Comparison of the Accuracy of PET/CT and PET/MRI Spatial Registration of Multiple Metastatic Lesions
Abstract
OBJECTIVE. The purpose of this study was to compare the accuracy of the spatial registration of conventional PET/CT with that of hybrid PET/MRI of patients with FDG-avid metastatic lesions.
SUBJECTS AND METHODS. Thirteen patients with known metastatic lesions underwent FDG PET/CT followed by PET/MRI with a hybrid whole-body system. The inclusion criterion for tumor analysis was spherical or oval FDG-avid tumor clearly identified with both CT and MRI. The spatial coordinates (x, y, z) of the visually estimated centers of the lesions were determined for PET/CT (PET and CT independently) and PET/MRI (PET, T1-weighted gradient-echo sequence with radial stack-of-stars trajectory, T2-weighted sequence), and the b0 images of an echo-planar imaging (EPI) diffusion-weighted imaging (DWI) acquisition. All MRI sequences were performed in the axial plane with free breathing. The spatial coordinates of the estimated centers of the lesions were determined for PET and CT and PET and MRI sequences. Distance between the isocenter of the lesion on PET images and on the images obtained with the anatomic modalities was measured, and misregistration (in millimeters) was calculated. The degree of misregistration was compared between PET/CT and PET/MRI with a paired Student t test.
RESULTS. Nineteen lesions were evaluated. On PET/CT images, the average of the total misregistration in all planes of CT compared with PET was 4.13 ± 4.24 mm. On PET/MR images, lesion misregistration between PET and T1-weighted gradient-echo images had a shift of 2.41 ± 1.38 mm and between PET and b0 DW images was 5.97 ± 2.83 mm. Similar results were calculated for 11 lesions on T2-weighted images. The shift on T2-weighted images compared with PET images was 2.24 ± 1.12 mm. Paired Student t test calculations for PET/CT compared with PET/MRI T1-weighted gradient-echo images with a radial stack-of-stars trajectory, b0 DW images, and T2-weighted images showed significant differences (p < 0.05). Similar results were seen in the analysis of six lung lesions.
CONCLUSION. PET/MRI T1-weighted gradient-echo images with a radial stack-of-stars trajectory and T2-weighted images had more accurate spatial registration than PET/CT images. This may be because that the whole-body PET/MRI system used can perform simultaneous acquisition, whereas the PET/CT system acquires data sequentially. However, the EPI-based b0 DWI datasets were significantly misregistered compared with the PET/CT datasets, especially in the thorax. Radiologists reading PET/MR images should be aware of the potential for misregistration on images obtained with EPI-based DWI sequences because of inherent spatial distortion associated with this type of MRI acquisition.
Combined PET/CT is currently the standard method of imaging used to diagnose and stage many cancers and to evaluate response to therapy [1–3]. PET/CT combines both functional and anatomic imaging, allowing higher spatial resolution and 3D views. PET/CT also has the potential advantage of accurate registration of PET and CT datasets [4]. However, the accuracy of registration is subject to errors due to mechanical misalignment, voluntary patient motion, and cardiac and respiratory motion between sequential acquisitions [4–7]. Knowledge of misregistration is important in understanding the potential artifacts and limitations of PET/CT coregistered imaging [4].
PET/MRI is a newer modality that has become a topic of growing interest in the imaging community. PET/MRI has the potential to add exceptional anatomic resolution and soft-tissue contrast while lowering total patient radiation exposure. However, MRI motion artifacts due to cardiac, respiratory, and involuntary patient motion have been described [7–11]. This may lead to the same technical limitations in registration in PET/MRI as in PET/CT. We hypothesized, however, that simultaneously acquired PET/MRI data may be more precisely registered than the sequentially acquired data of PET/CT. The aim of this study was to compare the accuracy of spatial registration of PET/CT with that of PET/MRI in imaging of patients with FDG-avid, well-circumscribed (rounded or oval) metastatic lesions.
Subjects and Methods
Local institutional review board approval was obtained before this prospective study was conducted. All patients provided written informed consent for the PET/MRI examinations. All PET/CT examinations were clinically indicated. Thirteen patients (two men, 11 women; average age, 64.7 years) with known metastatic cancer underwent FDG PET/CT immediately followed by whole-body PET/MRI. PET/MRI was performed after clinically indicated PET/CT with the residual FDG in patients from the PET/CT.
PET/CT
All patients fasted for a minimum of 4 hours before imaging. Insulin was withheld 6 hours before imaging, and blood glucose concentration was verified to be less than 200 mg/dL. All patients were adults receiving a fixed 15-mCi dose of FDG. For 45 minutes after the injection, patients were instructed to sit quietly in a dimly lit room. Patients were asked to void before imaging. Acquisitions were from the base of the skull to mid thighs. PET/CT images were acquired with a Biograph mCT system (Siemens Healthcare). The CT acquisition parameters were as follows: 120 kVp, 95 mA, 5.0-mm slice width, 50-cm transaxial FOV, 512 × 512 transaxial image matrix, B40f convolution kernel. The PET acquisition parameters were as follows: 15 mCi of FDG injected, 2 minutes per bed position, 814-mm transaxial FOV, 221-mm axial FOV, 200 × 200 transaxial matrix, and 3-mm gaussian postreconstruction image filter. PET images were reconstructed with CT for attenuation correction with the attenuation-weighting ordered subsets expectation-maximization 3D algorithm at two iterations and 24 subsets. With these parameters the transaxial voxel size was 4.07 × 4.07 mm and the axial voxel size 2.03 mm.
Immediately after PET/CT, patients were imaged with the whole-body PET/MRI system. Patients had to be transported to another building for PET/MRI. Thus PET/MRI acquisition was begun 120–150 minutes after the initial injection of the radiotracer for the PET/CT examination. PET/CT images were read by board-certified nuclear physicians, who provided clinical interpretation.
PET/MRI
PET/MRI studies were performed with a Biograph mMR system (Siemens Healthcare), with which PET and MRI data are acquired simultaneously. The PET detector is composed of lutetium oxyorthosilicate scintillation crystals attached to avalanche photodiodes replacing typical photomultiplier tubes for MRI compatibility. Each block detector consists of 64 crystal elements, and each crystal measures 4 × 4 × 20 mm. In each ring are 56 block detectors, and a total of 64 detector element rings are arranged on the z-axis. The MRI unit is equipped with a 3-T magnet.
PET and MRI data were acquired simultaneously. For each bed position, an approximately 20-second breath-hold dual-echo T1-weighted gradient-recalled echo sequence was performed to acquire the MRI attenuation-correction map based on a Dixon segmentation (background, lungs, fat, soft tissue). Afterward, the diagnostic MRI sequences (T1-weighted gradient-echo, T2-weighted turbo spin-echo [TSE], and transverse echo-planar imaging (EPI) diffusion-weighted imaging [DWI]) were performed while PET data were simultaneously acquired during free breathing. The T2-weighted TSE imaging with inversion recovery (STIR) parameters were TR/TE, 6250/56; inversion time, 220 ms; bandwidth, 300 Hz/pixel; voxel size, 2.2 × 2 × 6.5 mm; generalized autocalibrating partial parallel acquisition (GRAPPA) factor, 3. DWI was performed with a single-shot spin-echo EPI sequence with parameters as follows: TR/TE, 5900/54; slice thickness, 6 mm; number of axial slices, 30; bandwidth, 1628 Hz/pixel; voxel size, 2.6 × 2.1 × 6 mm; parallel imaging GRAPPA factor, 2; fat-saturation mode; b values, 0, 350, and 750 s/mm2. Radial fat-saturated gradient-echo T1-weighted volumetric interpolated breath-hold examination (VIBE) acquisition was performed with a stack-of-stars trajectory with the following parameters: TR/TE, 4.5/2; slice thickness, 2.5 mm; flip angle, 12°; number of axial slices, 80; bandwidth, 400 Hz/pixel; voxel size, 1.4 × 1.4 × 2.5 mm; quick fat-saturation mode. For each bed position, MRI and PET acquisition lasted approximately 5–10 minutes. Almost all patients had one bed position each for the pelvis, abdomen, thorax, and head and neck. Total imaging time for three to five bed positions ranged from 45 to 60 minutes.
The PET data were reconstructed with an iterative 3D ordinary Poisson ordered subsets expectation-maximization algorithm at 3 iterations and 21 subsets and with a 4-mm gaussian postreconstruction image filter. The transaxial image matrix size was 172 × 172 with a transaxial FOV of 717.2 mm and an axial FOV of 258 mm. The transaxial voxel size for the PET images was 4.17 × 4.17 mm and the axial voxel size 2.03 mm.
Image Analysis
Inclusion criteria for tumor analysis were as follows: FDG-avid tumor clearly identified on both CT and MR images by a nuclear physician with PET/CT experience, spherical or oval shape, and uniform FDG activity. PET/CT and PET/MR images were examined retrospectively by one observer using fusion software (MIM 5.4, MIM Software). The spatial coordinates (x, y, z) of the visually estimated centers of the lesions were determined for PET/CT (PET and CT independently) and PET/MRI (PET, T1-weighted radial VIBE, T2-weighted TSE, and b0 DW images independently) in a similar method suggested by Cohade et al. [4]. In addition, a subanalysis was performed for the six lung lesions, given the known deterioration in registration of lung findings on PET/CT images [1].
Statistical Analysis
Total distance between the isocenter of the lesions was calculated between PET and each specific anatomic modality with the method of Cohade et al. [4] based on the following formula: difference = (ΔX2 + ΔY2 + ΔZ2)0.5. In addition, averages, SD, and the significance of these shifts with a paired Student t test were calculated. Values of p < 0.05 were considered statistically significant.
Results
Nineteen lesions in 13 subjects met the inclusion criteria. A representative set of images is shown in Figure 1. For PET/CT, the average misalignment on CT images compared with PET images was 4.13 ± 4.24 (SD) mm. For PET/MRI the average misalignment on T1-weighted radial VIBE images was 2.41 ± 1.38 mm and on b0 DW images was 5.97 ± 2.83 mm compared with PET images. One lesion could not be measured with b0 DWI because the patient was only able to tolerate the imaging time for limited sequences. Similar results were calculated for 11 lesions with T2-weighted TSE sequences (T2-weighted TSE sequences were not performed for all patients); misregistration on T2-weighted images compared with PET images was 2.24 ± 1.12 mm (Table 1). Paired Student t test calculations of PET/CT compared with PET/MRI T1-weighted radial VIBE, b0 DWI, and T2-weighted sequences showed significant differences (p < 0.05).
Fig. 1A —66-year-old woman with lung lesion.
A, Fused PET/CT (A), T1-weighted radial volumetric interpolated breath-hold examination (VIBE) fused PET/MR (B), and b0 diffusion-weighted fused PET/MR (C) images show more accurate registration of left lower lung lesion with T1-weighted radial VIBE PET/MRI than with PET/CT and diffusion-weighted PET/MRI.
Fig. 1B —66-year-old woman with lung lesion.
B, Fused PET/CT (A), T1-weighted radial volumetric interpolated breath-hold examination (VIBE) fused PET/MR (B), and b0 diffusion-weighted fused PET/MR (C) images show more accurate registration of left lower lung lesion with T1-weighted radial VIBE PET/MRI than with PET/CT and diffusion-weighted PET/MRI.
Fig. 1C —66-year-old woman with lung lesion.
C, Fused PET/CT (A), T1-weighted radial volumetric interpolated breath-hold examination (VIBE) fused PET/MR (B), and b0 diffusion-weighted fused PET/MR (C) images show more accurate registration of left lower lung lesion with T1-weighted radial VIBE PET/MRI than with PET/CT and diffusion-weighted PET/MRI.
| PET/MRI (n = 19) | ||||
|---|---|---|---|---|
| Axis | PET/CT | T1-Weighted Radial VIBE | T2-Weighted | b0 Diffusion-Weighted Imaging |
| X | 0.74 ± 1.88 | −0.44 ± 2.09 | −0.13 ± 1.15 | −0.44 ± 1.87 |
| Y | 1.43 ± 2.51 | 0.22 ± 1.84 | −0.08 ± 1.56 | 1.28 ± 5.90 |
| Z | 1.63 ± 4.54 | 0 | 0.95 ± 1.42 | 1.25 ± 1.96 |
| Total | 4.13 ± 4.24 | 2.41 ± 1.38 | 2.24 ± 1.12 | 5.97 ± 2.83 |
Note—VIBE = volumetric interpolated breath-hold examination.
Discussion
Whereas first-generation PET cameras lacked CT components, fully integrated PET/CT systems have become the standard of care because of the synergies of both modalities. Integrating CT reduces the total PET acquisition time and improves the precision of the attenuation correction factors [3], allowing generation of PET attenuation correction maps. The importance of accurately registering lesions between modalities that yield anatomic and metabolic information cannot be understated. Precise registration is imperative for diagnosing pathologic conditions and for avoiding errors in interpretation [12]. Exact registration is also important for attenuation correction and quantification of metabolic activity, which has prognostic significance, and is paramount for assessment of treatment response comparisons [13]. Moreover, correct spatial registration is helpful for surgical planning and delineating radiation therapy margins [14].
Whole-body PET/MRI systems have been developed. One such integrated system that performs simultaneous PET and MRI acquisition (Biograph mMR) has been approved by the U.S. Food and Drug Administration and shows early promise for oncologic evaluation [15]. PET/MRI allows correlation of anatomic MRI data and metabolic PET data, allowing joint assessment of the metabolic and anatomic responses to treatment. Lesion registration between PET and anatomic CT in sequential PET/CT scanners and PET and anatomic MRI in a near-simultaneous scheme have not been compared previously, to our knowledge. Our results show that PET/MRI T1-weighted gradient-echo and T2-weighted sequences afford significantly more accurate registration than PET/CT.
The accuracy of registration depends on the positioning accuracy of the patient-handling system and the time between acquisition of the attenuation correction data and acquisition of the PET data. The manufacturer routinely measures the accuracy and reproducibility of the PET/CT patient-positioning system before introducing the product to the market. For the Biograph mCT system, the reported positioning accuracy is less than 0.25 mm and the positioning reproducibility is 0.25 mm after traveling 100 mm and returning. For the Biograph mMR system, the table-positioning accuracy is less than 0.8 mm. These performance figures are required to avoid CT reconstruction artifacts in the case of PET/CT and PET attenuation correction artifacts in the case of PET/MRI.
Although PET/CT had superior registration compared with PET/MRI b0 DW imaging, this result was most likely caused by the well-known inherent spatial distortion of EPI sequences. Although we used b0 images because of the highest signal intensity, the observed effect can be even more pronounced at higher b values. The main contributing factors to misalignment of EPI images are eddy current–induced image distortion and nonlinearities of the gradient coils. Although many methods and algorithms, such as readout segmented EPI, have been developed in attempts to correct this problem, these methods are not flawless or require additional acquisition time [16–18]. Future research is needed to improve the registration between DW and PET images.
Despite the benefits of coregistration and attenuation correction of PET with CT data, there are limitations to PET/CT. One of the most prevalent artifacts is respiratory motion–induced misregistration. This is principally due to the free-breathing acquisition of PET data, which are acquired over 10–20 minutes, whereas CT scans are acquired during a specific stage of breathing. The difference in acquisition techniques results in breathing artifacts [19]. Common examples of respiratory artifacts on PET/CT images [20] include curvilinear cold areas at the diaphragmatic interface and misplacement of liver lesions at the base of the lungs, mimicking a lung nodule. Several techniques have been attempted to diminish these artifacts.
Given the known deterioration in registration of lung findings on PET/CT images, particularly nodules at the lung bases [4], we performed a subanalysis of lung lesions. In accordance with published data [4], the PET/CT FDG-avid lung lesions had less accurate registration (Table 2) than did all of the lesions combined (Table 1). The misregistration was less pronounced on T1 radial VIBE and T2-weighted images than on PET/CT images but was slightly worse on DWI PET/MR images, similar to the results of the all-lesions analysis. This can have diagnostic and treatment implications in the evaluation of lung cancer patients. Although our analysis included only six lung lesions, these results suggest that more motion-robust techniques with respiratory and cardiac triggering may be required.
| PET/MRI (n = 6) | ||||
|---|---|---|---|---|
| Axis | PET/CT | T1-Weighted Radial VIBE | T2-Weighted | b0 Diffusion-Weighted Imaging |
| X | 2.43 ± 2.35 | −0.52 ± 2.09 | −1.22 ± 0.40 | −0.19 ± 1.46 |
| Y | 2.88 ± 3.03 | 0.07 ± 3.17 | −0.38 ± 2.79 | −2.95 ± 7.28 |
| Z | 1.50 ± 3.67 | 0 | 0 | 0 |
| Total | 5.47 ± 3.47 | 3.22 ± 1.52 | 2.32 ± 0.67 | 6.61 ± 1.6 |
Note—Average size of FDG-avid lung lesions was 1.65 cm. VIBE = volumetric interpolated breath-hold examination.
A limitation of this study was the potential for imprecise localization of the lesion isocenter, which is subjective and operator dependent, although previous studies of PET/CT were performed with similar methods [4]. To our knowledge, there is no readily available software for automating this process. For further analyses, the use of software-based methods is under investigation. The small number of lesions was another limitation. However, despite the small number of included lesions, we found statistically significant better registration between PET images and T1-weighted gradient-recalled echo or T2-weighted TSE images than we did on PET/CT images. Accurate PET/MRI spatial registration as described in our study may also play an important role in future research endeavors, such as assessing voxel-wise correlation between PET- and MRI-derived quantitative datasets. Robust registration may also assist in applications such as PET/MRI-guided biopsy and radiation planning.
Conclusion
T1-weighted gradient-echo with radial stack-of-stars trajectory and T2-weighted TSE PET/MR images had more accurate spatial registration than did PET/CT images. This finding is likely related to the simultaneous acquisition possible with the whole-body PET/MRI system used in the study, whereas PET/CT entails sequential acquisition. However, the EPI-based b0 DWI datasets had significant misregistration compared with the PET/CT datasets, especially in the thorax. Radiologists reading PET/MR images obtained with EPI-based sequences should be aware of the potential for misregistration due to inherent spatial distortion associated with this type of MRI acquisition.
References
1.
Gambhir SS, Czernin J, Schwimmer J, Silverman DH, Coleman RE, Phelps ME. A tabulated summary of the FDG PET literature. J Nucl Med 2001; 42(5 suppl):1S–93S
2.
Czernin J, Allen-Auerbach M, Schelbert HR. Improvements in cancer staging with PET/CT: literature-based evidence as of September 2006. J Nucl Med 2007; 48(suppl 1):78S–88S
3.
Beyer T, Townsend DW, Brun T, et al. A combined PET/CT scanner for clinical oncology. J Nucl Med 2000; 41:1369–1379
4.
Cohade C, Osman M, Marshall LN, et al. PET-CT: accuracy of PET and CT spatial registration of lung lesions. Eur J Nucl Med Mol Imaging 2003; 30:721–726
5.
Goerres GW, Kamel E, Heidelberg TN, Schwitter MR, Burger C, von Schulthess GK. PET-CT image co-registration in the thorax: influence of respiration. Eur J Nucl Med Mil Imaging 2002; 29:351–360
6.
Slomka PJ, Dey D, Przetak C, Aladl UE, Baum RP. Automated 3-dimensional registration of stand-alone 18F-FDG whole-body PET with CT. J Nucl Med 2003; 44:1156–1167
7.
Mäkelä T, Clarysse P, Sipilä O, et al. A review of cardiac image registration methods. IEEE Trans Med Imaging 2002; 21:1011–1021
8.
Sachs TS, Meyer CH, Irarrazabal P, Hu BS, Nishimura DG, Macovski A. The diminishing variance algorithm for real-time reduction of motion artifacts in MRI. Magn Reson Med 1995; 34:412–422
9.
Liao, JR, Pauly, JM., Brosnan, TJ et al. Reduction of motion artifacts in cine MRI using variable-density spiral trajectories. Magn Reson Med 2007; 37:569–575
10.
Glover GH, Pauly JM. Projection reconstruction techniques for reduction of motion effects in MRI. Magn Reson Med 1992; 28:275–289
11.
Le Bihan D, Poupon C, Amadon A, et al. Artifacts and pitfalls in diffusion MRI. J Magn Reson Imaging 2006; 24:478–488
12.
Eubank WB, Mankoff DA, Schmiedl UP, et al. Imaging of oncologic patients: benefit of combined CT and FDG PET in the diagnosis of malignancy. AJR 1998; 171:1103–1110
13.
Kinahan PE, Townsend DW, Beyer T, et al. Attenuation correction for a combined 3D PET/CT scanner. Med Phys 1998; 25:2046–2053
14.
Erdi YE, Rosenzweig K, Erdi AK, et al. Radiotherapy treatment planning for patients with non-small cell lung cancer using positron emission tomography (PET). Radiother Oncol 2002; 62:51–60
15.
Schwenzer NF, Schraml C, Müller M, et al. Pulmonary lesion assessment: comparison of whole-body hybrid MR/PET and PET/CT imaging—pilot study. Radiology 2012; 264:551–558
16.
Rohde GK, Barnett AS, Basser PJ, Marenco S, Pierpaoli C. Comprehensive approach for correction of motion and distortion in diffusion-weighted MRI. Magn Res Med 2004; 51:103–114
17.
Heidemann RM, Porter DA, Anwander T, et al. Diffusion imaging in humans at 7T using readout-segmented EPI and GRAPPA. Magn Reson Med 2010; 64:9–14
18.
Frost R, Porter DA, Miller K, et al. Implementation and assessment of diffusion-weighted partial Fourier readout-segmented echo-planar imaging. Magn Reson Med 2012; 68:441–451
19.
Osman MM, Cohade C, Nakamoto Y, et al. Respiratory motion artifacts on PET emission images obtained using CT attenuation correction on PET-CT. Eur J Nucl Med Mol Imaging 2003; 30:603– 606
20.
von Schulthess GK. Normal PET and PET/CT body scans: imaging pitfalls and artifacts. In: von-Schulthess GK, ed. Clinical molecular anatomic imaging: PET, PET/CT and SPECT/CT. Baltimore, MD: Lippincott, Williams & Wilkins, 2002:252–270
Information & Authors
Information
Published In
Copyright
© American Roentgen Ray Society.
History
Submitted: May 22, 2013
Accepted: May 29, 2013
First published: October 22, 2013
Keywords
Authors
Metrics & Citations
Metrics
Citations
Export Citations
To download the citation to this article, select your reference manager software.