Premium
IC‐P‐135: The effect of head positioning and repositioning on SIENA‐generated measures of brain volume change: Results from simulated Z‐shifts
Author(s) -
Caramanos Zografos,
Fonov Vladimir,
Francis Simon J.,
Narayanan Sridar,
Collins D. Louis,
Arnold Douglas L.
Publication year - 2009
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1016/j.jalz.2009.05.107
Subject(s) - scanner , isocenter , nuclear medicine , imaging phantom , physics , volume (thermodynamics) , coordinate system , nuclear magnetic resonance , mathematics , medicine , optics , geometry , quantum mechanics
Background: Accurate, reliable quantification of brain volume change (BVC) in AD patients is important. This can be affected, however, by local-volume changes related to MRI-scanner-gradient nonlinearities and inconsistent positioning of subjects within the scanner (particularly along the Z-axis). Furthermore, standard canthomeatal (CM) alignment results in the cerebrum centered several cms away from isocenter (Fig-A). Objective: Use simulated MRI data to model the effect of Z-shift-associated gradientdistortion (GD) on SIENA-generated measures of %-BVC (PBVC). Methods: High-resolution, volumetric T1-weighted MRI data were acquired in 9 normal adults on a Siemens Sonata 1.5 T scanner after: (i) CM-alignment (CM, Fig-A), (ii) moving the scanner bed 50-mm out of the magnet (Z50), and (iii) accurate-as-possible CM repositioning (Repos). A GD field was generated using spherical harmonic expansion to map coordinates from an ‘‘ideal’’ coordinate system (a Lego-DUPLO phantom) to the imaging coordinate system of the scanner (Fig-B). For each CM image, Z-shifts were simulated in 5-mm steps from -50-mm to þ50-mm by applying the inverse of the GD, Z-shifting the image, reapplying the GD, and shifting the resulting images back. SIENA v2.5 was used to quantify the effect of a range of simulated between-scan Z-shifts similar to those seen in multicenter clinical-trial data in terms of: (i) magnitude, and (ii) location around which they were centered. Evidence for the validity of this approach is shown in Fig-C. Results: As shown in Fig-D, (i) simulated 10-mm Z-shifts resulted in absolute errors (AE) greater than Repos-vs-CM (dashed horizontal lines) when they were centered furthest into the magnet (i.e., with the cerebrum centered even farther away from isocenter); (ii) simulated Z-shifts of 20-, 30-, and 40-mm resulted in similar findings, but with AEs become progressively greater and being seen with Z-shifts centered progressively less far into the magnet; and (iii) simulated 50-mm Z-shifts consistently resulted in substantial AEs in almost all subjects throughout the range of simulations. Conclusions: Z-shift-associated GDs can have significant effects on observed SIENAPBVC values, with Z-shifts of similar magnitude having different effects depending on where they are centered. Consistent alignment of the centers of the cerebrum and magnet should decrease the effects of GD on observed BVC and increase statistical power.