Research ArticleIMAGING

Effect of Prefiltering Cutoff Frequency and Scatter and Attenuation Corrections During Normal Database Creation for Statistical Imaging Analysis of the Brain

Hideo Onishi, Yuki Matsutake, Norikazu Matsutomo, Yuji Kai and Hizuru Amijima
Journal of Nuclear Medicine Technology September 2011, 39 (3) 231-236; DOI: https://doi.org/10.2967/jnmt.110.079871

Article Figures & Data

Figures

  • FIGURE 1.
    FIGURE 1.

    Method of determining area of hypoperfused region extracted from patient data to brain surface from area of trapezoidal shape. We calculated distance (r1; mm) of hypoperfused lesion and distance (r2; mm) of next slice using slice thickness (d) in 3 slice images. (A) Distance r1 (mm) measured on surface of nonperfused region in each SPECT slice after anatomic standardization (voxel size, 2.25 mm). (B) Area determined using distances r1 and r2 extracted to brain surface and trapezoidal formula for trapezoidal approximation.

  • FIGURE 2.
    FIGURE 2.

    3D-SSP (A) and eZIS (B) histograms of SD in normal database obtained at each fc of Butterworth filter. Histogram distribution of SD was generated from SD of each pixel of brain using data from all participants. Each (3D-SSP, eZIS) histogram for SD scores of normal database was obtained at different fc of Butterworth filter. Horizontal axis is SD score; vertical axis is frequency.

  • FIGURE 3.
    FIGURE 3.

    3D-SSP (A) and eZIS (B) histograms of SD in normal database with and without correction. Each histogram of SD scores of normal database was obtained with and without scatter and attenuation correction. Horizontal axis is SD score; vertical axis is frequency. AC+ = with attenuation correction; SC+ = with scatter correction.

  • FIGURE 4.
    FIGURE 4.

    3D-SSP (A) and eZIS (B) z score map images of left lateral brain surface when fc of Butterworth filter of normal database is changed while patient data remain constant. Pixel values are normalized to global brain, and images show decrease in z score. z score images confirm hypoperfused region. z score in each z score map image is scaled to 10.0 from 2.0.

Tables

  • TABLE 1

    Variation in Average z Scores in Region of Interest at Various Butterworth Filter fc Settings

    Processing methodfc (cycles/cm)
    0.40.450.50.550.6
    3D-SSP14.5514.5212.8211.2410.17
    eZIS14.213.9113.5112.8312.24
  • TABLE 2

    Error in Areas of Hypoperfused Region Compared with True Values

    Processing methodfc of Butterworth filter (cycles/cm)
    0.40.450.50.550.6
    3D-SSP40.2125.657.51−9.5−13.33
    eZIS70.8666.2655.3453.0941.92

    • Units are % error of measured, compared with true, region size.

  • TABLE 3

    Evaluation of Error of Hypoperfused Region According to Differences Between Compensation Methods

    Processing methodNDB−, PD−NDB+, PD+NDB−, PD+NDB+, PD−
    ACSCACSCACSCACSC
    3D-SSP7.817.96−2.15−3.3−13.5419.27−20.58−19.79
    eZIS4.534.53−4.12−4.36−13.1710.7−27.57−23.87

    • Data are ratios. NDB− = without correction in normal database; PD− = without correction in patient data; NDB+ = with correction in normal database; PD+ = with correction in patient data.

Back to top

In this issue

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

Jump to section