data preprocessing, data management, segmentation of relevant image regions, calculation as well as selection of quantitative image features, and the use of statistical models. During this project, radiomics analyses are to be applied using the example of a prospective study (HEADNUT study) that has already been carried out. The HEADNUT study is a randomized controlled clinical study that examined the influence of regular nutritional advice in 61 head and neck cancer patients who were irradiated in the clinic for radiation therapy and radiation oncology after surgical tumor removal. The nutritional status of all patients was examined before therapy and at the end of therapy and the tumor stage was determined in order to establish the dose concept for radiation therapy. The endpoints of the study included freedom from recurrence and survival. The patients are currently being followed up. Several CT images are available for all patients: one before the start of treatment (for treatment planning) and at least one per week for checking the patient's position. The influence of diet is of particular interest in radiomics analysis of these CT images. Several studies have shown that up to 52% of patients with head and neck tumors suffer from malnutrition at the time of diagnosis and that the proportion increases to up to 80% during radiation therapy. The potential relationship between nutrition and radiomics image features is therefore investigated in this work. However, in addition to the possible existence of multicollinearity between the image features, the reproducibility, interpretability and visualization of the results of the radiomics method represent a major challenge and will be thoroughly investigated in this project.
Doktorandin
Diana Schweizer, Institut für medizinische Biometrie und Epidemiologie, Universitätsklinikum Hamburg-Eppendorf
Projektleitung
Dr. Elisabetta Gargioni, Klinik für Strahlentherapie und Radioonkologie, Universitätsklinikum Hamburg-Eppendorf
Dr. rer. nat. Linda Krause, Institut für Medizinische Biometrie und Epidemiologie, Universitätsklinikum Hamburg-Eppendorf