Clinical Evaluation of Deep Learning for Tumor Delineation on 18F-FDG PET/CT of Head and Neck Cancer.

Artificial intelligence (AI) may decrease 18F-FDG PET/CT-based gross tumor volume (GTV) delineation variability and automate tumor-volume-derived image biomarker extraction. Hence, we aimed to identify and evaluate promising state-of-the-art deep learning methods for head and neck cancer (HNC) PET GTV delineation. Methods: We trained and evaluated deep learning methods using retrospectively included scans of HNC patients referred for radiotherapy between January 2014 and December 2019 (ISRCTN16907234). We used 3 test datasets: an internal set to compare methods, another internal set to compare AI-to-expert variability and expert interobserver variability (IOV), and an external set to compare internal and external AI-to-expert variability. Expert PET GTVs were used as the reference standard. Our benchmark IOV was measured using the PET GTV of 6 experts. The primary outcome was the Dice similarity coefficient (DSC). ANOVA was used to compare methods, a paired t test was used to compare AI-to-expert variability and expert IOV, an unpaired t test was used to compare internal and external AI-to-expert variability, and post hoc Bland-Altman analysis was used to evaluate biomarker agreement. Results: In total, 1,220 18F-FDG PET/CT scans of 1,190 patients (mean age ± SD, 63 ± 10 y; 858 men) were included, and 5 deep learning methods were trained using 5-fold cross-validation (n = 805). The nnU-Net method achieved the highest similarity (DSC, 0.80 [95% CI, 0.77-0.86]; n = 196). We found no evidence of a difference between expert IOV and AI-to-expert variability (DSC, 0.78 for AI vs. 0.82 for experts; mean difference of 0.04 [95% CI, -0.01 to 0.09]; P = 0.12; n = 64). We found no evidence of a difference between the internal and external AI-to-expert variability (DSC, 0.80 internally vs. 0.81 externally; mean difference of 0.004 [95% CI, -0.05 to 0.04]; P = 0.87; n = 125). PET GTV-derived biomarkers of AI were in good agreement with experts. Conclusion: Deep learning can be used to automate 18F-FDG PET/CT tumor-volume-derived imaging biomarkers, and the deep-learning-based volumes have the potential to assist clinical tumor volume delineation in radiation oncology.

An Investigation of Lesion Detection Accuracy for Artificial Intelligence-Based Denoising of Low-Dose 64Cu-DOTATATE PET Imaging in Patients with Neuroendocrine Neoplasms.

Frequent somatostatin receptor PET, for example, 64Cu-DOTATATE PET, is part of the diagnostic work-up of patients with neuroendocrine neoplasms (NENs), resulting in high accumulated radiation doses. Scan-related radiation exposure should be minimized in accordance with the as-low-as-reasonably achievable principle, for example, by reducing injected radiotracer activity. Previous investigations found that reducing 64Cu-DOTATATE activity to below 50 MBq results in inadequate image quality and lesion detection. We therefore investigated whether image quality and lesion detection of less than 50 MBq of 64Cu-DOTATATE PET could be restored using artificial intelligence (AI). Methods: We implemented a parameter-transferred Wasserstein generative adversarial network for patients with NENs on simulated low-dose 64Cu-DOTATATE PET images corresponding to 25% (PET25%), or about 48 MBq, of the injected activity of the reference full dose (PET100%), or about 191 MBq, to generate denoised PET images (PETAI). We included 38 patients in the training sets for network optimization. We analyzed PET intensity correlation, peak signal-to-noise ratio (PSNR), structural similarity index (SSIM), and mean-square error (MSE) of PETAI/PET100% versus PET25%/PET100% Two readers assessed Likert scale-defined image quality (1, very poor; 2, poor; 3, moderate; 4, good; 5, excellent) and identified lesion-suspicious foci on PETAI and PET100% in a subset of the patients with no more than 20 lesions per organ (n = 33) to allow comparison of all foci on a 1:1 basis. Detected foci were scored (C1, definite lesion; C0, lesion-suspicious focus) and matched with PET100% as the reference. True-positive (TP), false-positive (FP), and false-negative (FN) lesions were assessed. Results: For PETAI/PET100% versus PET25%/PET100%, PET intensity correlation had a goodness-of-fit value of 0.94 versus 0.81, PSNR was 58.1 versus 53.0, SSIM was 0.908 versus 0.899, and MSE was 2.6 versus 4.7. Likert scale-defined image quality was rated good or excellent in 33 of 33 and 32 of 33 patients on PET100% and PETAI, respectively. Total number of detected lesions was 118 on PET100% and 115 on PETAI Only 78 PETAI lesions were TP, 40 were FN, and 37 were FP, yielding detection sensitivity (TP/(TP+FN)) and a false discovery rate (FP/(TP+FP)) of 66% (78/118) and 32% (37/115), respectively. In 62% (23/37) of cases, the FP lesion was scored C1, suggesting a definite lesion. Conclusion: PETAI improved visual similarity with PET100% compared with PET25%, and PETAI and PET100% had similar Likert scale-defined image quality. However, lesion detection analysis performed by physicians showed high proportions of FP and FN lesions on PETAI, highlighting the need for clinical validation of AI algorithms.

The performance of machine learning approaches for attenuation correction of PET in neuroimaging: A meta-analysis.

PURPOSE: This systematic review provides a consensus on the clinical feasibility of machine learning (ML) methods for brain PET attenuation correction (AC). Performance of ML-AC were compared to clinical standards. METHODS: Two hundred and eighty studies were identified through electronic searches of brain PET studies published between January 1, 2008, and August 1, 2022. Reported outcomes for image quality, tissue classification performance, regional and global bias were extracted to evaluate ML-AC performance. Methodological quality of included studies and the quality of evidence of analysed outcomes were assessed using QUADAS-2 and GRADE, respectively. RESULTS: A total of 19 studies (2371 participants) met the inclusion criteria. Overall, the global bias of ML methods was 0.76 ± 1.2%. For image quality, the relative mean square error (RMSE) was 0.20 ± 0.4 while for tissues classification, the Dice similarity coefficient (DSC) for bone/soft tissue/air were 0.82 ± 0.1 / 0.95 ± 0.03 / 0.85 ± 0.14. CONCLUSIONS: In general, ML-AC performance is within acceptable limits for clinical PET imaging. The sparse information on ML-AC robustness and its limited qualitative clinical evaluation may hinder clinical implementation in neuroimaging, especially for PET/MRI or emerging brain PET systems where standard AC approaches are not readily available.

Deep learning for Dixon MRI-based attenuation correction in PET/MRI of head and neck cancer patients.

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.

Robustness and Generalizability of Deep Learning Synthetic Computed Tomography for Positron Emission Tomography/Magnetic Resonance Imaging-Based Radiation Therapy Planning of Patients With Head and Neck Cancer.

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.

Deep-learning-based attenuation correction in dynamic [15O]H2O studies using PET/MRI in healthy volunteers.

Quantitative [15O]H2O positron emission tomography (PET) is the accepted reference method for regional cerebral blood flow (rCBF) quantification. To perform reliable quantitative [15O]H2O-PET studies in PET/MRI scanners, MRI-based attenuation-correction (MRAC) is required. Our aim was to compare two MRAC methods (RESOLUTE and DeepUTE) based on ultrashort echo-time with computed tomography-based reference standard AC (CTAC) in dynamic and static [15O]H2O-PET. We compared rCBF from quantitative perfusion maps and activity concentration distribution from static images between AC methods in 25 resting [15O]H2O-PET scans from 14 healthy men at whole-brain, regions of interest and voxel-wise levels. Average whole-brain CBF was 39.9 ± 6.0, 39.0 ± 5.8 and 40.0 ± 5.6 ml/100 g/min for CTAC, RESOLUTE and DeepUTE corrected studies respectively. RESOLUTE underestimated whole-brain CBF by 2.1 ± 1.50% and rCBF in all regions of interest (range -2.4%- -1%) compared to CTAC. DeepUTE showed significant rCBF overestimation only in the occipital lobe (0.6 ± 1.1%). Both MRAC methods showed excellent correlation on rCBF and activity concentration with CTAC, with slopes of linear regression lines between 0.97 and 1.01 and R2 over 0.99. In conclusion, RESOLUTE and DeepUTE provide AC information comparable to CTAC in dynamic [15O]H2O-PET but RESOLUTE is associated with a small but systematic underestimation.

Feasibility of Multiparametric Positron Emission Tomography/Magnetic Resonance Imaging as a One-Stop Shop for Radiation Therapy Planning for Patients with Head and Neck Cancer.

PURPOSE: Multiparametric positron emission tomography (PET)/magnetic resonance imaging (MRI) as a one-stop shop for radiation therapy (RT) planning has great potential but is technically challenging. We studied the feasibility of performing multiparametric PET/MRI of patients with head and neck cancer (HNC) in RT treatment position. As a step toward planning RT based solely on PET/MRI, a deep learning approach was employed to generate synthetic computed tomography (sCT) from MRI. This was subsequently evaluated for dose calculation and PET attenuation correction (AC). METHODS AND MATERIALS: Eleven patients, including 3 pilot patients referred for RT of HNC, underwent PET/MRI in treatment position after a routine fluorodeoxyglucose-PET/CT planning scan. The PET/MRI scan protocol included multiparametric imaging. A convolutional neural network was trained in a leave-one-out process to predict sCT from the Dixon MRI. The clinical CT-based dose plans were recalculated on sCT, and the plans were compared in terms of relative differences in mean, maximum, near-maximum, and near-minimum absorbed doses for different volumes of interest. Comparisons between PET with sCT-based AC and PET with CT-based AC were assessed based on the relative differences in mean and maximum standardized uptake values (SUVmean and SUVmax) from the PET-positive volumes. RESULTS: All 11 patients underwent PET/MRI in RT treatment position. Apart from the 3 pilots, full multiparametric imaging was completed in 45 minutes for 7 out of 8 patients. One patient terminated the examination after 30 minutes. With the exception of 1 patient with an inserted tracheostomy tube, all dosimetric parameters of the sCT-based dose plans were within ±1% of the CT-based dose plans. For PET, the mean difference was 0.4 ± 1.2% for SUVmean and -0.5 ± 1.0% for SUVmax. CONCLUSIONS: Performing multiparametric PET/MRI of patients with HNC in RT treatment position was clinically feasible. The sCT generation resulted in AC of PET and dose calculations sufficiently accurate for clinical use. These results are an important step toward using multiparametric PET/MRI as a one-stop shop for personalized RT planning.

Comparison of simultaneous arterial spin labeling MRI and 15O-H2O PET measurements of regional cerebral blood flow in rest and altered perfusion states.

Arterial spin labelling (ASL) is a non-invasive magnetic resonance imaging (MRI) technique that may provide fully quantitative regional cerebral blood flow (rCBF) images. However, before its application in clinical routine, ASL needs to be validated against the clinical gold standard, 15O-H2O positron emission tomography (PET). We aimed to compare the two techniques by performing simultaneous quantitative ASL-MRI and 15O-H2O-PET examinations in a hybrid PET/MRI scanner. Duplicate rCBF measurements were performed in healthy young subjects (n = 14) in rest, during hyperventilation, and after acetazolamide (post-ACZ), yielding 63 combined PET/MRI datasets in total. Average global CBF by ASL-MRI and 15O-H2O-PET was not significantly different in any state (40.0 ± 6.5 and 40.6 ± 4.1 mL/100 g/min, respectively in rest, 24.5 ± 5.1 and 23.4 ± 4.8 mL/100 g/min, respectively, during hyperventilation, and 59.1 ± 10.4 and 64.7 ± 10.0 mL/100 g/min, respectively, post-ACZ). Overall, strong correlation between the two methods was found across all states (slope = 1.01, R2 = 0.82), while the correlations within individual states and of reactivity measures were weaker, in particular in rest (R2 = 0.05, p = 0.03). Regional distribution was similar, although ASL yielded higher perfusion and absolute reactivity in highly vascularized areas. In conclusion, ASL-MRI and 15O-H2O-PET measurements of rCBF are highly correlated across different perfusion states, but with variable correlation within and between hemodynamic states, and systematic differences in regional distribution.

Hybrid PET/MRI imaging in healthy unsedated newborn infants with quantitative rCBF measurements using 15O-water PET.

In this study, a new hybrid PET/MRI method for quantitative regional cerebral blood flow (rCBF) measurements in healthy newborn infants was assessed and the low values of rCBF in white matter previously obtained by arterial spin labeling (ASL) were tested. Four healthy full-term newborn subjects were scanned in a PET/MRI scanner during natural sleep after median intravenous injection of 14 MBq 15O-water. Regional CBF was quantified using a one-tissue-compartment model employing an image-derived input function (IDIF) from the left ventricle. PET rCBF showed the highest values in the thalami, mesencephalon and brain stem and the lowest in cortex and unmyelinated white matter. The average global CBF was 17.8 ml/100 g/min. The average frontal and occipital unmyelinated white matter CBF was 10.3 ml/100 g/min and average thalamic CBF 31.3 ml/100 g/min. The average white matter/thalamic ratio CBF was 0.36, significantly higher than previous ASL data. The rCBF ASL measurements were all unsuccessful primarily owing to subject movement. In this study, we demonstrated for the first time, a minimally invasive PET/MRI method using low activity 15O-water PET for quantitative rCBF assessment in unsedated healthy newborn infants and found a white/grey matter CBF ratio similar to that of the adult human brain.

Reproducibility of MR-Based Attenuation Maps in PET/MRI and the Impact on PET Quantification in Lung Cancer.

Quantitative PET/MRI is dependent on reliable and reproducible MR-based attenuation correction (MR-AC). In this study, we evaluated the quality of current vendor-provided thoracic MR-AC maps and further investigated the reproducibility of their impact on 18F-FDG PET quantification in patients with non-small cell lung cancer. Methods: Eleven patients with inoperable non-small cell lung cancer underwent 2-5 thoracic PET/MRI scan-rescan examinations within 22 d. 18F-FDG PET data were acquired along with 2 Dixon MR-AC maps for each examination. Two PET images (PETA and PETB) were reconstructed using identical PET emission data but with MR-AC from these intrasubject repeated attenuation maps. In total, 90 MR-AC maps were evaluated visually for quality and the occurrence of categorized artifacts by 2 PET/MRI-experienced physicians. Each tumor was outlined by a volume of interest (40% isocontour of maximum) on PETA, which was then projected onto the corresponding PETB SUVmean and SUVmax were assessed from the PET images. Within-examination coefficients of variation and Bland-Altman analyses were conducted for the assessment of SUV variations between PETA and PETBResults: Image artifacts were observed in 86% of the MR-AC maps, and 30% of the MR-AC maps were subjectively expected to affect the tumor SUV. SUVmean and SUVmax resulted in coefficients of variation of 5.6% and 6.6%, respectively, and scan-rescan SUV variations were within ±20% in 95% of the cases. Substantial SUV variations were seen mainly for scan-rescan examinations affected by respiratory motion. Conclusion: Artifacts occur frequently in standard thoracic MR-AC maps, affecting the reproducibility of PET/MRI. These, in combination with other well-known sources of error associated with PET/MRI examinations, lead to inconsistent SUV measurements in serial studies, which may affect the reliability of therapy response assessment. A thorough visual inspection of the thoracic MR-AC map and Dixon images from which it is derived remains crucial for the detection of MR-AC artifacts that may influence the reliability of SUV.

RESOLUTE PET/MRI Attenuation Correction for O-(2-18F-fluoroethyl)-L-tyrosine (FET) in Brain Tumor Patients with Metal Implants.

Aim: Positron emission tomography (PET) imaging is a useful tool for assisting in correct differentiation of tumor progression from reactive changes, and the radiolabeled amino acid analog tracer O-(2-18F-fluoroethyl)-L-tyrosine (FET)-PET is amongst the most frequently used. The FET-PET images need to be quantitatively correct in order to be used clinically, which require accurate attenuation correction (AC) in PET/MRI. The aim of this study was to evaluate the use of the subject-specific MR-derived AC method RESOLUTE in post-operative brain tumor patients. Methods: We analyzed 51 post-operative brain tumor patients (68 examinations, 200 MBq [18F]-FET) investigated in a PET/MRI scanner. MR-AC maps were acquired using: (1) the Dixon water fat separation sequence, (2) the ultra short echo time (UTE) sequences, (3) calculated using our new RESOLUTE methodology, and (4) a same day low-dose CT used as reference "gold standard." For each subject and each AC method the tumor was delineated by isocontouring tracer uptake above a tumor(T)-to-brain background (B) activity ratio of 1.6. We measured B, tumor mean and maximal activity (TMEAN, TMAX), biological tumor volume (BTV), and calculated the clinical metrics TMEAN/B and TMAX/B. Results: When using RESOLUTE 5/68 studies did not meet our predefined acceptance criteria of TMAX/B difference to CT-AC < ±0.1 or 5%, TMEAN/B < ±0.05 or 5%, and BTV < ±2 mL or 10%. In total, 46/68 studies failed our acceptance criteria using Dixon, and 26/68 using UTE. The 95% limits of agreement for TMAX/B was for RESOLUTE (-3%; 4%), Dixon (-9%; 16%), and UTE (-7%; 10%). The absolute error when measuring BTV was 0.7 ± 1.9 mL (N.S) with RESOLUTE, 5.3 ± 10 mL using Dixon, and 1.7 ± 3.7 mL using UTE. RESOLUTE performed best in the identification of the location of peak activity and in brain tumor follow-up monitoring using clinical FET PET metrics. Conclusions: Overall, we found RESOLUTE to be the AC method that most robustly reproduced the CT-AC clinical metrics per se, during follow-up, and when interpreted into defined clinical use cut-off criteria and into the patient history. RESOLUTE is especially suitable for brain tumor patients, as these often present with distorted anatomy where other methods based on atlas/template information might fail.

PET/MRI for Oncologic Brain Imaging: A Comparison of Standard MR-Based Attenuation Corrections with a Model-Based Approach for the Siemens mMR PET/MR System.

The aim of this study was to compare attenuation-correction (AC) approaches for PET/MRI in clinical neurooncology. Methods: Forty-nine PET/MRI brain scans were included: brain tumor studies using 18F-fluoro-ethyl-tyrosine (18F-FET) (n = 31) and 68Ga-DOTANOC (n = 7) and studies of healthy subjects using 18F-FDG (n = 11). For each subject, MR-based AC maps (MR-AC) were acquired using the standard DIXON- and ultrashort echo time (UTE)-based approaches. A third MR-AC was calculated using a model-based, postprocessing approach to account for bone attenuation values (BD, noncommercial prototype software by Siemens Healthcare). As a reference, AC maps were derived from patient-specific CT images (CTref). PET data were reconstructed using standard settings after AC with all 4 AC methods. We report changes in diagnosis for all brain tumor patients and the following relative differences values (RDs [%]), with regards to AC-CTref: for 18F-FET (A)-SUVs as well as volumes of interest (VOIs) defined by a 70% threshold of all segmented lesions and lesion-to-background ratios; for 68Ga-DOTANOC (B)-SUVs as well as VOIs defined by a 50% threshold for all lesions and the pituitary gland; and for 18F-FDG (C)-RD of SUVs of the whole brain and 10 anatomic regions segmented on MR images. Results: For brain tumor imaging (A and B), the standard PET-based diagnosis was not affected by any of the 3 MR-AC methods. For A, the average RDs of SUVmean were -10%, -4%, and -3% and of the VOIs 1%, 2%, and 7% for DIXON, UTE, and BD, respectively. Lesion-to-background ratios for all MR-AC methods were similar to that of CTref. For B, average RDs of SUVmean were -11%, -11%, and -3% and of the VOIs 1%, -4%, and -3%, respectively. In the case of 18F-FDG PET/MRI (C), RDs for the whole brain were -11%, -8%, and -5% for DIXON, UTE, and BD, respectively. Conclusion: The diagnostic reading of PET/MR patients with brain tumors did not change with the chosen AC method. Quantitative accuracy of SUVs was clinically acceptable for UTE- and BD-AC for group A, whereas for group B BD was in accordance with CTref. Nevertheless, for the quantification of individual lesions large deviations to CTref can be observed independent of the MR-AC method used.

A multi-centre evaluation of eleven clinically feasible brain PET/MRI attenuation correction techniques using a large cohort of patients.

AIM: To accurately quantify the radioactivity concentration measured by PET, emission data need to be corrected for photon attenuation; however, the MRI signal cannot easily be converted into attenuation values, making attenuation correction (AC) in PET/MRI challenging. In order to further improve the current vendor-implemented MR-AC methods for absolute quantification, a number of prototype methods have been proposed in the literature. These can be categorized into three types: template/atlas-based, segmentation-based, and reconstruction-based. These proposed methods in general demonstrated improvements compared to vendor-implemented AC, and many studies report deviations in PET uptake after AC of only a few percent from a gold standard CT-AC. Using a unified quantitative evaluation with identical metrics, subject cohort, and common CT-based reference, the aims of this study were to evaluate a selection of novel methods proposed in the literature, and identify the ones suitable for clinical use. METHODS: In total, 11 AC methods were evaluated: two vendor-implemented (MR-ACDIXON and MR-ACUTE), five based on template/atlas information (MR-ACSEGBONE (Koesters et al., 2016), MR-ACONTARIO (Anazodo et al., 2014), MR-ACBOSTON (Izquierdo-Garcia et al., 2014), MR-ACUCL (Burgos et al., 2014), and MR-ACMAXPROB (Merida et al., 2015)), one based on simultaneous reconstruction of attenuation and emission (MR-ACMLAA (Benoit et al., 2015)), and three based on image-segmentation (MR-ACMUNICH (Cabello et al., 2015), MR-ACCAR-RiDR (Juttukonda et al., 2015), and MR-ACRESOLUTE (Ladefoged et al., 2015)). We selected 359 subjects who were scanned using one of the following radiotracers: [18F]FDG (210), [11C]PiB (51), and [18F]florbetapir (98). The comparison to AC with a gold standard CT was performed both globally and regionally, with a special focus on robustness and outlier analysis. RESULTS: The average performance in PET tracer uptake was within ±5% of CT for all of the proposed methods, with the average±SD global percentage bias in PET FDG uptake for each method being: MR-ACDIXON (-11.3±3.5)%, MR-ACUTE (-5.7±2.0)%, MR-ACONTARIO (-4.3±3.6)%, MR-ACMUNICH (3.7±2.1)%, MR-ACMLAA (-1.9±2.6)%, MR-ACSEGBONE (-1.7±3.6)%, MR-ACUCL (0.8±1.2)%, MR-ACCAR-RiDR (-0.4±1.9)%, MR-ACMAXPROB (-0.4±1.6)%, MR-ACBOSTON (-0.3±1.8)%, and MR-ACRESOLUTE (0.3±1.7)%, ordered by average bias. The overall best performing methods (MR-ACBOSTON, MR-ACMAXPROB, MR-ACRESOLUTE and MR-ACUCL, ordered alphabetically) showed regional average errors within ±3% of PET with CT-AC in all regions of the brain with FDG, and the same four methods, as well as MR-ACCAR-RiDR, showed that for 95% of the patients, 95% of brain voxels had an uptake that deviated by less than 15% from the reference. Comparable performance was obtained with PiB and florbetapir. CONCLUSIONS: All of the proposed novel methods have an average global performance within likely acceptable limits (±5% of CT-based reference), and the main difference among the methods was found in the robustness, outlier analysis, and clinical feasibility. Overall, the best performing methods were MR-ACBOSTON, MR-ACMAXPROB, MR-ACRESOLUTE and MR-ACUCL, ordered alphabetically. These methods all minimized the number of outliers, standard deviation, and average global and local error. The methods MR-ACMUNICH and MR-ACCAR-RiDR were both within acceptable quantitative limits, so these methods should be considered if processing time is a factor. The method MR-ACSEGBONE also demonstrates promising results, and performs well within the likely acceptable quantitative limits. For clinical routine scans where processing time can be a key factor, this vendor-provided solution currently outperforms most methods. With the performance of the methods presented here, it may be concluded that the challenge of improving the accuracy of MR-AC in adult brains with normal anatomy has been solved to a quantitatively acceptable degree, which is smaller than the quantification reproducibility in PET imaging.

Optimized MLAA for quantitative non-TOF PET/MR of the brain.

For quantitative tracer distribution in positron emission tomography, attenuation correction is essential. In a hybrid PET/CT system the CT images serve as a basis for generation of the attenuation map, but in PET/MR, the MR images do not have a similarly simple relationship with the attenuation map. Hence attenuation correction in PET/MR systems is more challenging. Typically either of two MR sequences are used: the Dixon or the ultra-short time echo (UTE) techniques. However these sequences have some well-known limitations. In this study, a reconstruction technique based on a modified and optimized non-TOF MLAA is proposed for PET/MR brain imaging. The idea is to tune the parameters of the MLTR applying some information from an attenuation image computed from the UTE sequences and a T1w MR image. In this MLTR algorithm, an [Formula: see text] parameter is introduced and optimized in order to drive the algorithm to a final attenuation map most consistent with the emission data. Because the non-TOF MLAA is used, a technique to reduce the cross-talk effect is proposed. In this study, the proposed algorithm is compared to the common reconstruction methods such as OSEM using a CT attenuation map, considered as the reference, and OSEM using the Dixon and UTE attenuation maps. To show the robustness and the reproducibility of the proposed algorithm, a set of 204 [18F]FDG patients, 35 [11C]PiB patients and 1 [18F]FET patient are used. The results show that by choosing an optimized value of [Formula: see text] in MLTR, the proposed algorithm improves the results compared to the standard MR-based attenuation correction methods (i.e. OSEM using the Dixon or the UTE attenuation maps), and the cross-talk and the scale problem are limited.

Dental artifacts in the head and neck region: implications for Dixon-based attenuation correction in PET/MR.

BACKGROUND: In the absence of CT or traditional transmission sources in combined clinical positron emission tomography/magnetic resonance (PET/MR) systems, MR images are used for MR-based attenuation correction (MR-AC). The susceptibility effects due to metal implants challenge MR-AC in the neck region of patients with dental implants. The purpose of this study was to assess the frequency and magnitude of subsequent PET image distortions following MR-AC. METHODS: A total of 148 PET/MR patients with clear visual signal voids on the attenuation map in the dental region were included in this study. Patients were injected with [(18)F]-FDG, [(11)C]-PiB, [(18)F]-FET, or [(64)Cu]-DOTATATE. The PET/MR data were acquired over a single-bed position of 25.8 cm covering the head and neck. MR-AC was based on either standard MR-ACDIXON or MR-ACINPAINTED where the susceptibility-induced signal voids were substituted with soft tissue information. Our inpainting algorithm delineates the outer contour of signal voids breaching the anatomical volume using the non-attenuation-corrected PET image and classifies the inner air regions based on an aligned template of likely dental artifact areas. The reconstructed PET images were evaluated visually and quantitatively using regions of interests in reference regions. The volume of the artifacts and the computed relative differences in mean and max standardized uptake value (SUV) between the two PET images are reported. RESULTS: The MR-based volume of the susceptibility-induced signal voids on the MR-AC attenuation maps was between 1.6 and 520.8 mL. The corresponding/resulting bias of the reconstructed tracer distribution was localized mainly in the area of the signal void. The mean and maximum SUVs averaged across all patients increased after inpainting by 52% (± 11%) and 28% (± 11%), respectively, in the corrected region. SUV underestimation decreased with the distance to the signal void and correlated with the volume of the susceptibility artifact on the MR-AC attenuation map. CONCLUSIONS: Metallic dental work may cause severe MR signal voids. The resulting PET/MR artifacts may exceed the actual volume of the dental fillings. The subsequent bias in PET is severe in regions in and near the signal voids and may affect the conspicuity of lesions in the mandibular region.

Region specific optimization of continuous linear attenuation coefficients based on UTE (RESOLUTE): application to PET/MR brain imaging.

The reconstruction of PET brain data in a PET/MR hybrid scanner is challenging in the absence of transmission sources, where MR images are used for MR-based attenuation correction (MR-AC). The main challenge of MR-AC is to separate bone and air, as neither have a signal in traditional MR images, and to assign the correct linear attenuation coefficient to bone. The ultra-short echo time (UTE) MR sequence was proposed as a basis for MR-AC as this sequence shows a small signal in bone. The purpose of this study was to develop a new clinically feasible MR-AC method with patient specific continuous-valued linear attenuation coefficients in bone that provides accurate reconstructed PET image data. A total of 164 [(18)F]FDG PET/MR patients were included in this study, of which 10 were used for training. MR-AC was based on either standard CT (reference), UTE or our method (RESOLUTE). The reconstructed PET images were evaluated in the whole brain, as well as regionally in the brain using a ROI-based analysis. Our method segments air, brain, cerebral spinal fluid, and soft tissue voxels on the unprocessed UTE TE images, and uses a mapping of R(*)2 values to CT Hounsfield Units (HU) to measure the density in bone voxels. The average error of our method in the brain was 0.1% and less than 1.2% in any region of the brain. On average 95% of the brain was within ±10% of PETCT, compared to 72% when using UTE. The proposed method is clinically feasible, reducing both the global and local errors on the reconstructed PET images, as well as limiting the number and extent of the outliers.

Automatic correction of dental artifacts in PET/MRI.

A challenge when using current magnetic resonance (MR)-based attenuation correction in positron emission tomography/MR imaging (PET/MRI) is that the MRIs can have a signal void around the dental fillings that is segmented as artificial air-regions in the attenuation map. For artifacts connected to the background, we propose an extension to an existing active contour algorithm to delineate the outer contour using the nonattenuation corrected PET image and the original attenuation map. We propose a combination of two different methods for differentiating the artifacts within the body from the anatomical air-regions by first using a template of artifact regions, and second, representing the artifact regions with a combination of active shape models and k-nearest-neighbors. The accuracy of the combined method has been evaluated using 25 [Formula: see text]-fluorodeoxyglucose PET/MR patients. Results showed that the approach was able to correct an average of [Formula: see text] of the artifact areas.

Positron emission tomography/magnetic resonance hybrid scanner imaging of cerebral blood flow using (15)O-water positron emission tomography and arterial spin labeling magnetic resonance imaging in newborn piglets.

Abnormality in cerebral blood flow (CBF) distribution can lead to hypoxic-ischemic cerebral damage in newborn infants. The aim of the study was to investigate minimally invasive approaches to measure CBF by comparing simultaneous (15)O-water positron emission tomography (PET) and single TI pulsed arterial spin labeling (ASL) magnetic resonance imaging (MR) on a hybrid PET/MR in seven newborn piglets. Positron emission tomography was performed with IV injections of 20 MBq and 100 MBq (15)O-water to confirm CBF reliability at low activity. Cerebral blood flow was quantified using a one-tissue-compartment-model using two input functions: an arterial input function (AIF) or an image-derived input function (IDIF). The mean global CBF (95% CI) PET-AIF, PET-IDIF, and ASL at baseline were 27 (23; 32), 34 (31; 37), and 27 (22; 32) mL/100 g per minute, respectively. At acetazolamide stimulus, PET-AIF, PET-IDIF, and ASL were 64 (55; 74), 76 (70; 83) and 79 (67; 92) mL/100 g per minute, respectively. At baseline, differences between PET-AIF, PET-IDIF, and ASL were 22% (P<0.0001) and -0.7% (P=0.9). At acetazolamide, differences between PET-AIF, PET-IDIF, and ASL were 19% (P=0.001) and 24% (P=0.0003). In conclusion, PET-IDIF overestimated CBF. Injected activity of 20 MBq (15)O-water had acceptable concordance with 100 MBq, without compromising image quality. Single TI ASL was questionable for regional CBF measurements. Global ASL CBF and PET CBF were congruent during baseline but not during hyperperfusion.