OtherCONTINUING EDUCATION

Partial-Volume Effect in PET Tumor Imaging

Marine Soret, Stephen L. Bacharach and Irène Buvat
Journal of Nuclear Medicine June 2007, 48 (6) 932-945; DOI: https://doi.org/10.2967/jnumed.106.035774

Article Figures & Data

Figures

  • FIGURE 1. 
    FIGURE 1. 

    Circular source (diameter of 10 mm) of uniform activity (100 arbitrary units) in nonradioactive background yields measured image in which part of signal emanating from source is seen outside actual source. Maximum activity in measured image is reduced to 85.

  • FIGURE 2. 
    FIGURE 2. 

    Influence of image sampling on PVE. Pixels on edges of source include both source and background tissues. Signal intensity in these pixels is mean of signal intensities of underlying tissues. Part of signal emanating from source is seen outside actual object and therefore is described as spilling out.

  • FIGURE 3. 
    FIGURE 3. 

    The measured image (D) of the activity distribution (A) results from mixture of spilling out (B) and spilling in (C). Image sampling affects background activity, creating spilling in within tumor (C). Resulting image is sum of spilling in and spilling out (D).

  • FIGURE 4. 
    FIGURE 4. 

    (A) PET image. (B) Corresponding CT image. (C) PET/CT image. Discrepancy between tumor contours as seen on CT and PET images is clearly visible.

  • FIGURE 5. 
    FIGURE 5. 

    Transverse PET slice of 6 radioactive spheres with different diameters (10, 12, 16, 22, 28, and 34 mm) and filled with same radioactivity concentrations in uniform radioactive background (left) and corresponding CT slice (right). PVE makes apparent uptake decrease when sphere size decreases.

  • FIGURE 6. 
    FIGURE 6. 

    PET slice (left) and corresponding CT slice (right). Tumor is close to 3 types of tissues (lung, liver, and mediastinum).

  • FIGURE 7. 
    FIGURE 7. 

    Transverse PET slices of same PET data reconstructed with ordered-subset expectation maximization with different numbers of iterations (8 subsets). SUVmax varies substantially at early iterations.

  • FIGURE 8. 
    FIGURE 8. 

    Transverse PET slices from same patient with image sampling at 4 and 6 mm, resulting SUVmax, and corresponding CT slice.

  • FIGURE 9. 
    FIGURE 9. 

    Different measurement methods yield different SUVs. SUVmax was calculated from maximum uptake in tumor. SUV75% and SUV50% were mean values in ROI corresponding to isocontours equal to 75% and 50% SUVmax, respectively. SUV15×15 was measured in fixed rectangular region of 15 × 15 mm. SUVmean was measured in manually drawn region (represented in red on CT slice [right]).

  • FIGURE 10. 
    FIGURE 10. 

    Restored activity measured in actual contour of spheres in cold background as function of sphere diameter and spatial resolution of imaging system.

  • FIGURE 11. 
    FIGURE 11. 

    Calculation of transfer coefficients for 2 compartments (tumor [t] and background [b]). Image of each binary compartment as seen by imaging system is obtained by modeling imaging system response. Resulting image is nonbinary image from which 4 transfer coefficients can be calculated. For example, Wtt corresponds to fraction of signal emanating from tumor and detected in tumor, whereas Wtb corresponds to fraction of signal emanating from tumor and detected in background.

  • FIGURE 12. 
    FIGURE 12. 

    CT image (left), corresponding PET image (middle), and PET/CT image (right) of tumor with no uptake in center. Delineation of tumor from CT image would yield inappropriate definition of metabolically active part of tumor.

  • FIGURE 13. 
    FIGURE 13. 

    CT (A and C) and PET (B and D) images corresponding to 2 slices through lung tumor. Compartment contours as drawn from CT are shown in red on PET images.

Tables

  • TABLE 1

    Summary of Main Properties of 8 PVE Correction Approaches*

    PropertyRCGTMDeconvolutionPartition basedMultiresolutionFittingAnatomic maximum a posterioriKinetic modeling
    Assumes tumor is spheric?NoNoNoNoNoYesNoNo
    Assumes known tumor volume?YesYesNoYesNoYesYesNo
    Assumes uniform tumor uptake?YesYesNot necessarilyYesNoYesNoNo
    Assumption(s) regarding tissues surrounding tumorLocally uniform, known uptakePiecewise constant, unknown uptakeNoPiecewise constant, known uptakeGray levels correlated with those of anatomic dataLocally uniform, unknown uptakePartially knownNo
    Needs anatomic data?Not necessarilyYes, registeredNoYes, registeredYes, registeredNoYesNot necessarily
    Mode of actionAfter reconstructionAfter reconstructionAfter reconstructionAfter reconstructionAfter reconstructionAfter reconstructionDuring reconstructionAfter reconstruction, time series required
    Type of resultsTumor uptake valueTumor average uptake valueTumor average uptake valueImage of tumor compartmentPVE-corrected imageTumor average uptake valuePVE-corrected imageKinetic parameters compensated for PVE
    Reference(s) for PET tumor imaging7,17,18,22,23,50,51,53,5430373847,50

    • ↵* Desirable properties are shown in bold type.

Back to top

In this issue

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

Jump to section

Related Articles

Cited By...

More in this TOC Section

Similar Articles