3d fuzzy adaptive unsupervised Bayesian segmentation for volume determination in pet

Accurate volume contouring in PET is crucial for quantitation in numerous oncology applications. The objective of this study was to assess the performance of a segmentation algorithm for automatic lesion volume delineation that allows noise modelling and have not previously been applied to PET data. The method is based on unsupervised Bayesian segmentation using an adaptive local model and a fuzzy measure. The algorithm takes into account noise, voxel's intensity and local spatial information, in order to classify a voxel as "background" or "functional volume". Its performance was compared to a reference thresholding methodology and the Fuzzy C-Means (FCM), as well as the previously proposed Fuzzy Hidden Markov Chain (FHMC) model, using realistic simulated images. Results demonstrate that the proposed algorithm performs better than all of the other three approaches for functional volume determination under different imaging conditions.