David Faul

Suppression of Metal Artifacts in CT Using a Reconstruction Procedure That Combines MAP and Projection Completion

Metal implants such as hip prostheses and dental fillings produce streak and star artifacts in the reconstructed computed tomography (CT) images. Due to these artifacts, the CT image may not be diagnostically usable. A new reconstruction procedure is proposed that reduces the streak artifacts and that might improve the diagnostic value of the CT images. The procedure starts with a maximum a posteriori (MAP) reconstruction using an iterative reconstruction algorithm and a multimodal prior. This produces an artifact-free constrained image. This constrained image is the basis for an image-based projection completion procedure. The algorithm was validated on simulations, phantom and patient data, and compared with other metal artifact reduction algorithms.

Evaluation of the attenuation properties of MR equipment for its use in a whole-body PET/MR scanner

The combination of magnetic resonance imaging (MR) and positron emission tomography (PET) scanners can provide a powerful tool for clinical diagnosis and investigation. Among the challenges of developing a combined scanner, obtaining attenuation maps for PET reconstruction is of critical importance. This requires accounting for the presence of MR hardware in the field of view. The attenuation introduced by this hardware cannot be obtained from MR data. We propose the creation of attenuation models of MR hardware, to be registered into the MR-based attenuation map prior to PET reconstruction. Two steps were followed to assess the viability of this method. First, transmission and emission measurements were performed on MR components (RF coils and medical probes). The severity of the artifacts in the reconstructed PET images was evaluated. Secondly, a high-exposure computed tomography (CT) scan was used to obtain a model of a head coil. This model was registered into the attenuation map of PET/CT scans of a uniform phantom fitted with the coil. The resulting PET images were compared to the PET/CT reconstruction in the absence of coils. The artifacts introduced by misregistration of the model were studied. The transmission scans revealed 17% count loss due to the presence of head and neck coils in the field of view. Important sources of attenuation were found in the lock, signal cables and connectors. However, the worst source of attenuation was the casing between both coils. None of the measured medical probes introduced a significant amount of attenuation. Concerning the attenuation model of the head coil, reconstructed PET images with model-based correction were comparable to the reference PET/CT reconstruction. However, inaccuracies greater than 1-2 mm in the axial positioning of the model led to important artifacts. In conclusion, the results show that model-based attenuation correction is possible. Using a high-exposure scan to create an attenuation model of the coils has been proved feasible. However, adequate registration of the model is mandatory.

Whole-Body PET/MR Imaging: Quantitative Evaluation of a Novel Model-Based MR Attenuation Correction Method Including Bone

In routine whole-body PET/MR hybrid imaging, attenuation correction (AC) is usually performed by segmentation methods based on a Dixon MR sequence providing up to 4 different tissue classes. Because of the lack of bone information with the Dixon-based MR sequence, bone is currently considered as soft tissue. Thus, the aim of this study was to evaluate a novel model-based AC method that considers bone in whole-body PET/MR imaging. Methods: The new method (“Model”) is based on a regular 4-compartment segmentation from a Dixon sequence (“Dixon”). Bone information is added using a model-based bone segmentation algorithm, which includes a set of prealigned MR image and bone mask pairs for each major body bone individually. Model was quantitatively evaluated on 20 patients who underwent whole-body PET/MR imaging. As a standard of reference, CT-based μ-maps were generated for each patient individually by nonrigid registration to the MR images based on PET/CT data. This step allowed for a quantitative comparison of all μ-maps based on a single PET emission raw dataset of the PET/MR system. Volumes of interest were drawn on normal tissue, soft-tissue lesions, and bone lesions; standardized uptake values were quantitatively compared. Results: In soft-tissue regions with background uptake, the average bias of SUVs in background volumes of interest was 2.4% ± 2.5% and 2.7% ± 2.7% for Dixon and Model, respectively, compared with CT-based AC. For bony tissue, the −25.5% ± 7.9% underestimation observed with Dixon was reduced to −4.9% ± 6.7% with Model. In bone lesions, the average underestimation was −7.4% ± 5.3% and −2.9% ± 5.8% for Dixon and Model, respectively. For soft-tissue lesions, the biases were 5.1% ± 5.1% for Dixon and 5.2% ± 5.2% for Model. Conclusion: The novel MR-based AC method for whole-body PET/MR imaging, combining Dixon-based soft-tissue segmentation and model-based bone estimation, improves PET quantification in whole-body hybrid PET/MR imaging, especially in bony tissue and nearby soft tissue.

Field of view extension and truncation correction for MR-based human attenuation correction in simultaneous MR/PET imaging

PURPOSE In quantitative PET imaging, it is critical to accurately measure and compensate for the attenuation of the photons absorbed in the tissue. While in PET/CT the linear attenuation coefficients can be easily determined from a low-dose CT-based transmission scan, in whole-body MR/PET the computation of the linear attenuation coefficients is based on the MR data. However, a constraint of the MR-based attenuation correction (AC) is the MR-inherent field-of-view (FoV) limitation due to static magnetic field (B0) inhomogeneities and gradient nonlinearities. Therefore, the MR-based human AC map may be truncated or geometrically distorted toward the edges of the FoV and, consequently, the PET reconstruction with MR-based AC may be biased. This is especially of impact laterally where the patient arms rest beside the body and are not fully considered. METHODS A method is proposed to extend the MR FoV by determining an optimal readout gradient field which locally compensates B0 inhomogeneities and gradient nonlinearities. This technique was used to reduce truncation in AC maps of 12 patients, and the impact on the PET quantification was analyzed and compared to truncated data without applying the FoV extension and additionally to an established approach of PET-based FoV extension. RESULTS The truncation artifacts in the MR-based AC maps were successfully reduced in all patients, and the mean body volume was thereby increased by 5.4%. In some cases large patient-dependent changes in SUV of up to 30% were observed in individual lesions when compared to the standard truncated attenuation map. CONCLUSIONS The proposed technique successfully extends the MR FoV in MR-based attenuation correction and shows an improvement of PET quantification in whole-body MR/PET hybrid imaging. In comparison to the PET-based completion of the truncated body contour, the proposed method is also applicable to specialized PET tracers with little uptake in the arms and might reduce the computation time by obviating the need for iterative calculations of the PET emission data beyond those required for reconstructing images.

Correlation Between Standardized Uptake Value and Apparent Diffusion Coefficient of Neoplastic Lesions Evaluated With Whole-Body Simultaneous Hybrid PET/MRI

OBJECTIVE The purpose of this study was to assess the correlation between standardized uptake value (SUV) and apparent diffusion coefficient (ADC) of neoplastic lesions in the use of a simultaneous PET/MRI hybrid system. SUBJECTS AND METHODS Twenty-four patients with known primary malignancies underwent FDG PET/CT. They then underwent whole-body PET/MRI. Diffusion-weighted imaging was performed with free breathing and a single-shot spin-echo echo-planar imaging sequence with b values of 0, 350, and 750 s/mm(2). Regions of interest were manually drawn along the contours of neoplastic lesions larger than 1 cm, which were clearly identified on PET and diffusion-weighted images. Maximum SUV (SUVmax) on PET/MRI and PET/CT images, mean SUV (SUVmean), minimum ADC (ADCmin), and mean ADC (ADCmean) were recorded on PET/MR images for each FDG-avid neoplastic soft-tissue lesion with a maximum of three lesions per patient. Pearson correlation coefficient was used to asses the following relations: SUVmax versus ADCmin on PET/MR and PET/CT images, SUVmean versus ADCmean, and ratio of SUVmax to mean liver SUV (SUV ratio) versus ADCmin. A subanalysis of patients with progressive disease versus partial treatment response was performed with the ratio of SUVmax to ADCmin for the most metabolically active lesion. RESULTS Sixty-nine neoplastic lesions (52 nonosseous lesions, 17 bone metastatic lesions) were evaluated. The mean SUVmax from PET/MRI was 7.0 ± 6.0; SUVmean, 5.6 ± 4.6; mean ADCmin, 1.10 ± 0.58; and mean ADCmean, 1.48 ± 0.72. A significant inverse Pearson correlation coefficient was found between PET/MRI SUVmax and ADCmin (r = -0.21, p = 0.04), between SUVmean and ADCmean (r = -0.18, p = 0.07), and between SUV ratio and ADCmin (r = -0.27, p = 0.01). A similar inverse Pearson correlation coefficient was found between the PET/CT SUVmax and ADCmin. Twenty of 24 patients had previously undergone PET/CT; five patients had a partial treatment response, and six had progressive disease according to Response Evaluation Criteria in Solid Tumors 1.1. The ratio between SUVmax and ADCmin was higher among patients with progressive disease than those with a partial treatment response. CONCLUSION Simultaneous PET/MRI is a promising technology for the detection of neoplastic disease. There are inverse correlations between SUVmax and ADCmin and between SUV ratio and ADCmin. Correlation coefficients between SUVmax and ADCmin from PET/MRI were similar to values obtained with SUVmax from the same-day PET/CT. Given that both SUV and ADC are related to malignancy and that the correlation between the two biomarkers is relatively weak, SUV and ADC values may offer complementary information to aid in determination of prognosis and treatment response. The combined tumoral biomarker, ratio between SUVmax and ADCmin, may be useful for assessing progressive disease versus partial treatment response.

Comparison of the accuracy of PET/CT and PET/MRI spatial registration of multiple metastatic lesions.

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.

Does reducing CT artifacts from dental implants influence the PET interpretation in PET/CT studies of oral cancer and head and neck cancer?

In patients with oral head and neck cancer, the presence of metallic dental implants produces streak artifacts in the CT images. These artifacts negate the utility of CT for the spatial localization of PET findings and may propagate through the CT-based attenuation correction into the PET images. In this study, we evaluated the efficacy of an algorithm that reduces metallic artifacts in CT images and the impact of this approach on the quantification of PET images. Methods: Fifty-one patients with and 9 without dental implants underwent a PET/CT study. CT images through the patients dental implants were reconstructed using both standard CT reconstruction and an algorithm that reduces metallic artifacts. Attenuation correction factors were calculated from both sets of CT images and applied to the PET data. The CT images were evaluated for any reduction of the artifacts. The PET images were assessed for any quantitative change introduced by metallic artifact reduction. Results: For each reconstruction, 2 regions of interest were defined in areas where the standard CT reconstruction overestimated the Hounsfield units (HU), 2 were defined in underestimated areas, and 1 was defined in a region unaffected by the artifacts. The 5 regions of interest were transferred to the other 3 reconstructions. Mean HU or mean Bq/cm3 were obtained for all regions. In the CT reconstructions, metallic artifact reduction decreased the overestimated HUs by approximately 60% and increased the underestimated HUs by approximately 90%. There was no change in quantification in the PET images between the 2 algorithms (Spearman coefficient of rank correlation, 0.99). Although the distribution of attenuation (HU) changed considerably in the CT images, the distribution of activity did not change in the PET images. Conclusion: Our study demonstrated that the algorithm can enhance the structural and spatial content of CT images in the presence of metallic artifacts. The CT artifacts do not propagate through the CT-based attenuation correction into the PET images, confirming the robustness of CT-based attenuation correction in the presence of metallic artifacts. The study also demonstrated that considerable changes in CT images do not change the PET images.

A method for calibrating the CT-based attenuation correction of PET in human tissue

The use of x-ray computed tomography (CT)-based attenuation correction for positron emission tomography (PET) in PET/CT systems requires the transformation of CT Hounsfield units (HU) to linear attenuation coefficients at 511 keV (LAC/sub 511/). This cannot be done perfectly from a single peak kilovolt (kVp) CT scan due to variability in Compton and photoelectric composition and, thus, an approximate transformation must be employed. One difficulty in this lies in accurately determining the linear attenuation coefficients (LAC) in actual human tissue. Typically, phantoms consisting of synthetic materials thought to be approximate human tissue equivalents are employed instead. A potentially more accurate approach would be to use dual kVp CT scans to estimate LAC/sub 511/ in actual human tissue and then base the single kVp transformation on these data. This approach would also permit an assessment of the dispersion of actual tissue values about the two-component trend lines typically used for the single kVp transformation. In this paper, we develop and assess this methodology.

Suppression of metal streak artifacts in CT using a MAP reconstruction procedure

Metal implants such as hip prostheses and dental fillings produce streak artifacts in the reconstructed CT images. Due to these streaks, the CT image may not be diagnostically usable. Therefore we propose a reconstruction procedure that diminishes the streak artifacts and that may improve the diagnostic value of the CT. The procedure starts with a MAP reconstruction using an iterative reconstruction algorithm and a multi-modal prior. This produces a streak-free starting image. This starting image will be the basis for a projection completion MAR procedure. The patient results are very promising but further investigation and validation is needed.

Markerless attenuation correction for carotid MRI surface receiver coils in combined PET/MR imaging.

The purpose of the study was to evaluate the effect of attenuation of MR coils on quantitative carotid PET/MR exams. Additionally, an automated attenuation correction method for flexible carotid MR coils was developed and evaluated. The attenuation of the carotid coil was measured by imaging a uniform water phantom injected with 37 MBq of 18F-FDG in a combined PET/MR scanner for 24 min with and without the coil. In the same session, an ultra-short echo time (UTE) image of the coil on top of the phantom was acquired. Using a combination of rigid and non-rigid registration, a CT-based attenuation map was registered to the UTE image of the coil for attenuation and scatter correction. After phantom validation, the effect of the carotid coil attenuation and the attenuation correction method were evaluated in five subjects. Phantom studies indicated that the overall loss of PET counts due to the coil was 6.3% with local region-of-interest (ROI) errors reaching up to 18.8%. Our registration method to correct for attenuation from the coil decreased the global error and local error (ROI) to 0.8% and 3.8%, respectively. The proposed registration method accurately captured the location and shape of the coil with a maximum spatial error of 2.6 mm. Quantitative analysis in human studies correlated with the phantom findings, but was dependent on the size of the ROI used in the analysis. MR coils result in significant error in PET quantification and thus attenuation correction is needed. The proposed strategy provides an operator-free method for attenuation and scatter correction for a flexible MRI carotid surface coil for routine clinical use.