Accuracy of Image Coregistration of Pulmonary Lesions in Patients with Non-Small Cell Lung Cancer Using an Integrated PET/CT System

Accuracy of Image Coregistration of Pulmonary Lesions in Patients with Non-Small Cell Lung Cancer Using an Integrated PET/CT System

Gerhard W. Goerres, MD¹, Ehab Kamel, MD¹, Burkhardt Seifert, PhD², Cyrill Burger, PhD¹, Alfred Buck, MD¹, Thomas F. Hany, MD¹ and Gustav K. von Schulthess, MD, PhD¹

¹ Division of Nuclear Medicine, University Hospital Zurich, Zurich, Switzerland
² Department of Biostatistics, University of Zurich, Zurich, Switzerland

View larger version (73K):

[in a new window]

FIGURE 1. Four different lung lesions: apex, peripheral, central, and base of lung. Lesions were visually associated with a region. Apex corresponds to approximately upper 25% of lung and base corresponds to lower 25% of lung. Central was defined as area surrounding lung hila up to half of distance between hila and lateral border of lung. Peripheral was remaining lateral, anterior, and posterior space around this central area.

View larger version (18K):

[in a new window]

FIGURE 2. Box plot illustrates incongruence of COGs for all lung lesions in 4 different regions and for 2 patient groups. In this graph each box is composed of 5 horizontal lines that display the 10th, 25th, 50th, 75th, and 90th percentile. Diamonds indicate 95% confidence interval around median. B = group that breathed normally during CT scanning; H = group that held its breath in normal expiration level.

View larger version (20K):

[in a new window]

FIGURE 3. Box plot illustrates that incongruence between PET and CT for lesions in apex and in central parts of lung is less than that for lesions in periphery and in base of lung. This is the case with both respiration protocols: Breathing = group that breathed normally during CT scanning; Holding = group that held its breath in normal expiration level. Refer to legend to Figure 2 for explanation of horizontal lines and diamonds.

View larger version (16K):

[in a new window]

FIGURE 4. Taking both patient groups together, discrepancy of COGs between upper and central parts and lower and peripheral parts of lung is even more evident. Box plot shows that PET/CT match of lesions is worse in base and in periphery of lung. Refer to legend to Figure 2 for explanation of horizontal lines and diamonds.

View larger version (66K):

[in a new window]

FIGURE 5. Patient with solitary lung lesion caused by NSCLC in upper peripheral part of right lung. Coregistration of CT (A) and PET (B) images is visually adequate, but measured incongruence was 1.8 mm. This is due to different shapes of lesions, with more star-like appearance on CT and roundly shaped lesion on PET. (C) Final coregistered PET/CT image.

View larger version (112K):

[in a new window]

FIGURE 6. Coronal (A) and sagittal (B) CT (left), PET (middle), and PET/CT (right) images of patient with solitary lung lesion adjacent to diaphragmatic pleura just above liver. Lesion seen on PET image seems to be larger than lesion visible on CT image with soft-tissue windowing. Lesion cannot be discriminated from pleura or liver and direct infiltration is not ruled out with this image, although ¹⁸F-FDG uptake in lung lesion is higher than normal liver uptake.

View larger version (96K):

[in a new window]

FIGURE 7. Same lesion as in Figure 6 in transversal view. On CT image with lung window (A) lesion is larger and fits better to size seen on ¹⁸F-FDG PET image (B). (C) Incongruence in this case was measured as 6.2 mm because shape of lesion was measured on CT images with soft-tissue windowing. (D) There is perfect match of lesion on coregistered PET/CT image of this patient, who performed normal expiration protocol during CT scanning.