Multi-scale hierarchical approach for parametric mapping: assessment on multi-compartmental models

This paper investigates a new hierarchical method to apply basis function to mono- and multi-compartmental models (Hierarchical-Basis Function Method, H-BFM) at a voxel level. This method identifies the parameters of the compartmental model in its nonlinearized version, integrating information derived at the region of interest (ROI) level by segmenting the cerebral volume based on anatomical definition or functional clustering. We present the results obtained by using a two tissue-four rate constant model with two different tracers ([(11)C]FLB457 and [carbonyl-(11)C]WAY100635), one of the most complex models used in receptor studies, especially at the voxel level. H-BFM is robust and its application on both [(11)C]FLB457 and [carbonyl-(11)C]WAY100635 allows accurate and precise parameter estimates, good quality parametric maps and a low percentage of voxels out of physiological bound (<8%). The computational time depends on the number of basis functions selected and can be compatible with clinical use (~6h for a single subject analysis). The novel method is a robust approach for PET quantification by using compartmental modeling at the voxel level. In particular, different from other proposed approaches, this method can also be used when the linearization of the model is not appropriate. We expect that applying it to clinical data will generate reliable parametric maps.