Cortical surface-based analysis reduces bias and variance in kinetic modeling of brain PET data

Abstract

Exploratory (i.e., voxelwise) spatial methods are commonly used in neuroimaging to identify areas that show an effect when a region-of-interest (ROI) analysis cannot be performed because no strong a priori anatomical hypothesis exists. However, noise at a single voxel is much higher than noise in a ROI making noise management critical to successful exploratory analysis. This work explores how preprocessing choices affect the bias and variability of voxelwise kinetic modeling analysis of brain positron emission tomography (PET) data. These choices include the use of volume- or cortical surface-based smoothing, level of smoothing, use of voxelwise partial volume correction (PVC), and PVC masking threshold. PVC was implemented using the Muller-Gartner method with the masking out of voxels with low gray matter (GM) partial volume fraction. Dynamic PET scans of an antagonist serotonin-4 receptor radioligand ([11C]SB2307145) were collected on sixteen healthy subjects using a Siemens HRRT PET scanner. Kinetic modeling was used to compute maps of non-displaceable binding potential (BPND) after preprocessing. The results showed a complicated interaction between smoothing, PVC, and masking on BPND estimates. Volume-based smoothing resulted in large bias and intersubject variance because it smears signal across tissue types. In some cases, PVC with volume smoothing paradoxically caused the estimated BPND to be less than when no PVC was used at all. When applied in the absence of PVC, cortical surface-based smoothing resulted in dramatically less bias and the least variance of the methods tested for smoothing levels 5mm and higher. When used in combination with PVC, surface-based smoothing minimized the bias without significantly increasing the variance. Surface-based smoothing resulted in 2-4 times less intersubject variance than when volume smoothing was used. This translates into more than 4 times fewer subjects needed in a group analysis to achieve similarly powered statistical tests. Surface-based smoothing has less bias and variance because it respects cortical geometry by smoothing the PET data only along the cortical ribbon and so does not contaminate the GM signal with that of white matter and cerebrospinal fluid. The use of surface-based analysis in PET should result in substantial improvements in the reliability and detectability of effects in exploratory PET analysis, with or without PVC.

A High-ResolutionIn VivoAtlas of the Human Brain's Serotonin System

“…Leveraging high-resolution structural MRI (ϳ1 mm resolution) in combination with molecular images acquired with a high-resolution PET scanner with a resolution of 2 mm allows for precise segmentation of brain regions and accurate intersubject normalization. The surface-based approach used in this work has also been shown to lead to a reduction in bias and variance of PET-derived measurements (Greve et al, 2013). A main advantage of the surface-based method is to diminish partial volume effects introduced by smoothing in the volume; smoothing on the surface drastically reduces the blurring of neighboring tissues with cortical gray matter and blurring across adjacent gyri (Hagler et al, 2006).…”

“…The pial surfaces were further refined using T2-weighted structural images and corrected manually where necessary. PET-MR coregistration was estimated using boundary-based registration (Greve and Fischl, 2009) between the timeweighted sum of the PET time-activity curves (TACs) and the structural MRI. Additionally, the transformation from individual MR space to normal MNI152 space was estimated with combined volume-surface (CVS) registration (Postelnicu et al, 2009).…”

“…Regional TACs for the cortical regions were extracted by resampling the TACs to the cortical surface (Greve and Fischl, 2009) and taking the average within each of the 34 regions defined by the Desikan-Killiany Greve et al (2016). In addition, a segmentation of cerebellum including different lobules (e.g., vermis) were created using SUIT 2.7 (Diedrichsen, 2006) and SPM12 ͑http://www.fil.ion.ucl.ac.uk/spm͒.…”

“…For all radioligands, modeling of the parametric and regional nondisplaceable binding potential (BP ND ) was performed using the FS PET pipeline (Greve et al, 2013) with a Multilinear Reference Tissue Model 2 (MRTM2) (Ichise et al, 2003) using cerebellar gray matter, excluding vermis, as a reference region. The reference region washout rate (k 2 Ј) was computed using MRTM (Ichise et al, 1996); the high-binding TAC was obtained from a surface-weighted average of neocortical regions for C]DASB.…”

Role of Serotonin Transporter Changes in Depressive Responses to Sex-Steroid Hormone Manipulation: A Positron Emission Tomography Study

“…As primary VOI analyses indicated that mainly cortical areas were related to GnRHa-induced depressive symptoms, we chose to conduct the analysis on a surface representation of cortex using FreeSurfer (version 5.1; A. Martinos Center for Biomedical Imaging, Boston, Massachusetts) by an approach described recently (41) and detailed in Supplement 1. Statistical analyses across subjects were performed by fitting general linear models independently at each surface node in standard space (MATLAB 8.1; Mathworks, Natick, Massachusetts).…”