Fully automated dual-run manufacturing of [11C]PIB on FASTlab

This report describes an updated, fully automated method for the production of [11C]PIB on a cassette-based automated synthesis module. The method allows for two separate productions of [11C]PIB, both of which meet all specification for use in clinical studies. The GE FASTlab developer system was used to create the cassette design as well as the controlling tracer package. The method takes 16 min from the delivery of [11C]MeOTf to the FASTlab, or 35 min from the End of Bombardment; and reliably produces 3547 ± 586 MBq of [11C]PIB in high radiochemical purity (> 98 %). This methodology increases the production capacity of radiopharmaceutical facilities for [11C]PIB, and can easily produce 4 batches in a single day with limited infrastructure footprint.

89Zr-panitumumab Combined With 18F-FDG PET Improves Detection and Staging of Head and Neck Squamous Cell Carcinoma

PURPOSE

Determine the safety and specificity of a tumor-targeted radiotracer (89Zr-pan) in combination with 18F-FDG PET/CT to improve diagnostic accuracy in head and neck squamous cell carcinoma (HNSCC).

EXPERIMENTAL DESIGN

Adult patients with biopsy-proven HNSCC scheduled for standard-of-care surgery were enrolled in a clinical trial and underwent systemic administration of 89Zirconium-panitumumab and panitumumab-IRDye800 followed by preoperative 89Zr-pan PET/CT and intraoperative fluorescence imaging. The sensitivity, specificity, and AUC were evaluated.

RESULTS

A total of fourteen patients were enrolled and completed the study. Four patients (28.5%) had areas of high 18F-FDG uptake outside the head and neck region with maximum standardized uptake values (SUVmax) greater than 2.0 that were not detected on 89Zr-pan PET/CT. These four patients with incidental findings underwent further workup and had no evidence of cancer on biopsy or clinical follow-up. Forty-eight lesions (primary tumor, LNs, incidental findings) with SUVmax ranging 2.0-23.6 were visualized on 18F-FDG PET/CT; 34 lesions on 89Zr-pan PET/CT with SUVmax ranging 0.9-10.5. The combined ability of 18F-FDG PET/CT and 89Zr-pan PET/CT to detect HNSCC in the whole body was improved with higher specificity of 96.3% [confidence interval (CI), 89.2%-100%] compared to 18F-FDG PET/CT alone with specificity of 74.1% (CI, 74.1%-90.6%). One possibly related grade 1 adverse event of prolonged QTc (460 ms) was reported but resolved in follow-up.

CONCLUSIONS

89Zr-pan PET/CT imaging is safe and may be valuable in discriminating incidental findings identified on 18F-FDG PET/CT from true positive lesions and in localizing metastatic LNs.

Performance Comparison of Individual and Ensemble CNN Models for the Classification of Brain 18F-FDG-PET Scans

The high-background glucose metabolism of normal gray matter on [18F]-fluoro-2-D-deoxyglucose (FDG) positron emission tomography (PET) of the brain results in a low signal-to-background ratio, potentially increasing the possibility of missing important findings in patients with intracranial malignancies. To explore the strategy of using a deep learning classifier to aid in distinguishing normal versus abnormal findings on PET brain images, this study evaluated the performance of a two-dimensional convolutional neural network (2D-CNN) to classify FDG PET brain scans as normal (N) or abnormal (A).

METHODS

Two hundred eighty-nine brain FDG-PET scans (N; n = 150, A; n = 139) resulting in a total of 68,260 images were included. Nine individual 2D-CNN models with three different window settings for axial, coronal, and sagittal axes were trained and validated. The performance of these individual and ensemble models was evaluated and compared using a test dataset. Odds ratio, Akaike's information criterion (AIC), and area under curve (AUC) on receiver-operative-characteristic curve, accuracy, and standard deviation (SD) were calculated.

RESULTS

An optimal window setting to classify normal and abnormal scans was different for each axis of the individual models. An ensembled model using different axes with an optimized window setting (window-triad) showed better performance than ensembled models using the same axis and different windows settings (axis-triad). Increase in odds ratio and decrease in SD were observed in both axis-triad and window-triad models compared with individual models, whereas improvements of AUC and AIC were seen in window-triad models. An overall model averaging the probabilities of all individual models showed the best accuracy of 82.0%.

CONCLUSIONS

Data ensemble using different window settings and axes was effective to improve 2D-CNN performance parameters for the classification of brain FDG-PET scans. If prospectively validated with a larger cohort of patients, similar models could provide decision support in a clinical setting.

Artificial intelligence for automatic cerebral ventricle segmentation and volume calculation: a clinical tool for the evaluation of pediatric hydrocephalus

OBJECTIVE

Imaging evaluation of the cerebral ventricles is important for clinical decision-making in pediatric hydrocephalus. Although quantitative measurements of ventricular size, over time, can facilitate objective comparison, automated tools for calculating ventricular volume are not structured for clinical use. The authors aimed to develop a fully automated deep learning (DL) model for pediatric cerebral ventricle segmentation and volume calculation for widespread clinical implementation across multiple hospitals.

METHODS

The study cohort consisted of 200 children with obstructive hydrocephalus from four pediatric hospitals, along with 199 controls. Manual ventricle segmentation and volume calculation values served as "ground truth" data. An encoder-decoder convolutional neural network architecture, in which T2-weighted MR images were used as input, automatically delineated the ventricles and output volumetric measurements. On a held-out test set, segmentation accuracy was assessed using the Dice similarity coefficient (0 to 1) and volume calculation was assessed using linear regression. Model generalizability was evaluated on an external MRI data set from a fifth hospital. The DL model performance was compared against FreeSurfer research segmentation software.

RESULTS

Model segmentation performed with an overall Dice score of 0.901 (0.946 in hydrocephalus, 0.856 in controls). The model generalized to external MR images from a fifth pediatric hospital with a Dice score of 0.926. The model was more accurate than FreeSurfer, with faster operating times (1.48 seconds per scan).

CONCLUSIONS

The authors present a DL model for automatic ventricle segmentation and volume calculation that is more accurate and rapid than currently available methods. With near-immediate volumetric output and reliable performance across institutional scanner types, this model can be adapted to the real-time clinical evaluation of hydrocephalus and improve clinician workflow.

Application of Deep Learning to Predict Standardized Uptake Value Ratio and Amyloid Status on 18F-Florbetapir PET Using ADNI Data

BACKGROUND AND PURPOSE

Cortical amyloid quantification on PET by using the standardized uptake value ratio is valuable for research studies and clinical trials in Alzheimer disease. However, it is resource intensive, requiring co-registered MR imaging data and specialized segmentation software. We investigated the use of deep learning to automatically quantify standardized uptake value ratio and used this for classification.

MATERIALS AND METHODS

Using the Alzheimer's Disease Neuroimaging Initiative dataset, we identified 2582 18F-florbetapir PET scans, which were separated into positive and negative cases by using a standardized uptake value ratio threshold of 1.1. We trained convolutional neural networks (ResNet-50 and ResNet-152) to predict standardized uptake value ratio and classify amyloid status. We assessed performance based on network depth, number of PET input slices, and use of ImageNet pretraining. We also assessed human performance with 3 readers in a subset of 100 randomly selected cases.

RESULTS

We have found that 48% of cases were amyloid positive. The best performance was seen for ResNet-50 by using regression before classification, 3 input PET slices, and pretraining, with a standardized uptake value ratio root-mean-square error of 0.054, corresponding to 95.1% correct amyloid status prediction. Using more than 3 slices did not improve performance, but ImageNet initialization did. The best trained network was more accurate than humans (96% versus a mean of 88%, respectively).

CONCLUSIONS

Deep learning algorithms can estimate standardized uptake value ratio and use this to classify 18F-florbetapir PET scans. Such methods have promise to automate this laborious calculation, enabling quantitative measurements rapidly and in settings without extensive image processing manpower and expertise.

Five percent dextrose maximizes dose delivery of Yttrium-90 resin microspheres and reduces rates of premature stasis compared to sterile water

Resin Yttrium-90 (Y90) microspheres have historically been infused using sterile water (H₂O). In 2013, recommendations expanded to allow delivery with 5% dextrose in water (D₅W). In this retrospective study, we hypothesized that D₅W would improve Y90 delivery with a lower incidence of stasis. We reviewed 190 resin Y90 infusions using H₂O (n=137) or D₅W (n=53). Y90 dosimetry was calculated using the body surface area method. Infusion was halted if intra-arterial stasis was fluoroscopically identified prior to clearing the vial. Differences between H₂O and D₅W groups were calculated for activity prescription, percentage of cases reaching stasis, and percentage delivery of prescribed activity using z- and t-test comparisons, with α=0.05. Thirty-one of 137 H₂O infusions developed stasis compared to 2 of 53 with D₅W (z=3.07, p=1.05E-03). D₅W also had a significantly higher prescribed activity than H₂O [28.2 millicuries (mCi) vs. 20.4 mCi, respectively; t=5.0, p=1.1E-6]. D₅W had a higher delivery percentage of the prescribed dose compared to H₂O (101.5 vs. 92.7%, respectively; t=3.8, p=1.92E-4). In conclusion, resin microsphere infusion utilizing D₅W has a significantly lower rate of stasis than H₂O and results in more complete dose delivery. D₅W is preferable to H₂O for resin microsphere infusion.

Procedural Impact of a Dedicated Interventional Oncology Service Line in a National Cancer Institute Comprehensive Cancer Center

PURPOSE

We tested the hypothesis that establishing a dedicated interventional oncology (IO) clinical service line would increase clinic visits and procedural volumes at a single quaternary care academic medical center.

METHODS

Two time periods were defined: July 2012 to June 2013 (pre-IO clinic) and July 2013 to June 2014 (first year of dedicated IO service). Staff was recruited, and clinic space was provided in the institution's comprehensive cancer center. Clinic visits and procedure numbers were documented using the institution's electronic medical record and billing forms. IO procedures included were transarterial chemoembolization, Y-90 radioembolization, perfusion mapping for Y-90, portal vein embolization, and bland embolization. We compared changes in clinic visit and procedure numbers using paired t tests. Changes after IO initiation were compared to 1-year changes in the Medicare 5% Limited Data Set by cross-referencing Current Procedure Terminology and International Classification of Diseases codes in 2012 and 2013.

RESULTS

Clinic visits increased from 9 to 204 (P = .003, t = 8.89, df = 3). Procedures increased from 60 to 239 (P = .018, t = 3.85, df = 4). Procedural volumes increased at least 150% for each subtype. The volumes in the 5% Limited Data Set did not change significantly over the 2-year period (443 to 385, P > .05).

CONCLUSIONS

The establishment of a dedicated IO service significantly increased clinic visits and procedural volumes. National trends were unchanged, suggesting that the impact of our program was not part of a sudden increase of IO procedures.

Impact of family structure and common environment on heritability estimation for neuroimaging genetics studies using Sequential Oligogenic Linkage Analysis Routines

Imaging genetics is an emerging methodological field that combines genetic information with medical imaging-derived metrics to understand how genetic factors impact observable phenotypes. In order for a trait to be a reasonable phenotype in an imaging genetics study, it must be heritable: at least some proportion of its variance must be due to genetic influences. The Sequential Oligogenic Linkage Analysis Routines (SOLAR) imaging genetics software can estimate the heritability of a trait in complex pedigrees. We investigate the ability of SOLAR to accurately estimate heritability and common environmental effects on simulated imaging phenotypes in various family structures. We found that heritability is reliably estimated with small family-based studies of 40 to 80 individuals, though subtle differences remain between the family structures. In an imaging application analysis, we found that with 80 subjects in any of the family structures, estimated heritability of white matter fractional anisotropy was biased by <10% for every region of interest. Results from these studies can be used when investigators are evaluating power in planning genetic analyzes.

Epistatic genetic effects among Alzheimer's candidate genes

Background

Novel risk variants for late-onset Alzheimer’s disease (AD) have been identified and replicated in genome-wide association studies. Recent work has begun to address the relationship between these risk variants and biomarkers of AD, though results have been mixed. The aim of the current study was to characterize single marker and epistatic genetic effects between the top candidate Single Nucleotide Polymorphisms (SNPs) in relation to amyloid deposition.

Methods

We used a combined dataset across ADNI-1 and ADNI-2, and looked within each dataset separately to validate identified genetic effects. Amyloid was quantified using data acquired by Positron Emission Tomography (PET) with ¹⁸F-AV-45.

Results

Two SNP-SNP interactions reached significance when correcting for multiple comparisons, BIN1 (rs7561528, rs744373) xPICALM (rs7851179). Carrying the minor allele in BIN1 was related to higher levels of amyloid deposition, however only in non-carriers of the protective PICALM minor allele.

Conclusions

Our results support previous research suggesting these candidate SNPs do not show single marker associations with amyloid pathology. However, we provide evidence for a novel interaction between PICALM and BIN1 in relation to amyloid deposition. Risk related to the BIN1 minor allele appears to be mitigated in the presence of the PICALM protective variant. In that way, variance in amyloid plaque burden can be better classified within the context of a complex genetic background. Efforts to model cumulative risk for AD should explicitly account for this epistatic effect, and future studies should explicitly test for such effects whenever statistically feasible.

Associations between KCNJ6 (GIRK2) gene polymorphisms and pain-related phenotypes

G-protein coupled inwardly rectifying potassium (GIRK) channels are effectors determining degree of analgesia experienced upon opioid receptor activation by endogenous and exogenous opioids. The impact of GIRK-related genetic variation on human pain responses has received little research attention. We used a tag single nucleotide polymorphism (SNP) approach to comprehensively examine pain-related effects of KCNJ3 (GIRK1) and KCNJ6 (GIRK2) gene variation. Forty-one KCNJ3 and 69 KCNJ6 tag SNPs were selected, capturing the known variability in each gene. The primary sample included 311 white patients undergoing total knee arthroplasty in whom postsurgical oral opioid analgesic medication order data were available. Primary sample findings were then replicated in an independent white sample of 63 healthy pain-free individuals and 75 individuals with chronic low back pain (CLBP) who provided data regarding laboratory acute pain responsiveness (ischemic task) and chronic pain intensity and unpleasantness (CLBP only). Univariate quantitative trait analyses in the primary sample revealed that 8 KCNJ6 SNPs were significantly associated with the medication order phenotype (P < .05); overall effects of the KCNJ6 gene (gene set-based analysis) just failed to reach significance (P = .054). No significant KCNJ3 effects were observed. A continuous GIRK Related Risk Score (GRRS) was derived in the primary sample to summarize each individual's number of KCNJ6 "pain risk" alleles. This GRRS was applied to the replication sample, which revealed significant associations (P < .05) between higher GRRS values and lower acute pain tolerance and higher CLBP intensity and unpleasantness. Results suggest further exploration of the impact of KCNJ6 genetic variation on pain outcomes is warranted.