Ultra-low foetal radiation exposure in 18F-FDG PET/CT imaging with a long axial field-of-view PET/CT system.

PURPOSE: Long axial field-of-view (LAFOV) PET/CT systems enable PET/CT scans with reduced injected activities because of improved sensitivity. With this study, we aimed to examine the foetal radiation dose from an 18F-FDG PET/CT scan on a LAFOV PET/CT system with reduced injected activity. METHODS: Two pregnant women were retrospectively included and received an 18F-FDG PET/CT scan on a LAFOV PET/CT system with an intravenous bolus injection of 0.30 MBq/kg. Foetal radiation exposure from the PET was estimated using dose conversion factors from three published papers. Radiation exposure from the CT scans was estimated using CT-Expo. RESULTS: Foetal radiation dose from the PET scan ranged between 0.11 and 0.44 mGy. Foetal radiation exposure from the CT scan ranged between < 0.10 - 0.90 mGy depending if the foetus was included in the field-of-view. CONCLUSION: Foetal radiation dose could be reduced to < 1.5 mGy when scanning pregnant patients on a LAFOV PET/CT system. The radiation dose to the foetus was reduced significantly in our study due to the increased sensitivity of the LAFOV PET/CT system.

The development process of 'fit-for-purpose' imaging biomarkers to characterize the tumor microenvironment.

Immune-based treatment approaches are successfully used for the treatment of patients with cancer. While such therapies can be highly effective, many patients fail to benefit. To provide optimal therapy choices and to predict treatment responses, reliable biomarkers for the assessment of immune features in patients with cancer are of significant importance. Biomarkers (BM) that enable a comprehensive and repeatable assessment of the tumor microenvironment (TME), the lymphoid system, and the dynamics induced by drug treatment can fill this gap. Medical imaging, notably positron emission tomography (PET) and magnetic resonance imaging (MRI), providing whole-body imaging BMs, might deliver such BMs. However, those imaging BMs must be well characterized as being 'fit for purpose' for the intended use. This review provides an overview of the key steps involved in the development of 'fit-for-purpose' imaging BMs applicable in drug development, with a specific focus on pharmacodynamic biomarkers for assessing the TME and its modulation by immunotherapy. The importance of the qualification of imaging BMs according to their context of use (COU) as defined by the Food and Drug Administration (FDA) and National Institutes of Health Biomarkers, EndpointS, and other Tools (BEST) glossary is highlighted. We elaborate on how an imaging BM qualification for a specific COU can be achieved.

Baseline and on Treatment Biodistribution Variability of 18 F-FLT Uptake in Patients With Advanced Melanoma: Brief Communication.

PURPOSE: This prospective study evaluates the biodistribution of 18 F-FLT PET in patients with advanced melanoma before and after treatment with BRAF/MEK inhibitors. PATIENTS AND METHODS: Eighteen BRAF-positive unresectable stage IIIc or IV melanoma patients referred for 18 F-FLT PET/CT before (BL) and during (D14) BRAF/MEK inhibition were included. 18 F-FLT accumulation in the liver, bone marrow, blood, and muscle was quantified. RESULTS: Baseline interpatient 18 F-FLT uptake had a coefficient-of-variation between 17.5% and 21.5%. During treatment, liver uptake increased (SUV meanBL = 4.86 ± 0.98, SUV meanD14 = 6.31 ± 1.36, P < 0.001) and bone marrow uptake decreased (SUV meanBL = 7.67 ± 1.65, SUV meanD14 = 6.78 ± 1.19, P < 0.025). Both changes were unrelated to baseline metabolic tumor volume or tumor response. CONCLUSIONS: To assess 18 F-FLT PET, both liver and bone marrow uptake may be used as normal tissue references at baseline, but 18 F-FLT biodistribution significantly changes in longitudinal response studies when treated with BRAF/MEK inhibitors.

Investigating reliable amyloid accumulation in Centiloids: Results from the AMYPAD Prognostic and Natural History Study.

INTRODUCTION: To support clinical trial designs focused on early interventions, our study determined reliable early amyloid-ß (Aß) accumulation based on Centiloids (CL) in pre-dementia populations. METHODS: A total of 1032 participants from the Amyloid Imaging to Prevent Alzheimer's Disease-Prognostic and Natural History Study (AMYPAD-PNHS) and Insight46 who underwent [18F]flutemetamol, [18F]florbetaben or [18F]florbetapir amyloid-PET were included. A normative strategy was used to define reliable accumulation by estimating the 95th percentile of longitudinal measurements in sub-populations (NPNHS = 101/750, NInsight46 = 35/382) expected to remain stable over time. The baseline CL threshold that optimally predicts future accumulation was investigated using precision-recall analyses. Accumulation rates were examined using linear mixed-effect models. RESULTS: Reliable accumulation in the PNHS was estimated to occur at >3.0 CL/year. Baseline CL of 16 [12,19] best predicted future Aß-accumulators. Rates of amyloid accumulation were tracer-independent, lower for APOE ε4 non-carriers, and for subjects with higher levels of education. DISCUSSION: Our results support a 12-20 CL window for inclusion into early secondary prevention studies. Reliable accumulation definition warrants further investigations.

Is PET Radiomics Useful to Predict Pathologic Tumor Response and Prognosis in Locally Advanced Cervical Cancer?

This study investigated whether radiomic features extracted from pretreatment [18F]FDG PET could improve the prediction of both histopathologic tumor response and survival in patients with locally advanced cervical cancer (LACC) treated with neoadjuvant chemoradiotherapy followed by surgery compared with conventional PET parameters and histopathologic features. Methods: The medical records of all consecutive patients with LACC referred between July 2010 and July 2016 were reviewed. [18F]FDG PET/CT was performed before neoadjuvant chemoradiotherapy. Radiomic features were extracted from the primary tumor volumes delineated semiautomatically on the PET images and reduced by factor analysis. A receiver-operating-characteristic analysis was performed, and conventional and radiomic features were dichotomized with Liu's method according to pathologic response (pR) and cancer-specific death. According to the study protocol, only areas under the curve of more than 0.70 were selected for further analysis, including logistic regression analysis for response prediction and Cox regression analysis for survival prediction. Results: A total of 195 patients fulfilled the inclusion criteria. At pathologic evaluation after surgery, 131 patients (67.2%) had no or microscopic (≤3 mm) residual tumor (pR0 or pR1, respectively); 64 patients (32.8%) had macroscopic residual tumor (>3 mm, pR2). With a median follow-up of 76.0 mo (95% CI, 70.7-78.7 mo), 31.3% of patients had recurrence or progression and 20.0% died of the disease. Among conventional PET parameters, SUVmean significantly differed between pathologic responders and nonresponders. Among radiomic features, 1 shape and 3 textural features significantly differed between pathologic responders and nonresponders. Three radiomic features significantly differed between presence and absence of recurrence or progression and between presence and absence of cancer-specific death. Areas under the curve were less than 0.70 for all parameters; thus, univariate and multivariate regression analyses were not performed. Conclusion: In a large series of patients with LACC treated with neoadjuvant chemoradiotherapy followed by surgery, PET radiomic features could not predict histopathologic tumor response and survival. It is crucial to further explore the biologic mechanism underlying imaging-derived parameters and plan a large, prospective, multicenter study with standardized protocols for all phases of the process of radiomic analysis to validate radiomics before its use in clinical routine.

Validation of image-derived input function using a long axial field of view PET/CT scanner for two different tracers.

BACKGROUND: Accurate image-derived input function (IDIF) from highly sensitive large axial field of view (LAFOV) PET/CT scanners could avoid the need of invasive blood sampling for kinetic modelling. The aim is to validate the use of IDIF for two kinds of tracers, 3 different IDIF locations and 9 different reconstruction settings. METHODS: Eight [18F]FDG and 10 [18F]DPA-714 scans were acquired respectively during 70 and 60 min on the Vision Quadra PET/CT system. PET images were reconstructed using various reconstruction settings. IDIFs were taken from ascending aorta (AA), descending aorta (DA), and left ventricular cavity (LV). The calibration factor (CF) extracted from the comparison between the IDIFs and the manual blood samples as reference was used for IDIFs accuracy and precision assessment. To illustrate the effect of various calibrated-IDIFs on Patlak linearization for [18F]FDG and Logan linearization for [18F]DPA-714, the same target time-activity curves were applied for each calibrated-IDIF. RESULTS: For [18F]FDG, the accuracy and precision of the IDIFs were high (mean CF ≥ 0.82, SD ≤ 0.06). Compared to the striatum influx (Ki) extracted using calibrated AA IDIF with the updated European Association of Nuclear Medicine Research Ltd. standard reconstruction (EARL2), Ki mean differences were < 2% using the other calibrated IDIFs. For [18F]DPA714, high accuracy of the IDIFs was observed (mean CF ≥ 0.86) except using absolute scatter correction, DA and LV (respectively mean CF = 0.68, 0.47 and 0.44). However, the precision of the AA IDIFs was low (SD ≥ 0.10). Compared to the distribution volume (VT) in a frontal region obtained using calibrated continuous arterial sampler input function as reference, VT mean differences were small using calibrated AA IDIFs (for example VT mean difference = -5.3% using EARL2), but higher using calibrated DA and LV IDIFs (respectively + 12.5% and + 19.1%). CONCLUSIONS: For [18F]FDG, IDIF do not need calibration against manual blood samples. For [18F]DPA-714, AA IDIF can replace continuous arterial sampling for simplified kinetic quantification but only with calibration against arterial blood samples. The accuracy and precision of IDIF from LAFOV PET/CT system depend on tracer, reconstruction settings and IDIF VOI locations, warranting careful optimization.

The Image Biomarker Standardization Initiative: Standardized Convolutional Filters for Reproducible Radiomics and Enhanced Clinical Insights.

Filters are commonly used to enhance specific structures and patterns in images, such as vessels or peritumoral regions, to enable clinical insights beyond the visible image using radiomics. However, their lack of standardization restricts reproducibility and clinical translation of radiomics decision support tools. In this special report, teams of researchers who developed radiomics software participated in a three-phase study (September 2020 to December 2022) to establish a standardized set of filters. The first two phases focused on finding reference filtered images and reference feature values for commonly used convolutional filters: mean, Laplacian of Gaussian, Laws and Gabor kernels, separable and nonseparable wavelets (including decomposed forms), and Riesz transformations. In the first phase, 15 teams used digital phantoms to establish 33 reference filtered images of 36 filter configurations. In phase 2, 11 teams used a chest CT image to derive reference values for 323 of 396 features computed from filtered images using 22 filter and image processing configurations. Reference filtered images and feature values for Riesz transformations were not established. Reproducibility of standardized convolutional filters was validated on a public data set of multimodal imaging (CT, fluorodeoxyglucose PET, and T1-weighted MRI) in 51 patients with soft-tissue sarcoma. At validation, reproducibility of 486 features computed from filtered images using nine configurations × three imaging modalities was assessed using the lower bounds of 95% CIs of intraclass correlation coefficients. Out of 486 features, 458 were found to be reproducible across nine teams with lower bounds of 95% CIs of intraclass correlation coefficients greater than 0.75. In conclusion, eight filter types were standardized with reference filtered images and reference feature values for verifying and calibrating radiomics software packages. A web-based tool is available for compliance checking.

EANM practice guidelines for an appropriate use of PET and SPECT for patients with epilepsy.

Epilepsy is one of the most frequent neurological conditions with an estimated prevalence of more than 50 million people worldwide and an annual incidence of two million. Although pharmacotherapy with anti-seizure medication (ASM) is the treatment of choice, ~30% of patients with epilepsy do not respond to ASM and become drug resistant. Focal epilepsy is the most frequent form of epilepsy. In patients with drug-resistant focal epilepsy, epilepsy surgery is a treatment option depending on the localisation of the seizure focus for seizure relief or seizure freedom with consecutive improvement in quality of life. Beside examinations such as scalp video/electroencephalography (EEG) telemetry, structural, and functional magnetic resonance imaging (MRI), which are primary standard tools for the diagnostic work-up and therapy management of epilepsy patients, molecular neuroimaging using different radiopharmaceuticals with single-photon emission computed tomography (SPECT) and positron emission tomography (PET) influences and impacts on therapy decisions. To date, there are no literature-based praxis recommendations for the use of Nuclear Medicine (NM) imaging procedures in epilepsy. The aims of these guidelines are to assist in understanding the role and challenges of radiotracer imaging for epilepsy; to provide practical information for performing different molecular imaging procedures for epilepsy; and to provide an algorithm for selecting the most appropriate imaging procedures in specific clinical situations based on current literature. These guidelines are written and authorized by the European Association of Nuclear Medicine (EANM) to promote optimal epilepsy imaging, especially in the presurgical setting in children, adolescents, and adults with focal epilepsy. They will assist NM healthcare professionals and also specialists such as Neurologists, Neurophysiologists, Neurosurgeons, Psychiatrists, Psychologists, and others involved in epilepsy management in the detection and interpretation of epileptic seizure onset zone (SOZ) for further treatment decision. The information provided should be applied according to local laws and regulations as well as the availability of various radiopharmaceuticals and imaging modalities.

How to obtain the image-derived blood concentration from 89Zr-immuno-PET scans.

BACKGROUND: PET scans using zirconium-89 labelled monoclonal antibodies (89Zr-mAbs), known as 89Zr-immuno-PET, are made to measure uptake in tumour and organ tissue. Uptake is related to the supply of 89Zr-mAbs in the blood. Measuring activity concentrations in blood, however, requires invasive blood sampling. This study aims to identify the best delineation strategy to obtain the image-derived blood concentration (IDBC) from 89Zr-immuno-PET scans. METHODS: PET imaging and blood sampling of two 89Zr-mAbs were included, 89Zr-cetuximab and 89Zr-durvalumab. For seven patients receiving 89Zr-cetuximab, PET scans on 1-2 h, 2 and 6 days post-injection (p.i.) were analysed. Five patients received three injections of 89Zr-durvalumab. The scanning protocol for the first two injections consisted of PET scanning on 2, 5 and 7 days p.i. and for the third injection only on 7 days p.i. Blood samples were drawn with every PET scan and the sample-derived blood concentration (SDBC) was used as gold standard for the IDBC. According to an in-house developed standard operating procedure, the aortic arch, ascending aorta, descending aorta and left ventricle were delineated. Bland-Altman analyses were performed to assess the bias (mean difference) and variability (1.96 times the standard deviation of the differences) between IDBC and SDBC. RESULTS: Overall, the activity concentration obtained from the IDBC was lower than from the SDBC. When comparing IDBC with SDBC, variability was smallest for the ascending aorta (20.3% and 17.0% for 89Zr-cetuximab and 89Zr-durvalumab, respectively). Variability for the other regions ranged between 17.9 and 30.8%. Bias for the ascending aorta was - 10.9% and - 11.4% for 89Zr-cetuximab and 89Zr-durvalumab, respectively. CONCLUSIONS: Image-derived blood concentrations should be obtained from delineating the ascending aorta in 89Zr-immuno-PET scans, as this results in the lowest variability with respect to sample-derived blood concentrations.

Non-specific irreversible 89Zr-mAb uptake in tumours: evidence from biopsy-proven target-negative tumours using 89Zr-immuno-PET.

BACKGROUND: Distribution of mAbs into tumour tissue may occur via different processes contributing differently to the 89Zr-mAb uptake on PET. Target-specific binding in tumours is of main interest; however, non-specific irreversible uptake may also be present, which influences quantification. The aim was to investigate the presence of non-specific irreversible uptake in tumour tissue using Patlak linearization on 89Zr-immuno-PET data of biopsy-proven target-negative tumours. Data of two studies, including target status obtained from biopsies, were retrospectively analysed, and Patlak linearization provided the net rate of irreversible uptake (Ki). RESULTS: Two tumours were classified as CD20-negative and two as CD20-positive. Four tumours were classified as CEA-negative and nine as CEA-positive. Ki values of CD20-negative (0.43 µL/g/h and 0.92 µL/g/h) and CEA-negative tumours (mdn = 1.97 µL/g/h, interquartile range (IQR) = 1.50-2.39) were higher than zero. Median Ki values of target-negative tumours were lower than CD20-positive (1.87 µL/g/h and 1.90 µL/g/h) and CEA-positive tumours (mdn = 2.77 µL/g/h, IQR = 2.11-3.65). CONCLUSION: Biopsy-proven target-negative tumours showed irreversible uptake of 89Zr-mAbs measured in vivo using 89Zr-immuno-PET data, which suggests the presence of non-specific irreversible uptake in tumours. Consequently, for 89Zr-immuno-PET, even if the target is absent, a tumour-to-plasma ratio always increases over time.

Mitigating SUV uncertainties using total body PET imaging.

PURPOSE: Standardised uptake values (SUV) are commonly used to quantify 18F-FDG lesion uptake. However, SUVs may suffer from several uncertainties and errors. Long-axial field-of-view (LAFOV) PET/CT systems might enable image-based quality control (QC) by deriving 18F-FDG activity and weight from total body (TB) 18F-FDG PET images. In this study, we aimed to develop these image-based QC to reduce errors and mitigate SUV uncertainties. METHODS: Twenty-five out of 81 patient scans from a LAFOV PET/CT system were used to determine regression fits for deriving of image-derived activity and weight. Thereafter, the regression fits were applied to 56 independent 18F-FDG PET scans from the same scanner to determine if injected activity and weight could be obtained accurately from TB and half-body (HB) scans. Additionally, we studied the impact of image-based values on the precision of liver SUVmean and lesion SUVpeak. Finally, 20 scans were acquired from a short-axial field-of-view (SAFOV) PET/CT system to determine if the regression fits also applied to HB scans from a SAFOV system. RESULTS: Both TB and HB 18F-FDG activity and weight significantly predicted reported injected activity (r = 0.999; r = 0.984) and weight (r = 0.999; r = 0.987), respectively. After applying the regression fits, 18F-FDG activity and weight were accurately derived within 4.8% and 3.2% from TB scans and within 4.9% and 3.1% from HB, respectively. Image-derived values also mitigated liver and lesion SUV variability compared with reported values. Moreover, 18F-FDG activity and weight obtained from a SAFOV scanner were derived within 6.7% and 4.5%, respectively. CONCLUSION: 18F-FDG activity and weight can be derived accurately from TB and HB scans, and image-derived values improved SUV precision and corrected for lesion SUV errors. Therefore, image-derived values should be included as QC to generate a more reliable and reproducible quantitative uptake measurement.

The Predictive Value of FDG PET/CT for Determining Progression-Free Survival in Advanced Stage III-IV BRAF -Mutated Melanoma Patients Treated With Targeted Therapy-What Can Be Learned From Progression?

PURPOSE: The aims of this study were to investigate whether (early) PERCIST response monitoring with 18 F-FDG PET/CT is predictive for progression-free survival (PFS) in unresectable stage III or IV melanoma patients treated with BRAF/MEK inhibitor (MEKi) and to define dissemination patterns at progression with a lesion-based evaluation in direct comparison to baseline to improve our understanding of 18 F-FDG PET/CT during BRAF/MEKi. PATIENTS AND METHODS: This prospective multicenter single-arm study included 70 patients with unresectable stage III/IV BRAF -mutated melanoma who underwent contrast-enhanced CT and 18 F-FDG PET/CT at baseline and 2 and 7 weeks during treatment with vemurafenib plus cobimetinib and at progression if possible. Tumor response assessment was done with RECIST1.1 and PERCIST. Follow-up PET/CT scans were visually compared with baseline to assess dissemination patterns. RESULTS: Using RECIST1.1, PFS was not significantly different between the response groups ( P = 0.26). At 2 weeks, PERCIST median PFS was 15.7 months for patients with complete metabolic response (CMR) versus 8.3 months for non-CMR ( P = 0.035). The hazards ratio (HR) for progression/death in non-CMR versus CMR was 1.99 (95% confidence interval [CI], 1.03-3.84; P = 0.040) and 1.77 (95% CI, 0.91-3.43; P = 0.0935) when adjusting for lactate dehydrogenase (LDH). At 7 weeks, median PFS for PERCIST CMR was 16.7 months versus 8.5 months for non-CMR ( P = 0.0003). The HR for progression/death in the non-CMR group was significantly increased (HR, 2.94; 95% CI, 1.60-5.40; P = 0.0005), even when adjusting for LDH (HR, 2.65; 95% CI, 1.43-4.91; P = 0.0020). At week 7, 18 F-FDG PET/CT was false-positive in all 4 (6%) patients with new FDG-avid lesions but CMR of known metastases. When 18 F-FDG PET/CT was performed at progressive disease, 18/22 (82%) patients had progression of known metastases with or without new 18 F-FDG-avid lesions. CONCLUSIONS: This study shows that PERCIST response assessment at week 7 is predictive for PFS, regardless of LDH. At 2 weeks, patients with CMR have longer PFS than patients with non-CMR, but different PET parameters should be investigated to further evaluate the added value of early 18 F-FDG PET/CT. Disease progression on PET/CT is predominated by progression of known metastases, and new 18 F-FDG-avid lesions during BRAF/MEKi are not automatically a sign of recurrent disease.

89Zr-Immuno-PET with Immune Checkpoint Inhibitors: Measuring Target Engagement in Healthy Organs.

INTRODUCTION: 89Zr-immuno-PET (positron emission tomography with zirconium-89-labeled monoclonal antibodies ([89Zr]Zr-mAbs)) can be used to study the biodistribution of mAbs targeting the immune system. The measured uptake consists of target-specific and non-specific components, and it can be influenced by plasma availability of the tracer. To find evidence for target-specific uptake, i.e., target engagement, we studied five immune-checkpoint-targeting [89Zr]Zr-mAbs to (1) compare the uptake with previously reported baseline values for non-specific organ uptake (ns-baseline) and (2) look for saturation effects of increasing mass doses. METHOD: 89Zr-immuno-PET data from five [89Zr]Zr-mAbs, i.e., nivolumab and pembrolizumab (anti-PD-1), durvalumab (anti-PD-L1), BI 754,111 (anti-LAG-3), and ipilimumab (anti-CTLA-4), were analysed. For each mAb, 2-3 different mass doses were evaluated. PET scans and blood samples from at least two time points 24 h post injection were available. In 35 patients, brain, kidneys, liver, spleen, lungs, and bone marrow were delineated. Patlak analysis was used to account for differences in plasma activity concentration and to quantify irreversible uptake (Ki). To identify target engagement, Ki values were compared to ns-baseline Ki values previously reported, and the effect of increasing mass doses on Ki was investigated. RESULTS: All mAbs, except ipilimumab, showed Ki values in spleen above the ns-baseline for the lowest administered mass dose, in addition to decreasing Ki values with higher mass doses, both indicative of target engagement. For bone marrow, no ns-baseline was established previously, but a similar pattern was observed. For kidneys, most mAbs showed Ki values within the ns-baseline for both low and high mass doses. However, with high mass doses, some saturation effects were seen, suggestive of a lower ns-baseline value. Ki values were near zero in brain tissue for all mass doses of all mAbs. CONCLUSION: Using Patlak analysis and the established ns-baseline values, evidence for target engagement in (lymphoid) organs for several immune checkpoint inhibitors could be demonstrated. A decrease in the Ki values with increasing mass doses supports the applicability of Patlak analysis for the assessment of target engagement for PET ligands with irreversible uptake behavior.

Optimization and validation of 18F-DCFPyL PET radiomics-based machine learning models in intermediate- to high-risk primary prostate cancer.

INTRODUCTION: Radiomics extracted from prostate-specific membrane antigen (PSMA)-PET modeled with machine learning (ML) may be used for prediction of disease risk. However, validation of previously proposed approaches is lacking. We aimed to optimize and validate ML models based on 18F-DCFPyL-PET radiomics for the prediction of lymph-node involvement (LNI), extracapsular extension (ECE), and postoperative Gleason score (GS) in primary prostate cancer (PCa) patients. METHODS: Patients with intermediate- to high-risk PCa who underwent 18F-DCFPyL-PET/CT before radical prostatectomy with pelvic lymph-node dissection were evaluated. The training dataset included 72 patients, the internal validation dataset 24 patients, and the external validation dataset 27 patients. PSMA-avid intra-prostatic lesions were delineated semi-automatically on PET and 480 radiomics features were extracted. Conventional PET-metrics were derived for comparative analysis. Segmentation, preprocessing, and ML methods were optimized in repeated 5-fold cross-validation (CV) on the training dataset. The trained models were tested on the combined validation dataset. Combat harmonization was applied to external radiomics data. Model performance was assessed using the receiver-operating-characteristics curve (AUC). RESULTS: The CV-AUCs in the training dataset were 0.88, 0.79 and 0.84 for LNI, ECE, and GS, respectively. In the combined validation dataset, the ML models could significantly predict GS with an AUC of 0.78 (p<0.05). However, validation AUCs for LNI and ECE prediction were not significant (0.57 and 0.63, respectively). Conventional PET metrics-based models had comparable AUCs for LNI (0.59, p>0.05) and ECE (0.66, p>0.05), but a lower AUC for GS (0.73, p<0.05). In general, Combat harmonization improved external validation AUCs (-0.03 to +0.18). CONCLUSION: In internal and external validation, 18F-DCFPyL-PET radiomics-based ML models predicted high postoperative GS but not LNI or ECE in intermediate- to high-risk PCa. Therefore, the clinical benefit seems to be limited. These results underline the need for external and/or multicenter validation of PET radiomics-based ML model analyses to assess their generalizability.

Head-to-head comparison of relative cerebral blood flow derived from dynamic [18F]florbetapir and [18F]flortaucipir PET in subjects with subjective cognitive decline.

BACKGROUND: Dynamic PET imaging studies provide accurate estimates of specific binding, but also measure the relative tracer delivery (R1), which is a proxy for relative cerebral blood flow (rCBF). Recently, studies suggested that R1 obtained from different tracers could be used interchangeably and is irrespective of target tissue. However, the similarities or differences of R1 obtained from different PET tracers still require validation. Therefore, the goal of the current study was to compare R1 estimates, derived from dynamic [18F]florbetapir (amyloid) and [18F]flortaucipir (tau) PET, in the same subjects with subjective cognitive decline (SCD). RESULTS: Voxel-wise analysis presented a small cluster (1.6% of the whole brain) with higher R1 values for [18F]flortaucipir compared to [18F]florbetapir in the Aß-negative group. These voxels were part of the hippocampus and the left middle occipital gyrus. In part of the thalamus, midbrain and cerebellum, voxels (2.5% of the whole brain) with higher R1 values for [18F]florbetapir were observed. In the Aß-positive group, a cluster (0.2% of the whole brain) of higher R1 values was observed in part of the hippocampus, right parahippocampal gyrus and in the left sagittal stratum for [18F]flortaucipir compared to [18F]florbetapir. Furthermore, in part of the thalamus, left amygdala, midbrain and right parahippocampal gyrus voxels (0.4% of the whole brain) with higher R1 values for [18F]florbetapir were observed. Despite these differences, [18F]florbetapir R1 had high correspondence with [18F]flortaucipir R1 across all regions of interest (ROIs) and subjects (Aß-:r2 = 0.79, slope = 0.85, ICC = 0.76; Aß+: r2 = 0.87, slope = 0.93, ICC = 0.77). CONCLUSION: [18F]flortaucipir and [18F]florbetapir showed similar R1 estimates in cortical regions. This finding, put together with previous studies, indicates that R1 could be considered a surrogate for relative cerebral blood flow (rCBF) in the cortex and may be used interchangeably, but with caution, regardless of the choice of these two tracers.

Harmonization of brain PET images in multi-center PET studies using Hoffman phantom scan.

BACKGROUND: Image harmonization has been proposed to minimize heterogeneity in brain PET scans acquired in multi-center studies. However, standard validated methods and software tools are lacking. Here, we assessed the performance of a framework for the harmonization of brain PET scans in a multi-center European clinical trial. METHOD: Hoffman 3D brain phantoms were acquired in 28 PET systems and reconstructed using site-specific settings. Full Width at Half Maximum (FWHM) of the Effective Image Resolution (EIR) and harmonization kernels were estimated for each scan. The target EIR was selected as the coarsest EIR in the imaging network. Using "Hoffman 3D brain Analysis tool," indicators of image quality were calculated before and after the harmonization: The Coefficient of Variance (COV%), Gray Matter Recovery Coefficient (GMRC), Contrast, Cold-Spot RC, and left-to-right GMRC ratio. A COV% ≤ 15% and Contrast ≥ 2.2 were set as acceptance criteria. The procedure was repeated to achieve a 6-mm target EIR in a subset of scans. The method's robustness against typical dose-calibrator-based errors was assessed. RESULTS: The EIR across systems ranged from 3.3 to 8.1 mm, and an EIR of 8 mm was selected as the target resolution. After harmonization, all scans met acceptable image quality criteria, while only 13 (39.4%) did before. The harmonization procedure resulted in lower inter-system variability indicators: Mean ± SD COV% (from 16.97 ± 6.03 to 7.86 ± 1.47%), GMRC Inter-Quartile Range (0.040-0.012), and Contrast SD (0.14-0.05). Similar results were obtained with a 6-mm FWHM target EIR. Errors of ± 10% in the DRO activity resulted in differences below 1 mm in the estimated EIR. CONCLUSION: Harmonizing the EIR of brain PET scans significantly reduced image quality variability while minimally affecting quantitative accuracy. This method can be used prospectively for harmonizing scans to target sharper resolutions and is robust against dose-calibrator errors. Comparable image quality is attainable in brain PET multi-center studies while maintaining quantitative accuracy.

Ethical Considerations for Artificial Intelligence in Medical Imaging: Data Collection, Development, and Evaluation.

The development of artificial intelligence (AI) within nuclear imaging involves several ethically fraught components at different stages of the machine learning pipeline, including during data collection, model training and validation, and clinical use. Drawing on the traditional principles of medical and research ethics, and highlighting the need to ensure health justice, the AI task force of the Society of Nuclear Medicine and Molecular Imaging has identified 4 major ethical risks: privacy of data subjects, data quality and model efficacy, fairness toward marginalized populations, and transparency of clinical performance. We provide preliminary recommendations to developers of AI-driven medical devices for mitigating the impact of these risks on patients and populations.

Sensitivity of an AI method for [18F]FDG PET/CT outcome prediction of diffuse large B-cell lymphoma patients to image reconstruction protocols.

BACKGROUND: Convolutional neural networks (CNNs), applied to baseline [18F]-FDG PET/CT maximum intensity projections (MIPs), show potential for treatment outcome prediction in diffuse large B-cell lymphoma (DLBCL). The aim of this study is to investigate the robustness of CNN predictions to different image reconstruction protocols. Baseline [18F]FDG PET/CT scans were collected from 20 DLBCL patients. EARL1, EARL2 and high-resolution (HR) protocols were applied per scan, generating three images with different image qualities. Image-based transformation was applied by blurring EARL2 and HR images to generate EARL1 compliant images using a Gaussian filter of 5 and 7 mm, respectively. MIPs were generated for each of the reconstructions, before and after image transformation. An in-house developed CNN predicted the probability of tumor progression within 2 years for each MIP. The difference in probabilities per patient was then calculated between both EARL2 and HR with respect to EARL1 (delta probabilities or ΔP). We compared these to the probabilities obtained after aligning the data with ComBat using the difference in median and interquartile range (IQR). RESULTS: CNN probabilities were found to be sensitive to different reconstruction protocols (EARL2 ΔP: median = 0.09, interquartile range (IQR) = [0.06, 0.10] and HR ΔP: median = 0.1, IQR = [0.08, 0.16]). Moreover, higher resolution images (EARL2 and HR) led to higher probability values. After image-based and ComBat transformation, an improved agreement of CNN probabilities among reconstructions was found for all patients. This agreement was slightly better after image-based transformation (transformed EARL2 ΔP: median = 0.022, IQR = [0.01, 0.02] and transformed HR ΔP: median = 0.029, IQR = [0.01, 0.03]). CONCLUSION: Our CNN-based outcome predictions are affected by the applied reconstruction protocols, yet in a predictable manner. Image-based harmonization is a suitable approach to harmonize CNN predictions across image reconstruction protocols.

Prognostic model using 18F-FDG PET radiomics predicts progression-free survival in relapsed/refractory Hodgkin lymphoma.

Investigating prognostic factors in patients with relapsed or primary refractory classical Hodgkin lymphoma (R/R cHL) is essential to optimize risk-adapted treatment strategies. We built a prognostic model using baseline quantitative 18F-fluorodeoxyglucose positron emission tomography (PET) radiomics features and clinical characteristics to predict the progression-free survival (PFS) among patients with R/R cHL treated with salvage chemotherapy followed by autologous stem cell transplantation. Metabolic tumor volume and several novel radiomics dissemination features, representing interlesional differences in distance, volume, and standard uptake value, were extracted from the baseline PET. Machine learning using backward selection and logistic regression were applied to develop and train the model on a total of 113 patients from 2 clinical trials. The model was validated on an independent external cohort of 69 patients. In addition, we validated 4 different PET segmentation methods to calculate radiomics features. We identified a subset of patients at high risk for progression with significant inferior 3-year PFS outcomes of 38.1% vs 88.4% for patients in the low-risk group in the training cohort (P < .001) and 38.5% vs 75.0% in the validation cohort (P = .015), respectively. The overall survival was also significantly better in the low-risk group (P = .022 and P < .001). We provide a formula to calculate a risk score for individual patients based on the model. In conclusion, we developed a prognostic model for PFS combining radiomics and clinical features in a large cohort of patients with R/R cHL. This model calculates a PET-based risk profile and can be applied to develop risk-stratified treatment strategies for patients with R/R cHL. These trials were registered at www.clinicaltrials.gov as #NCT02280993, #NCT00255723, and #NCT01508312.