Deep-Inspiration Breath-Hold PET/CT of the Thorax

MATERIALS AND METHODS

RESULTS

The patient's sex, age, site of primary disease, CT volumes, and locations of the 10 lesions detected in the DIBH PET FOV are summarized in Table 1. The average CT volume was ∼11 mL (range, 1.3–73 mL). The breathing signal of one of the patients is presented in Figure 1. Figure 1A shows the breath-hold period during the DIBH CT session, as well as the hold period for the first DIBH PET session. The hold threshold is also displayed. The lung inflation level was consistent thorough out the DIBH PET/CT study (Fig. 1B). DIBH PET/CT yielded an improved registration between PET and CT images. This is elaborated in the transaxial, coronal, and sagittal views of the fused clinical (Fig. 2A) and DIBH (Fig. 2B) PET/CT images for 1 patient. DIBH also enabled correction for diaphragm artifacts due to respiration on the CT images (Figs. 3A and 3B). When using the standard clinical CT images for attenuation correction, these artifacts translate into a region of low activity concentration due to the apparent mixing between lung and diaphragm tissues in that region (Fig. 3C). This artifact was minimized on the DIBH CT attenuation-corrected DIBH PET images (Fig. 3D). The SUV_max measurements on both clinical and DIBH PET images as well as the percent difference between the 2 measurements are reported in Table 1. DIBH yielded an increase in median SUV_max, compared with clinical measurements, of ∼32.6% (range, 4%–83%). On the other hand, the DIBH technique resulted in decreasing the mismatch between PET and CT lesions. This is elaborated in the negative percent differences between d_DIBH and d_clinical, as shown in Table 1. DIBH PET/CT enabled reducing the distance—thus the mismatch—between PET and CT lesions by a median of 26.6% (range, 3%–50%).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1. 

(A) RPM breathing signal of 1 patient showing “Hold” periods on CT and the first PET frame are displayed. Dashed line corresponds to inflation level at which patient was instructed to hold the breath. This was defined on the fly on the basis of the lung inflation level read during DIBH CT session. The “Hold” signal exhibited a relaxation period of 1–2 s before it stabilized (arrows). Therefore, it was necessary to wait for 1–2 s before starting the acquisition. (B) Consistency in lung inflation levels throughout DIBH PET/CT sessions is shown.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2. 

Transaxial, coronal, and sagittal views of clinical (A) and DIBH (B) fused PET and CT images are shown. Arrows point to a lesion in rib, based on CT images. PET lesion appears partially in lung on clinical images and just partially matching CT lesion because of respiratory motion. DIBH technique allowed improvement of coregistration between PET and CT and lesion localization.

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 3. 

DIBH enabled correction for respiratory motion artifacts in diaphragm on CT images (A and B). Because of breathing motion, tissue density in region where the diaphragm moves in and out drops. This yielded a reduced activity concentration in corresponding areas on PET images when CT attenuation correction is performed. This artifact was minimized on DIBH CT attenuation-corrected DIBH PET images.

DISCUSSION

In this study, we described the procedure to acquire and clinically implement PET/CT images with the DIBH technique. This method enabled reduced breathing-induced artifacts on PET/CT images of the thorax, which include PET/CT spatial mismatch and target blurring and, consequently, increase the SUV. In 4D PET, data are acquired in several bins, each corresponding to one breathing phase. Because of the limitation on the acquisition time per bin from a clinical practice point of view, a total of 1 min per bin is usually acquired, for a total of 10 bins—that is, a total of 10 min (2,3). Because of irregular cycles, the RPM may not trigger the acquisition process on the PET scanner, which will result in prolonging the total time the patient is on the PET table. Comparison with the standard clinical protocol (3 min/bed), the reduction in the acquisition time (1 min/bin) yields a decrease in the signal-to-noise ratio. A 4D acquisition also requires major training for the patient to follow verbal instructions (breath-in, breath-out) to maintain a regular breathing motion. Unlike radiotherapy, information about the extent of tumor motion is not required for diagnosis, and DIBH PET/CT may have major advantages over 4D PET/CT in this setting. First, no patient training was required. PET data are acquired at one single phase (EI) rather than at all phases. Hence, the patient will have to reproduce the breathing characteristics of just the breath-hold phase rather than for all phases. To ensure reproducibility of the “Hold” period at the same breathing amplitude, the physicist or technologist needs to recognize the patient's breathing pattern and then coach the patient to hold the breath at the same amplitude. The patient should be able to easily follow the instructions and reproduce the breathing amplitude, especially because he or she gets a break period (“Relax”) after every “Hold” period. Second, 3 min per bed (equals the standard clinical acquisition time per bed) could be acquired in DIBH, which corresponds to at least 30 min of acquisition time in 4D PET for a total of 10 bins, as in the protocol suggested by Nehmeh et al. (3–5). Third, in case of sudden breathing irregularities, the corresponding data may be disregarded from the final DIBH dataset, which is not possible in 4D PET. Finally, the postprocessing time is minimal; DIBH PET and DIBH CT are acquired according to the standard clinical PET/CT protocol. Consequently, no further rebinning of the CT data to spatially match the PET slices is required, as reported by Nehmeh et al. (3).

Our results showed an increase in median SUV by 32.46% (range, 4%–83%), compared with SUVs measured on the clinical images. This finding is consistent with earlier reports (2,3). DIBH PET/CT also improved the spatial registration between PET and CT images of the thorax. Our results showed an average reduction in the distance between the PET and CT lesions' centroids of ∼24% (range, 3%–49%). This resulted in minimizing breathing artifacts especially in the diaphragm. No obvious correlation could be established between the change in the SUV and that in the distance between PET and CT centroids, as a result of DIBH compared with standard clinical acquisition. This is because the change in the SUV should correlate with the tumor size, its motion amplitude, and the accurate matching with CT. In DIBH, data are acquired at just one breathing phase, which does not enable the quantitation of the motion amplitude. This is a major disadvantage over the 4D technique, especially for radiotherapy applications. The reduction in the mismatch between PET and CT should overcome the unwanted effect of spatial misalignment on attenuation correction of PET emission images (5) and also help in determining the exact anatomic location of a given lesion.

Here we have shown that the proposed DIBH method is feasible in daily practice; the potential clinical implications will be investigated in future studies.

CONCLUSION

We have described the details to acquire PET/CT data using the DIBH technique to correct for respiration-induced artifacts on the PET/CT images. Lesions in the thorax usually move due to respiration. DIBH PET/CT enabled reducing breathing motion artifacts especially in the diaphragm, increasing the SUV, and improving the coregistration between PET and CT images. Consequently, this resulted in a better localization of the lesion and a more accurate attenuation correction. DIBH is an easy technique to clinically implement and practice, which requires minimal acquisition and processing time and effort. This procedure may improve the accuracy of PET/CT diagnosis for lung lesions.