RESUMEN
BACKGROUND: Quantitative whole-body PET/MRI relies on accurate patient-specific MRI-based attenuation correction (AC) of PET, which is a non-trivial challenge, especially for the anatomically complex head and neck region. We used a deep learning model developed for dose planning in radiation oncology to derive MRI-based attenuation maps of head and neck cancer patients and evaluated its performance on PET AC. METHODS: Eleven head and neck cancer patients, referred for radiotherapy, underwent CT followed by PET/MRI with acquisition of Dixon MRI. Both scans were performed in radiotherapy position. PET AC was performed with three different patient-specific attenuation maps derived from: (1) Dixon MRI using a deep learning network (PETDeep). (2) Dixon MRI using the vendor-provided atlas-based method (PETAtlas). (3) CT, serving as reference (PETCT). We analyzed the effect of the MRI-based AC methods on PET quantification by assessing the average voxelwise error within the entire body, and the error as a function of distance to bone/air. The error in mean uptake within anatomical regions of interest and the tumor was also assessed. RESULTS: The average (± standard deviation) PET voxel error was 0.0 ± 11.4% for PETDeep and -1.3 ± 21.8% for PETAtlas. The error in mean PET uptake in bone/air was much lower for PETDeep (-4%/12%) than for PETAtlas (-15%/84%) and PETDeep also demonstrated a more rapidly decreasing error with distance to bone/air affecting only the immediate surroundings (less than 1 cm). The regions with the largest error in mean uptake were those containing bone (mandible) and air (larynx) for both methods, and the error in tumor mean uptake was -0.6 ± 2.0% for PETDeep and -3.5 ± 4.6% for PETAtlas. CONCLUSION: The deep learning network for deriving MRI-based attenuation maps of head and neck cancer patients demonstrated accurate AC and exceeded the performance of the vendor-provided atlas-based method both overall, on a lesion-level, and in vicinity of challenging regions such as bone and air.
RESUMEN
PURPOSE: Radiotherapy planning based only on positron emission tomography/magnetic resonance imaging (PET/MRI) lacks computed tomography (CT) information required for dose calculations. In this study, a previously developed deep learning model for creating synthetic CT (sCT) from MRI in patients with head and neck cancer was evaluated in 2 scenarios: (1) using an independent external dataset, and (2) using a local dataset after an update of the model related to scanner software-induced changes to the input MRI. METHODS AND MATERIALS: Six patients from an external site and 17 patients from a local cohort were analyzed separately. Each patient underwent a CT and a PET/MRI with a Dixon MRI sequence over either one (external) or 2 (local) bed positions. For the external cohort, a previously developed deep learning model for deriving sCT from Dixon MRI was directly applied. For the local cohort, we adapted the model for an upgraded MRI acquisition using transfer learning and evaluated it in a leave-one-out process. The sCT mean absolute error for each patient was assessed. Radiotherapy dose plans based on sCT and CT were compared by assessing relevant absorbed dose differences in target volumes and organs at risk. RESULTS: The MAEs were 78 ± 13 HU and 76 ± 12 HU for the external and local cohort, respectively. For the external cohort, absorbed dose differences in target volumes were within ± 2.3% and within ± 1% in 95% of the cases. Differences in organs at risk were <2%. Similar results were obtained for the local cohort. CONCLUSIONS: We have demonstrated a robust performance of a deep learning model for deriving sCT from MRI when applied to an independent external dataset. We updated the model to accommodate a larger axial field of view and software-induced changes to the input MRI. In both scenarios dose calculations based on sCT were similar to those of CT suggesting a robust and reliable method.
RESUMEN
BACKGROUND: We present a quick and easy method to perform quantitatively accurate PET scans of typical water-filled PET plastic shell phantoms on the Siemens Biograph mMR PET/MR system. We perform regular cross-calibrations (Xcal) of our PET systems, including the PET/MR, using a Siemens mCT water phantom. LONG-TERM STABILITY: The mMR calibration stability was evaluated over a 3-year period where 54 cross-calibrations were acquired, showing that the mMR on average underestimated the concentration by 16 %, consistently due to the use of MR-based µ-maps. The mMR produced the narrowest calibration ratio range with the lowest standard deviation, implying it is the most stable of the six systems in the study over a 3-year period. MMR ACCURACY WITH PREDEFINED µ-MAPS: With the latest mMR software version, VB20P, it is possible to utilize predefined phantom µ-maps. We evaluated both the system-integrated, predefined µ-map of the long mMR water phantom and our own user-defined CT-based µ-map of the mCT water phantom, which is used for cross-calibration. For seven scans, which were reconstructed with correctly segmented µ-maps, the mMR produced cross-calibration ratios of 1.00-1.02, well within the acceptance range [0.95-1.05], showing high accuracy. CONCLUSIONS: The mMR is the most stable PET system in this study, and the mean underestimation is no longer an issue with the easily accessible µ-map, which resulted in correct cross-calibration ratios in all seven tests. We will share the user-defined µ-map of the mCT phantom and the protocol with interested mMR users.
RESUMEN
UNLABELLED: The recently released Biograph mMR is the first commercially available integrated whole-body PET/MR scanner. There are considerable advantages to integrating both modalities in a single scanner that enables truly simultaneous acquisition. However, there are also concerns about the possible degradation of both PET and MR performance in an integrated system. This paper evaluates the performance of the Biograph mMR during independent and simultaneous acquisition of PET and morphologic MR data. METHODS: The NEMA NU 2-2007 protocol was followed for studying the PET performance. The following measurements were performed: spatial resolution; scatter fraction, count losses, and randoms; sensitivity; accuracy of the correction for count losses and randoms; and image quality. The quality control manual of the American College of Radiology was followed for studying the MR performance. The following measurements were performed: geometric accuracy, spatial resolution, low-contrast detectability, signal-to-noise ratio, static field (B(0)) homogeneity, radiofrequency field (B(1)) homogeneity, and radiofrequency noise. RESULTS: An average spatial resolution of 4.3 mm in full width at half maximum was measured at 1 cm offset from the center of the field of view. The system sensitivity was 15.0 kcps/MBq along the center of the scanner. The scatter fraction was 37.9%, and the peak noise-equivalent count rate was 184 kcps at 23.1 kBq/mL. The maximum absolute value of the relative count rate error due to dead-time losses and randoms was 5.5%. The average residual error in scatter and attenuation correction was 12.1%. All MR parameters were within the tolerances defined by the American College of Radiology. B(0) inhomogeneities below 1 ppm were measured in a 120-mm radius. B(1) homogeneity and signal-to-noise ratio were equivalent to those of a standard MR scanner. No radiofrequency interference was detected. CONCLUSION: These results compare favorably with other state-of-the-art PET/CT and PET/MR scanners, indicating that the integration of the PET detectors in the MR scanner and their operation within the magnetic field do not have a perceptible impact on the overall performance. The MR subsystem performs essentially like a standalone system. However, further work is necessary to evaluate the more advanced MR applications, such as functional imaging and spectroscopy.
