Comorbid amyloid with cerebrovascular disease in domain-specific cognitive and neuropsychiatric disturbances: a cross-sectional memory clinic study.

Dementia is a multifactorial disorder that is likely influenced by both Alzheimer's disease (AD) and vascular pathologies. We evaluated domain-specific cognitive and neuropsychiatric dysfunction using a two-neuroimaging biomarker construct (beta-amyloid [Aß] and cerebrovascular disease [CeVD]). We analyzed data from 216 memory clinic participants (mean age = 75.9 ± 6.9; 56.5% female) with neuropsychological and neuropsychiatric assessments, 3T-MRI, and Aß-PET imaging. Structural equation modeling showed that the largest Aß (A+) effect was on memory (B = -1.50) and apathy (B = 0.26), whereas CeVD effects were largest on language (B = -1.62) and hyperactivity (B = 0.32). Group comparisons showed that the A+C+ group had greater memory impairment (B = -1.55), hyperactivity (B = 0.79), and apathy (B = 0.74) compared to A-C+; and greater language impairment (B = -1.26) compared to A+C-. These potentially additive effects of Aß and CeVD burden underline the importance of early detection and treatment of Aß alongside optimal control of vascular risk factors as a potential strategy in preventing cognitive and neurobehavioral impairment.

Longitudinal associations between ß-amyloid and cortical thickness in mild cognitive impairment.

How beta-amyloid accumulation influences brain atrophy in Alzheimer's disease remains contentious with conflicting findings. We aimed to elucidate the correlations of regional longitudinal atrophy with cross-sectional regional and global amyloid in individuals with mild cognitive impairment and no cognitive impairment. We hypothesized that greater cortical thinning over time correlated with greater amyloid deposition, particularly within Alzheimer's disease characteristic regions in mild cognitive impairment, and weaker or no correlations in those with no cognitive impairment. 45 patients with mild cognitive impairment and 12 controls underwent a cross-sectional [11C]-Pittsburgh Compound B PET and two retrospective longitudinal structural imaging (follow-up: 23.65 ± 2.04 months) to assess global/regional amyloid and regional cortical thickness, respectively. Separate linear mixed models were constructed to evaluate relationships of either global or regional amyloid with regional cortical thinning longitudinally. In patients with mild cognitive impairment, regional amyloid in the right banks of the superior temporal sulcus was associated with longitudinal cortical thinning in the right medial orbitofrontal cortex (P = 0.04 after False Discovery Rate correction). In the mild cognitive impairment group, greater right banks amyloid burden and less cortical thickness in the right medial orbitofrontal cortex showed greater visual and verbal memory decline over time, which was not observed in controls. Global amyloid was not associated with longitudinal cortical thinning in any locations in either group. Our findings indicate an increasing influence of amyloid on neurodegeneration and memory along the preclinical to prodromal spectrum. Future multimodal studies that include additional biomarkers will be well-suited to delineate the interplay between various pathological processes and amyloid and memory decline, as well as clarify their additive or independent effects along the disease deterioration.

Brain atrophy and white matter hyperintensities are independently associated with plasma neurofilament light chain in an Asian cohort of cognitively impaired patients with concomitant cerebral small vessel disease.

Introduction: Plasma neurofilament light chain (NfL) is a potential biomarker for neurodegeneration in Alzheimer's disease (AD), ischemic stroke, and non-dementia cohorts with cerebral small vessel disease (CSVD). However, studies of AD in populations with high prevalence of concomitant CSVD to evaluate associations of brain atrophy, CSVD, and amyloid beta (Aß) burden on plasma NfL are lacking. Methods: Associations were tested between plasma NfL and brain Aß, medial temporal lobe atrophy (MTA) as well as neuroimaging features of CSVD, including white matter hyperintensities (WMH), lacunes, and cerebral microbleeds. Results: We found that participants with either MTA (defined as MTA score ≥2; neurodegeneration [N]+WMH-) or WMH (cut-off for log-transformed WMH volume at 50th percentile; N-WMH+) manifested increased plasma NfL levels. Participants with both pathologies (N+WMH+) showed the highest NfL compared to N+WMH-, N-WMH+, and N-WMH- individuals. Discussion: Plasma NfL has potential utility in stratifying individual and combined contributions of AD pathology and CSVD to cognitive impairment.

Comparison of quantitative parameters and radiomic features as inputs into machine learning models to predict the Gleason score of prostate cancer lesions.

INTRODUCTION: The classification of prostate cancer (PCa) lesions using Prostate Imaging Reporting and Data System (PI-RADS) suffers from poor inter-reader agreement. This study compared quantitative parameters or radiomic features from multiparametric magnetic resonance imaging (mpMRI) or positron emission tomography (PET), as inputs into machine learning (ML) to predict the Gleason scores (GS) of detected lesions for improved PCa lesion classification. METHODS: 20 biopsy-confirmed PCa subjects underwent imaging before radical prostatectomy. A pathologist assigned GS from tumour tissue. Two radiologists and one nuclear medicine physician delineated the lesions on the mpMR and PET images, yielding 45 lesion inputs. Seven quantitative parameters were extracted from the lesions, namely T2-weighted (T2w) image intensity, apparent diffusion coefficient (ADC), transfer constant (KTRANS), efflux rate constant (Kep), and extracellular volume ratio (Ve) from mpMR images, and SUVmean and SUVmax from PET images. Eight radiomic features were selected out of 109 radiomic features from T2w, ADC and PET images. Quantitative parameters or radiomic features, with risk factors of age, prostate-specific antigen (PSA), PSA density and volume, of 45 different lesion inputs were input in different combinations into four ML models - Decision Tree (DT), Support Vector Machine (SVM), k-Nearest-Neighbour (kNN), Ensembles model (EM). RESULTS: SUVmax yielded the highest accuracy in discriminating detected lesions. Among the 4 ML models, kNN yielded the highest accuracies of 0.929 using either quantitative parameters or radiomic features with risk factors as input. CONCLUSIONS: ML models' performance is dependent on the input combinations and risk factors further improve ML classification accuracy.

PET image enhancement using artificial intelligence for better characterization of epilepsy lesions.

Introduction: [18F]fluorodeoxyglucose ([18F]FDG) brain PET is used clinically to detect small areas of decreased uptake associated with epileptogenic lesions, e.g., Focal Cortical Dysplasias (FCD) but its performance is limited due to spatial resolution and low contrast. We aimed to develop a deep learning-based PET image enhancement method using simulated PET to improve lesion visualization. Methods: We created 210 numerical brain phantoms (MRI segmented into 9 regions) and assigned 10 different plausible activity values (e.g., GM/WM ratios) resulting in 2100 ground truth high quality (GT-HQ) PET phantoms. With a validated Monte-Carlo PET simulator, we then created 2100 simulated standard quality (S-SQ) [18F]FDG scans. We trained a ResNet on 80% of this dataset (10% used for validation) to learn the mapping between S-SQ and GT-HQ PET, outputting a predicted HQ (P-HQ) PET. For the remaining 10%, we assessed Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index Measure (SSIM), and Root Mean Squared Error (RMSE) against GT-HQ PET. For GM and WM, we computed recovery coefficients (RC) and coefficient of variation (COV). We also created lesioned GT-HQ phantoms, S-SQ PET and P-HQ PET with simulated small hypometabolic lesions characteristic of FCDs. We evaluated lesion detectability on S-SQ and P-HQ PET both visually and measuring the Relative Lesion Activity (RLA, measured activity in the reduced-activity ROI over the standard-activity ROI). Lastly, we applied our previously trained ResNet on 10 clinical epilepsy PETs to predict the corresponding HQ-PET and assessed image quality and confidence metrics. Results: Compared to S-SQ PET, P-HQ PET improved PNSR, SSIM and RMSE; significatively improved GM RCs (from 0.29 ± 0.03 to 0.79 ± 0.04) and WM RCs (from 0.49 ± 0.03 to 1 ± 0.05); mean COVs were not statistically different. Visual lesion detection improved from 38 to 75%, with average RLA decreasing from 0.83 ± 0.08 to 0.67 ± 0.14. Visual quality of P-HQ clinical PET improved as well as reader confidence. Conclusion: P-HQ PET showed improved image quality compared to S-SQ PET across several objective quantitative metrics and increased detectability of simulated lesions. In addition, the model generalized to clinical data. Further evaluation is required to study generalization of our method and to assess clinical performance in larger cohorts.

The Associations Between Intracranial Stenosis, Brain Amyloid-beta, and Cognition in a Memory Clinic Sample.

BACKGROUND: Intracranial stenosis (ICS) and brain amyloid-beta (Aß) have been associated with cognition and dementia. We aimed to investigate the association between ICS and brain Aß and their independent and joint associations with cognition. METHODS: We conducted a cross-sectional study of 185 patients recruited from a memory clinic. ICS was measured on 3-dimensional time-of-flight magnetic resonance angiography and defined as stenosis ≥50%. Brain Aß was measured with [ 11 C] Pittsburgh compound B-positron emission tomography imaging. Cognition was assessed with a locally validated neuropsychological battery. RESULTS: A total of 17 (9.2%) patients had ICS, and the mean standardized uptake value ratio was 1.4 (±0.4 SD). ICS was not significantly associated with brain Aß deposition. ICS was significantly associated with worse global cognition (ß: -1.26, 95% CI: -2.25; -0.28, P =0.013), executive function (ß: -1.04, 95% CI: -1.86; -0.22, P =0.015) and visuospatial function (ß: -1.29, 95% CI: -2.30; -0.27, P =0.015). Moreover, in ICS patients without dementia (n=8), the presence of Aß was associated with worse performance on visuomotor speed. CONCLUSIONS: ICS was significantly associated with worse cognition and showed interaction with brain Aß such that patients with both pathologies performed worse on visuomotor speed specifically in those without dementia. Further studies may clarify if ICS and brain Aß deposition indeed have a synergistic association with cognition.

Evolving operational guidance and experiences for radiology and nuclear medicine facilities in response to and beyond the COVID-19 pandemic.

The resulting pandemic from the novel severe acute respiratory coronavirus 2, SARS-CoV-2 (COVID-19), continues to exert a strain on worldwide health services due to the incidence of hospitalization and mortality associated with infection. The aim of clinical services throughout the period of the pandemic and likely beyond to endemic infections as the situation stabilizes is to enhance safety aspects to mitigate transmission of COVID-19 while providing a high quality of service to all patients (COVID-19 positive and negative) while still upholding excellent medical standards. In order to achieve this, new strategies of clinical service operation are essential. Researchers have published peer-reviewed reference materials such as guidelines, experiences and advice to manage the resulting issues from the unpredictable challenges presented by the pandemic. There is a range of international guidance also from professional medical organizations, including best practice and advice in order to help imaging facilities adjust their standard operating procedures and workflows in line with infection control principles. This work provides a broad review of the main sources of advice and guidelines for radiology and nuclear medicine facilities during the pandemic, and also of rapidly emerging advice and local/national experiences as facilities begin to resume previously canceled non-urgent services as well as effects on imaging research.

Comparison of the performances of machine learning and deep learning in improving the quality of low dose lung cancer PET images.

PURPOSE: To compare the performances of machine learning (ML) and deep learning (DL) in improving the quality of low dose (LD) lung cancer PET images and the minimum counts required. MATERIALS AND METHODS: 33 standard dose (SD) PET images, were used to simulate LD PET images at seven-count levels of 0.25, 0.5, 1, 2, 5, 7.5 and 10 million (M) counts. Image quality transfer (IQT), a ML algorithm that uses decision tree and patch-sampling was compared to two DL networks-HighResNet (HRN) and deep-boosted regression (DBR). Supervised training was performed by training the ML and DL algorithms with matched-pair SD and LD images. Image quality evaluation and clinical lesion detection tasks were performed by three readers. Bias in 53 radiomic features, including mean SUV, was evaluated for all lesions. RESULTS: ML- and DL-estimated images showed higher signal and smaller error than LD images with optimal image quality recovery achieved using LD down to 5 M counts. True positive rate and false discovery rate were fairly stable beyond 5 M counts for the detection of small and large true lesions. Readers rated average or higher ratings to images estimated from LD images of count levels above 5 M only, with higher confidence in detecting true lesions. CONCLUSION: LD images with a minimum of 5 M counts (8.72 MBq for 10 min scan or 25 MBq for 3 min scan) are required for optimal clinical use of ML and DL, with slightly better but more varied performance shown by DL.

Validation of deep learning-based nonspecific estimates for amyloid burden quantification with longitudinal data.

PURPOSE: To validate our previously proposed method of quantifying amyloid-beta (Aß) load using nonspecific (NS) estimates generated with convolutional neural networks (CNNs) using [18F]Florbetapir scans from longitudinal and multicenter ADNI data. METHODS: 188 paired MR (T1-weighted and T2-weighted) and PET images were downloaded from the ADNI3 dataset, of which 49 subjects had 2 time-point scans. 40 Aß- subjects with low specific uptake were selected for training. Multimodal ScaleNet (SN) and monomodal HighRes3DNet (HRN), using either T1-weighted or T2-weighted MR images as inputs) were trained to map structural MR to NS-PET images. The optimized SN and HRN networks were used to estimate the NS for all scans and then subtracted from SUVr images to determine the specific amyloid load (SAßL) images. The association of SAßL with various cognitive and functional test scores was evaluated using Spearman analysis, as well as the differences in SAßL with cognitive test scores for 49 subjects with 2 time-point scans and sensitivity analysis. RESULTS: SAßL derived from both SN and HRN showed higher association with memory-related cognitive test scores compared to SUVr. However, for longitudinal scans, only SAßL estimated from multimodal SN consistently performed better than SUVr for all memory-related cognitive test scores. CONCLUSIONS: Our proposed method of quantifying Aß load using NS estimated from CNN correlated better than SUVr with cognitive decline for both static and longitudinal data, and was able to estimate NS of [18F]Florbetapir. We suggest employing multimodal networks with both T1-weighted and T2-weighted MR images for better NS estimation.

11C-Metomidate PET-CT versus adrenal vein sampling to subtype primary aldosteronism: a prospective clinical trial.

OBJECTIVE: Adrenal vein sampling (AVS) is recommended to subtype primary aldosteronism, but it is technically challenging. We compared 11C-Metomidate-PET-computed tomography (PET-CT) and AVS for subtyping of primary aldosteronism. METHODS: Patients with confirmed primary aldosteronism underwent both AVS and 11C-Metomidate PET-CT (post-dexamethasone). All results were reviewed at a multidisciplinary meeting to decide on final subtype diagnosis. Primary outcome was accuracy of PET versus AVS to diagnosis of unilateral primary aldosteronism based on post-surgical biochemical cure. Secondary outcome was accuracy of both tests to final subtype diagnosis. RESULTS: All 25 patients recruited underwent PET and successful AVS (100%). Final diagnosis was unilateral in 22 patients, bilateral in two and indeterminate in one due to discordant lateralization. Twenty patients with unilateral primary aldosteronism underwent surgery, with 100% complete biochemical success, and 75% complete/partial clinical success. For the primary outcome, sensitivity of PET was 80% [95% confidence interval (95% CI): 56.3-94.3] and AVS was 75% (95% CI: 50.9-91.3). For the secondary outcome, sensitivity and specificity of PET was 81.9% (95% CI: 59.7-94.8) and 100% (95% CI: 15.8-100), and AVS was 68.2% (95% CI: 45.1-86.1) and 100% (95% CI: 15.8-100), respectively. Twelve out of 20 (60%) patients had both PET and AVS lateralization, four (20%) PET-only, three (15%) AVS-only, while one patient did not lateralize on PET or AVS. Post-surgery outcomes did not differ between patients identified by either test. CONCLUSION: In our pilot study, 11C-Metomidate PET-CT performed comparably to AVS, and this should be validated in larger studies. PET identified patients with unilateral primary aldosteronism missed on AVS, and these tests could be used together to identify more patients with unilateral primary aldosteronism. VIDEO ABSTRACT: http://links.lww.com/HJH/B918.

Comparison of Three Automated Approaches for Classification of Amyloid-PET Images.

Automated amyloid-PET image classification can support clinical assessment and increase diagnostic confidence. Three automated approaches using global cut-points derived from Receiver Operating Characteristic (ROC) analysis, machine learning (ML) algorithms with regional SUVr values, and deep learning (DL) network with 3D image input were compared under various conditions: number of training data, radiotracers, and cohorts. 276 [11C]PiB and 209 [18F]AV45 PET images from ADNI database and our local cohort were used. Global mean and maximum SUVr cut-points were derived using ROC analysis. 68 ML models were built using regional SUVr values and one DL network was trained with classifications of two visual assessments - manufacturer's recommendations (gray-scale) and with visually guided reference region scaling (rainbow-scale). ML-based classification achieved similarly high accuracy as ROC classification, but had better convergence between training and unseen data, with a smaller number of training data. Naïve Bayes performed the best overall among the 68 ML algorithms. Classification with maximum SUVr cut-points yielded higher accuracy than with mean SUVr cut-points, particularly for cohorts showing more focal uptake. DL networks can support the classification of definite cases accurately but performed poorly for equivocal cases. Rainbow-scale standardized image intensity scaling and improved inter-rater agreement. Gray-scale detects focal accumulation better, thus classifying more amyloid-positive scans. All three approaches generally achieved higher accuracy when trained with rainbow-scale classification. ML yielded similarly high accuracy as ROC, but with better convergence between training and unseen data, and further work may lead to even more accurate ML methods.

Cerebrovascular disease in suspected non-Alzheimer's pathophysiology and cognitive decline over time.

BACKGROUND: The underlying cause of cognitive decline in individuals who are positive for biomarkers of neurodegeneration (N) but negative for biomarkers of amyloid-beta (A), designated as Suspected non-Alzheimer's pathophysiology (SNAP), remains unclear. We evaluate whether cerebrovascular disease (CeVD) is more prevalent in those with SNAP compared to A-N- and A+N+ individuals and whether CeVD is associated with cognitive decline over time in SNAP patients. METHODS: A total of 216 individuals from a prospective memory clinic cohort (mean [SD] age, 72.7 [7.3] years, 100 women [56.5%]) were included and were diagnosed as no cognitive impairment (NCI), cognitive impairment no dementia (CIND), Alzheimer's dementia (AD) or vascular dementia (VaD). All individuals underwent clinical evaluation and neuropsychological assessment annually for up to 5 years. Carbon 11-labeled Pittsburgh Compound B ([11 C]-PiB) or [18 F]-flutafuranol-positron emission spectrometry imaging was performed to ascertain amyloid-beta status. Magnetic resonance imaging was performed to assess neurodegeneration as measured by medial temporal atrophy ≥2, as well as significant CeVD (sCeVD) burden, defined by cortical infarct count ≥1, Fazekas score ≥2, lacune count ≥2 or cerebral microbleed count ≥2. RESULTS: Of the 216 individuals, 50 (23.1%) A-N+ were (SNAP), 93 (43.1%) A-N-, 36 (16.7%) A+N- and 37 (17.1%) A+N+. A+N+ individuals were significantly older, while A+N+ and SNAP individuals were more likely to have dementia. The SNAP group had a higher prevalence of sCeVD (90.0%) compared to A-N-. Moreover, SNAP individuals with sCeVD had significantly steeper decline in global cognition compared to A-N- over 5 years (p = 0.042). CONCLUSIONS: These findings suggest that CeVD is a contributing factor to cognitive decline in SNAP. Therefore, SNAP individuals should be carefully assessed and treated for CeVD.

The Association Between Standard Electrocardiography and Cerebral Small Vessel Disease in a Memory Clinic Study.

BACKGROUND: P-wave terminal force in lead V1 (PTFV1) on electrocardiography has been associated with atrial fibrillation and ischemic stroke. OBJECTIVE: To investigate whether PTFV1 is associated with cerebral small vessel disease (CSVD) markers and etiological subtypes of cognitive impairment and dementia. METHODS: Participants were recruited from ongoing memory clinic study between August 2010 to January 2019. All participants underwent physical and medical evaluation along with an electrocardiography and 3 T brain magnetic resonance imaging. Participants were classified as no cognitive impairment, cognitive impairment no dementia, vascular cognitive impairment no dementia, and dementia subtypes (Alzheimer's disease and vascular dementia). Elevated PTFV1 was defined as > 4,000µV×ms and measured manually on ECG. RESULTS: Of 408 participants, 78 (19.1%) had elevated PTFV1 (37 women [47%]; mean [SD] age, 73.8 [7.2] years). The participants with elevated PTFV1 had higher burden of lacunes, cerebral microbleeds (CMB), and cortical microinfarcts. As for the CMB location, persons with strictly deep CMB and mixed CMB had significantly higher PTFV1 than those with no CMB (p = 0.005, p = 0.007). Regardless of adjustment for cardiovascular risk factors and/or heart diseases, elevated PTFV1 was significantly associated with presence of CMB (odds ratio, 2.26; 95% CI,1.33-3.91). CONCLUSION: Elevated PTFV1 was associated with CSVD, especially deep CMB. PTFV1 in vascular dementia was also higher compared to Alzheimer's disease. Thus, PTFV1 might be a potential surrogate marker of brain-heart connection and vascular brain damage.

Parametric mapping for 11C-metomidate PET-computed tomography imaging in the study of primary aldosteronism.

OBJECTIVE: 11C-metomidate (11C-MTO) PET-computed tomography (CT) imaging has shown good sensitivity and specificity for the classification of bilateral or unilateral overexpression of aldosterone. This work seeks to investigate the usefulness of parametric maps via kinetic modeling of 11C-metomidate data into the clinical diagnosis pathway. METHODS: Twenty-five patients were injected with 172 ± 12 MBq of 11C-metomidate and a dynamic PET scan performed of the adrenal glands. A blood time-activity curve was drawn from a volume of interest in the left ventricle and converted to a plasma time-activity curve. Metabolite correction was performed with a population-based correction. We performed regional-based graphical Patlak analysis to calculate the regional uptake rate constant Ki(REG), and also calculated parametric maps of Ki(VOX) using a voxel-based technique. RESULTS: Comparison of Ki(REG), and the maximum lesion voxel from parametric maps Ki(mVOX) demonstrated a high correlation for all subjects (r2 = 0.96). Ki(mVOX) allowed differentiation between cases of active and inactive unilateral adenoma when compared to bilateral hyperplasia (P < 0.017), a feature not observed with standardized uptake ratios (SUVmax) analysis. Ki(mVOX) demonstrated a poor correlation of 0.68 with SUVmax, indicating the differences through the use of static and dynamic imaging. Three false-negative cases based on SUV analysis indicated that Ki(mVOX) was able to successfully differentiate the clinical presentation for these cases. CONCLUSION: Our work demonstrates that parametric Ki(VOX) was able to successfully differentiate between patients with bilateral hyperplasia and patients with unilateral adrenal adenoma in our cohort and that Ki may be considered be an additional useful metric to SUV in 11C-metomidate PET-CT imaging.

Comparison of metrics for the evaluation of medical segmentations using prostate MRI dataset.

Nine previously proposed segmentation evaluation metrics, targeting medical relevance, accounting for holes, and added regions or differentiating over- and under-segmentation, were compared with 24 traditional metrics to identify those which better capture the requirements for clinical segmentation evaluation. Evaluation was first performed using 2D synthetic shapes to highlight features and pitfalls of the metrics with known ground truths (GTs) and machine segmentations (MSs). Clinical evaluation was then performed using publicly-available prostate images of 20 subjects with MSs generated by 3 different deep learning networks (DenseVNet, HighRes3DNet, and ScaleNet) and GTs drawn by 2 readers. The same readers also performed the 2D visual assessment of the MSs using a dual negative-positive grading of -5 to 5 to reflect over- and under-estimation. Nine metrics that correlated well with visual assessment were selected for further evaluation using 3 different network ranking methods - based on a single metric, normalizing the metric using 2 GTs, and ranking the network based on a metric then averaging, including leave-one-out evaluation. These metrics yielded consistent ranking with HighRes3DNet ranked first then DenseVNet and ScaleNet using all ranking methods. Relative volume difference yielded the best positivity-agreement and correlation with dual visual assessment, and thus is better for providing over- and under-estimation. Interclass Correlation yielded the strongest correlation with the absolute visual assessment (0-5). Symmetric-boundary dice consistently yielded good discrimination of the networks for all three ranking methods with relatively small variations within network. Good rank discrimination may be an additional metric feature required for better network performance evaluation.

Plasma P-tau181 to Aß42 ratio is associated with brain amyloid burden and hippocampal atrophy in an Asian cohort of Alzheimer's disease patients with concomitant cerebrovascular disease.

INTRODUCTION: There is increasing evidence that phosphorylated tau (P-tau181) is a specific biomarker for Alzheimer's disease (AD) pathology, but its potential utility in non-White patient cohorts and patients with concomitant cerebrovascular disease (CeVD) is unknown. METHODS: Single molecule array (Simoa) measurements of plasma P-tau181, total tau, amyloid beta (Aß)40 and Aß42, as well as derived ratios were correlated with neuroimaging modalities indicating brain amyloid (Aß+), hippocampal atrophy, and CeVD in a Singapore-based cohort of non-cognitively impaired (NCI; n = 43), cognitively impaired no dementia (CIND; n = 91), AD (n = 44), and vascular dementia (VaD; n = 22) subjects. RESULTS: P-tau181/Aß42 ratio showed the highest area under the curve (AUC) for Aß+ (AUC = 0.889) and for discriminating between AD Aß+ and VaD Aß- subjects (AUC = 0.903). In addition, P-tau181/Aß42 ratio was associated with hippocampal atrophy. None of the biomarkers was associated with CeVD. DISCUSSION: Plasma P-tau181/Aß42 ratio may be a noninvasive means of identifying AD with elevated brain amyloid in populations with concomitant CeVD.

Improved amyloid burden quantification with nonspecific estimates using deep learning.

PURPOSE: Standardized uptake value ratio (SUVr) used to quantify amyloid-ß burden from amyloid-PET scans can be biased by variations in the tracer's nonspecific (NS) binding caused by the presence of cerebrovascular disease (CeVD). In this work, we propose a novel amyloid-PET quantification approach that harnesses the intermodal image translation capability of convolutional networks to remove this undesirable source of variability. METHODS: Paired MR and PET images exhibiting very low specific uptake were selected from a Singaporean amyloid-PET study involving 172 participants with different severities of CeVD. Two convolutional neural networks (CNN), ScaleNet and HighRes3DNet, and one conditional generative adversarial network (cGAN) were trained to map structural MR to NS PET images. NS estimates generated for all subjects using the most promising network were then subtracted from SUVr images to determine specific amyloid load only (SAßL). Associations of SAßL with various cognitive and functional test scores were then computed and compared to results using conventional SUVr. RESULTS: Multimodal ScaleNet outperformed other networks in predicting the NS content in cortical gray matter with a mean relative error below 2%. Compared to SUVr, SAßL showed increased association with cognitive and functional test scores by up to 67%. CONCLUSION: Removing the undesirable NS uptake from the amyloid load measurement is possible using deep learning and substantially improves its accuracy. This novel analysis approach opens a new window of opportunity for improved data modeling in Alzheimer's disease and for other neurodegenerative diseases that utilize PET imaging.

Validation of low-dose lung cancer PET-CT protocol and PET image improvement using machine learning.

PURPOSE: To conduct a simplified lesion-detection task of a low-dose (LD) PET-CT protocol for frequent lung screening using 30% of the effective PETCT dose and to investigate the feasibility of increasing clinical value of low-statistics scans using machine learning. METHODS: We acquired 33 SD PET images, of which 13 had actual LD (ALD) PET, and simulated LD (SLD) PET images at seven different count levels from the SD PET scans. We employed image quality transfer (IQT), a machine learning algorithm that performs patch-regression to map parameters from low-quality to high-quality images. At each count level, patches extracted from 23 pairs of SD/SLD PET images were used to train three IQT models - global linear, single tree, and random forest regressions with cubic patch sizes of 3 and 5 voxels. The models were then used to estimate SD images from LD images at each count level for 10 unseen subjects. Lesion-detection task was carried out on matched lesion-present and lesion-absent images. RESULTS: LD PET-CT protocol yielded lesion detectability with sensitivity of 0.98 and specificity of 1. Random forest algorithm with cubic patch size of 5 allowed further 11.7% reduction in the effective PETCT dose without compromising lesion detectability, but underestimated SUV by 30%. CONCLUSION: LD PET-CT protocol was validated for lesion detection using ALD PET scans. Substantial image quality improvement or additional dose reduction while preserving clinical values can be achieved using machine learning methods though SUV quantification may be biased and adjustment of our research protocol is required for clinical use.

Head-to-head comparison of amplified plasmonic exosome Aß42 platform and single-molecule array immunoassay in a memory clinic cohort.

BACKGROUND AND PURPOSE: Various blood biomarkers reflecting brain amyloid-ß (Aß) load have recently been proposed with promising results. However, to date, no comparative study amongst blood biomarkers has been reported. Our objective was to examine the diagnostic performance and cost effectiveness of three blood biomarkers on the same cohort. METHODS: Using the same cohort (n = 68), the performances of the single-molecule array (Simoa) Aß40, Aß42, Aß42/Aß40 and the amplified plasmonic exosome (APEX) Aß42 blood biomarkers were compared using amyloid positron emission tomography (PET) as the reference standard. The extent to which these blood tests can reduce the recruitment cost of clinical trials was also determined by identifying amyloid positive (Aß+) participants. RESULTS: Compared to Simoa biomarkers, APEX-Aß42 showed significantly higher correlations with amyloid PET retention values and excellent diagnostic performance (sensitivity 100%, specificity 93.3%, area under the curve 0.995). When utilized for clinical trial recruitment, our simulation showed that pre-screening with blood biomarkers followed by a confirmatory amyloid PET imaging would roughly half the cost (56.8% reduction for APEX-Aß42 and 48.6% for Simoa-Aß42/Aß40) compared to the situation where only PET imaging is used. Moreover, with 100% sensitivity, APEX-Aß42 pre-screening does not increase the required number of initial participants. CONCLUSIONS: With its high diagnostic performance, APEX is an ideal candidate for Aß+ subject identification, monitoring and primary care screening, and could efficiently enrich clinical trials with Aß+ participants whilst halving recruitment costs.

Evaluation of Multimodal Algorithms for the Segmentation of Multiparametric MRI Prostate Images.

Prostate segmentation in multiparametric magnetic resonance imaging (mpMRI) can help to support prostate cancer diagnosis and therapy treatment. However, manual segmentation of the prostate is subjective and time-consuming. Many deep learning monomodal networks have been developed for automatic whole prostate segmentation from T2-weighted MR images. We aimed to investigate the added value of multimodal networks in segmenting the prostate into the peripheral zone (PZ) and central gland (CG). We optimized and evaluated monomodal DenseVNet, multimodal ScaleNet, and monomodal and multimodal HighRes3DNet, which yielded dice score coefficients (DSC) of 0.875, 0.848, 0.858, and 0.890 in WG, respectively. Multimodal HighRes3DNet and ScaleNet yielded higher DSC with statistical differences in PZ and CG only compared to monomodal DenseVNet, indicating that multimodal networks added value by generating better segmentation between PZ and CG regions but did not improve the WG segmentation. No significant difference was observed in the apex and base of WG segmentation between monomodal and multimodal networks, indicating that the segmentations at the apex and base were more affected by the general network architecture. The number of training data was also varied for DenseVNet and HighRes3DNet, from 20 to 120 in steps of 20. DenseVNet was able to yield DSC of higher than 0.65 even for special cases, such as TURP or abnormal prostate, whereas HighRes3DNet's performance fluctuated with no trend despite being the best network overall. Multimodal networks did not add value in segmenting special cases but generally reduced variations in segmentation compared to the same matched monomodal network.