Research ArticleClinical Investigations

Multi-Atlas–Based Attenuation Correction for Brain 18F-FDG PET Imaging Using a Time-of-Flight PET/MR Scanner: Comparison with Clinical Single-Atlas– and CT-Based Attenuation Correction

Tetsuro Sekine, Ninon Burgos, Geoffrey Warnock, Martin Huellner, Alfred Buck, Edwin E.G.W. ter Voert, M. Jorge Cardoso, Brian F. Hutton, Sebastien Ourselin, Patrick Veit-Haibach and Gaspar Delso
Journal of Nuclear Medicine August 2016, 57 (8) 1258-1264; DOI: https://doi.org/10.2967/jnumed.115.169045

Article Figures & Data

Figures

  • FIGURE 1.
    FIGURE 1.

    Flowchart illustrating generation of the 3 AC maps. More detailed descriptions are provided in previously published studies (6,10,12).

  • FIGURE 2.
    FIGURE 2.

    Regression plots between CT-AC and s-Atlas (A) and CT-AC and m-Atlas (B) for 67 VOIs × 15 patients.

  • FIGURE 3.
    FIGURE 3.

    Bland–Altman plots of CT-AC and s-Atlas (A) and CT-AC and m-Atlas (B) for 67 VOIs × 15 patients. Average and SD of %diff are given in Table 1.

  • FIGURE 4.
    FIGURE 4.

    Scatterplot of %diff on s-Atlas and on m-Atlas.

  • FIGURE 5.
    FIGURE 5.

    Box plot of each generalized VOI of s-Atlas (blue box) and m-Atlas (green box). In temporal lobe and cerebellum, %diff on m-Atlas is significantly smaller than that on s-Atlas. **P < 0.01 using paired t test with Bonferroni adjustment.

  • FIGURE 6.
    FIGURE 6.

    Scatterplot of each generalized VOI of %diff on s-Atlas and on m-Atlas. Dot color corresponds to color of Supplemental Figure 3. R2 and slope of best fit line of each region are given in Table 3.

  • FIGURE 7.
    FIGURE 7.

    %diff on s-Atlas and on m-Atlas in 3 representative cases, all of which were normalized to a brain template (SPM5). In pt_01, %diff on m-Atlas is much smaller than on m-Atlas, especially in cerebellar and temporal regions. In pt_02, trend of %diff is somewhat different. In cerebellum, s-Atlas underestimated 18F-FDG uptake, whereas m-Atlas overestimated it. Absolute %diff on m-Atlas is smaller than on s-Atlas. Of all 15 patients, pt_03 is only one in whom %diff on m-Atlas was apparently larger than on s-Atlas, especially in cerebellum. Bottom row is respective high-resolution T1w image for %diff map.

Tables

  • TABLE 1

    %diff and |%diff| in 18F-FDG Uptake (kBq/mL) Between CT-AC and Each s-/m-Atlas in All Regions (67 × 15 = 1,005 VOIs)

    ParameterWith TOFWithout TOF
    %diff on s-Atlas*0.17% ± 1.82%0.06% ± 2.13%
    %diff on m-Atlas*−0.06% ± 1.50%0.03% ± 1.92%
    |%diff| on s-Atlas†1.49% ± 1.06%1.71% ± 1.28%
    |%diff| on m-Atlas†1.21% ± 0.89%1.51% ± 1.19%

    • ↵* s-Atlas vs. m-Atlas, P < 0.01.

    • ↵† With TOF vs. without TOF, P < 0.01.

    • %diff is CT-AC minus s-/m-AC, divided by CT-AC. |%diff| is absolute value of %diff.

  • TABLE 2

    %diff and |%diff| in 18F-FDG Uptake (kBq/mL) Between CT-AC and Each s-/m-Atlas in Each Merged Region

    ParameterFrontal lobeOccipital lobeParietal lobeInsula and cingulateCentral structureTemporal lobe*Cerebellum*
    %diff on s-Atlas−0.87% ± 1.47%0.21% ± 1.67%−0.37% ± 1.70%−0.20% ± 1.12%0.58% ± 1.09%1.49% ± 1.37%1.55% ± 1.97%
    %diff on m-Atlas−0.64% ± 1.13%0.04% ± 1.53%−0.06% ± 1.41%0.35% ± 0.92%0.53% ± 1.04%−0.37% ± 1.41%−1.15% ± 1.72%
    |%diff| on s-Atlas1.46% ± 0.83%1.15% ± 1.19%1.42% ± 0.92%0.87% ± 0.71%0.97% ± 0.74%1.77% ± 0.95%2.11% ± 1.28%
    |%diff| on m-Atlas1.12% ± 0.61%1.25% ± 0.82%1.18% ± 0.71%0.80% ± 0.55%0.91% ± 0.69%1.15% ± 0.85%1.49% ± 1.40%

    • ↵* s-Atlas vs. m-Atlas, P < 0.01.

    • %diff is CT-AC minus s-/m-AC divided by CT-AC. |%diff| is absolute value of %diff.

  • TABLE 3

    Correlation Between %diff on s-Atlas and on m-Atlas

    ParameterAll regions (1,005 VOIs)Frontal lobeOccipital lobeParietal lobeInsula and cingulateCentral structureTemporal lobeCerebellum
    R20.2210.1880.2350.7200.3290.5880.5000.385
    Slope of the best fit line0.3880.3340.4450.7050.4730.7270.7310.541
Back to top

In this issue

Print
Download PDF
Article Alerts
Email Article
Citation Tools
Bookmark this article

Jump to section

Keywords