A comparative study of AML, NEGML and OSEM based on experimental and clinical 90Y-PET data using the CASToR platform

Y TTRIUM-90 ( 90 Y) is an element used in nuclear medicine for the radioembolization of hepatic tumours. 90 Y is a β − emitter whose decay also produces positrons with a very low branching ratio of 3.2 × 10 − 5 . The possibility of using the emitted positrons for a post-treatment PET imaging has been established in 2004 [1] . PET imaging using 90 Y has then gained significant interest [2] , [3] , but remains challenging because the classical reconstruction algorithms, such as OSEM 1 , introduce some problems, among which a positive bias in regions of low activity. Some algorithms allowing negative values in the image space were introduced to mitigate this effect: ABEMML 2 [4] and NEG-ML 3 [5] . More recently, van Slambrouck et al. [6] derived new formulations for these algorithms (called AML 4 and NEGML respectively) with a focus on low statistics data. To our knowledge, there is no comparative study of these algorithms based on acquired data. The purpose of this work is to evaluate the benefits of using AML and NEGML in the context of 90 Y-PET by taking advantage of the new released CASToR platform for tomographic reconstruction.