DEFORMABLE REGISTRATION OF PET IN HEAD AND NECK CANCER TOMOTHERAPY TREATMENT: IS TEXTURE ANALYSIS A GOOD PREDICTOR? A.CIARMATORI, Post-graduate school in medical physics, University of Bologna, via Massarenti 9, 40138 Bologna, Italy. ciarmatori.alberto@policlinico.mo.it N.MAFFEI, C.VECCHI Physics Department, University of Bologna, via Irnerio 40 40138 Bologna, Italy G. GUIDI, B.FRANZONI, T. COSTI Medical Physics Department,University Hospital”Policlinico”, Modena Italy , via del Pozzo 71 40121 Modena, Italy guidi.gabriele@policlinico.mo.it 1. Abstract 18F-FDG positron emission tomography (PET) is widely used in Head&Neck radiation therapy target definition and tumor staging. In addition to these applications, this retrospective study focuses on the prediction of Texture Analysis respect to BTV (Biological Target Volume) deformation during radiation therapy treatment. 9 patients enrolled for Tomotherapy treatment have been included in this work. BTVs, based on Standardized Uptake Values (SUV), have been copied on treatment planning kVCT. Deformable registration have been performed along the different MVCT executed before each fraction of the treatment. SUV related texture features have been linked with tumor size variation. 8 of the 9 patients showed a substantial reduction of the size of the BTVs during the Tomotherapy treatment (range 8.6% - 33%). SUVmean and SUVmax, corrected for Partial Volume Effects, correlated (r>0.9) in case of BTVs greater than 7cc. For small lesion (<7cc) no relevant correlations have been founded. Keywords: Deformable registration, PET, Texture Analysis 2. Introduction Many studies suggest that 18 F-FDG PET may bring important advantages for staging and radiotherapy planning in Head & Neck cancer patient[1]. Image guided radiation therapy (IGRT) and adaptive radiotherapy (ART) are relatively new techniques that consider tumor motion and tumor deformation during radiation therapy planning and during intra- fraction and inter-fraction radiation treatment[2]. This is very important not only for the reproducibility of patient position between subsequent fraction but also for organ at risk and target dimensions evaluation. Tumor regression is, in fact, a major feature in radiotherapy treatment response evaluation[3]. In these context texture analysis of PET images, a tool that reflects the underlying spatial variation and heterogeneity of voxel intensities within a tumour, may yield important predictive and prognostic information[4]. The first part of this retrospective study focused on the procedure for biological target volumes (BTVs) definition on PET images of 9 patients enrolled for fractionated