Search

Introduction 18F-florbetaben is currently approved for the visual rule out of β-amyloid (Aβ) pathology. It is also used for recruitment and as an outcome measure in therapeutic trials, requiring accurate and reproducible quantification of Aβ burden in the brain. Methods Data from eighty-eight subjects (52 male subjects, aged 79.8 ± 10.6 years) who underwent antemortem 18F-florbetaben positron emission tomography scan and magnetic resonance imaging less than a year before neuropathological assessment at autopsy were evaluated. Image analysis was performed using the standard Centiloid (CL) statistical parametric mapping approach and CapAIBL®. Imaging results were compared against autopsy data. Results Against combined Bielschowsky silver staining and immunohistochemistry histopathological scores, statistical parametric mapping had 96% sensitivity, 96% specificity, and 95% accuracy, whereas magnetic resonance–less CapAIBL standardized uptake value ratioWhole Cerebellum had 94% sensitivity, 96% specificity, and 95% accuracy. Based on the combined histopathological scores, a CL threshold band of 19 ± 7 CL was determined. Discussion Quantification of 18F-florbetaben positron emission tomography scans using magnetic resonance–based and magnetic resonance–less CapAIBL® approaches showed high agreement, establishing a pathology-based threshold in CL.

BACKGROUND AND OBJECTIVES: This prospective study sought to determine the association of modifiable/nonmodifiable components included in the Cardiovascular Risk Factors, Aging, and Incidence of Dementia (CAIDE) risk score with hippocampal volume (HV) loss and episodic memory (EM) decline in cognitively normal (CN) older adults classified as brain β-amyloid (Aβ) negative (Aβ−) or positive (Aβ+). METHODS: Australian Imaging, Biomarkers and Lifestyle study participants (age 58–91 years) who completed ≥2 neuropsychological assessments and a brain Aβ PET scan (n = 592) were included in this study. We computed the CAIDE risk score (age, sex, APOE ε4 status, education, hypertension, body mass index [BMI], hypercholesterolemia, physical inactivity) and a modifiable CAIDE risk score (CAIDE-MR; education, hypertension, BMI, hypercholesterolemia, physical inactivity) for each participant. Aβ+ was classified using Centiloid >25. Linear mixed models assessed interactions between each CAIDE score, Aβ group, and time on HV loss and EM decline. Age, sex, and APOE ε4 were included as separate predictors in CAIDE-MR models to assess differential associations. Exploratory analyses examined relationships between individual modifiable risk factors and outcomes in Aβ− cognitively normal (CN) adults. RESULTS: We observed a significant Aβ group × CAIDE × time interaction on HV loss (β [SE] = –0.04 [0.01]; p < 0.000) but not EM decline (β [SE] = –2.33 [9.96]; p = 0.98). Decomposition revealed a significant CAIDE × time interaction in Aβ+ participants only. When modifiable/nonmodifiable CAIDE components were considered separately, we observed a significant Aβ group × CAIDE-MR × time interaction on EM decline only (β [SE] = 3.03 [1.18]; p = 0.01). A significant CAIDE-MR score × time interaction was observed in Aβ− participants only. Significant interactions between APOE ε4 and age × time on HV loss and EM decline were observed in both groups. Exploratory analyses in Aβ− CN participants revealed a significant interaction between BMI × time on EM decline (β [SE] = –3.30 [1.43]; p = 0.02). DISCUSSION: These results are consistent with studies showing that increasing age and APOE ε4 are associated with increased rates of HV loss and EM decline. In Aβ− CN adults, lower prevalence of modifiable cardiovascular risk factors was associated with less HV loss and EM decline over ∼10 years, suggesting interventions to reduce modifiable cardiovascular risk factors could be beneficial in this group.

BackgroundOngoing research seeks to identify blood-based biomarkers able to predict the onset and progression of Alzheimer’s disease (AD). A potential biomarker is the unfolded conformational variant of p53, previously observed in individuals in the prodromal and clinical AD stages. In this retrospective study, we compare diagnostic and prognostic performances of measures of the amyloid β load with those of a conformational variant of U-p53 in plasma samples from individuals participating in the Australian Imaging, Biomarkers and Lifestyle (AIBL) cohort.MethodsImmunoprecipitation (IP) followed by liquid chromatography (LC) tandem mass spectrometry (MS/MS) and protein sequencing in plasma samples from the AIBL study identified the clinically relevant AZ 284®peptide, representing a measure of the U-p53 conformational variant (U-p53AZ). Based on U-p53AZquantification via IP/LC electrospray ionisation-coupled MS/MS (AlzoSure®Predict test) on 515 samples from 482 individuals from the AIBL cohort, the predictive performance of U-p53AZwas assessed and compared with amyloid load as measured by amyloid β-positron emission tomography (Aβ-PET). Its predictive performance was determined at 36, 54, 72 and 90 months following baseline assessment.ResultsU-p53AZwas able to identify individuals with AD dementia with an area under the receiver operating characteristic curve (AUC) of 99%. U-p53AZoutperformed the conventional Aβ-PET measures in predicting the onset of AD dementia both from preclinical (AUC=98%) and prodromal stages (AUC=89%), even 90 months prior to onset (AUC=99%). Additionally, the estimated predictive performance of U-p53AZwas superior (AUC ≥98%) to other risk factors (i.e., gender, Aβ-PET andAPOEε4 allele status) in identifying individuals at high risk for progression to AD.ConclusionThese findings support use of U-p53AZas blood-based biomarker predicting if individuals, at both asymptomatic and MCI stages, would progress to AD at least six years prior to the onset of clinical AD dementia.

Despite the pervasive growth of deep neural networks in medical image analysis, methods to monitor and assess network outputs, such as segmentation or regression, remain limited. In this paper, we introduce SMOCAM (SMOoth Conditional Attention Mask), an optimization method that reveals the specific regions of the input image taken into account by the prediction of a trained neural network. We developed SMOCAM explicitly to perform saliency analysis for complex regression tasks in 3D medical imagery. Our formulation optimises an 3D-attention mask at a given layer of a convolutional neural network (CNN). Unlike previous attempts, our method is relatively fast (40s per output) and is suitable for large data such as 3D MRI. We applied SMOCAM on a CNN that predicts Brain morphometry from 3D MRI which was trained using more than 5000 3D brain MRIs. We show that SMOCAM highlights neural network’s limitations when cases are underrepresented and in cases with large volume asymmetry.

International audience; Brain morphometry performed with magnetic resonance (MR) imaging is affected by partial volume (PV) effects when single voxels contain the signal from two different tissues. This paper proposes a generalization of the MP2 RAGE sequence PV estimation model which accounts for transmitted magnetic field (B1 {+}) inhomogeneities at 7T. Our simulation experiments demonstrated that the PV estimation error of the proposed model is significantly lower than the error obtained with the same model neglecting B1 {+} inhomogeneities (p

1AbstractMagnetic Resonance Imaging (MRI) motion artefacts frequently complicate structural and diffusion MRI analyses. While diffusion imaging is easily ‘scrubbed’ of motion affected volumes, the same is not true for structural images. Structural images are critical to most diffusion-imaging pipelines thus their corruption can lead to disproportionate data loss. To enable diffusion-image processing when structural images have been corrupted, we propose a means by which synthetic structural images can be generated from diffusion MRI. This technique combines multi-tissue constrained spherical deconvolution, which is central to many existing diffusion analyses, with the Bloch equations which allow simulation of MRI intensities given scanner parameters and magnetic resonance (MR) tissue properties. We applied this technique to 32 scans, including those acquired on different scanners, with different protocols and with pathology present. The resulting synthetic T1w and T2w images were visually convincing and exhibited similar tissue contrast to acquired structural images. These were also of sufficient quality to drive a Freesurfer-based tractographic analysis. In this analysis, probabilistic tractography connecting the thalamus to the primary sensorimotor cortex was delineated with Freesurfer, using either real or synthetic structural images. Tractography for real and synthetic conditions was largely identical in terms of both voxels encountered (Dice 0.88 – 0.95) and mean fractional anisotropy (intrasubject absolute difference 0.00 – 0.02). We provide executables for the proposed technique in the hope that these may aid the community in analysing datasets where structural image corruption is common, such as studies of children or cognitively impaired persons.HighlightsWe propose a simple means of synthesizing T1w and T2w images from diffusion dataThe proposed method worked well for a variety of acquisitionsSynthetic images showed tissue contrast akin to acquired imagesSynthetic images were high enough quality to be used for Freesurfer seeded diffusion tractographyThis method enables analysis of datasets where motion has corrupted acquired structural MRIs

In clinics, the information about the appearance and location of brain tumors is essential to assist doctors in diagnosis and treatment. Automatic brain tumor segmentation on the images acquired by magnetic resonance imaging (MRI) is a common way to attain this information. However, MR images are not quantitative and can exhibit significant variation in signal depending on a range of factors, which increases the difficulty of training an automatic segmentation network and applying it to new MR images. To deal with this issue, this paper proposes to learn a sample-adaptive intensity lookup table (LuT) that dynamically transforms the intensity contrast of each input MR image to adapt to the following segmentation task. Specifically, the proposed deep SA-LuT-Net framework consists of a LuT module and a segmentation module, trained in an end-to-end manner: the LuT module learns a sample-specific nonlinear intensity mapping function through communication with the segmentation module, aiming at improving the final segmentation performance. In order to make the LuT learning sample-adaptive, we parameterize the intensity mapping function by exploring two families of non-linear functions (i.e., piece-wise linear and power functions) and predict the function parameters for each given sample. These sample-specific parameters make the intensity mapping adaptive to samples. We develop our SA-LuT-Nets separately based on two backbone networks for segmentation, i.e., DMFNet and the modified 3D Unet, and validate them on BRATS2018 and BRATS2019 datasets for brain tumor segmentation. Our experimental results clearly demonstrate the superior performance of the proposed SA-LuT-Nets using either single or multiple MR modalities. It not only significantly improves the two baselines (DMFNet and the modified 3D Unet), but also wins a set of state-of-the-art segmentation methods. Moreover, we show that, the LuTs learnt using one segmentation model could also be applied to improving the performance of another segmentation model, indicating the general segmentation information captured by LuTs.

Highlights • Detailed protocol of the Prospective Imaging Study of Ageing (PISA) Study. • Genetic risk prediction to identify those at differing risk of Alzheimer’s disease. • Longitudinal cohort for the study of precursors and lifestyle risk factors. • Use of online surveys and cognitive testing for large scale phenotyping. • Functional, structural and molecular neuroimaging with neurocognitive testing., This prospective cohort study, “Prospective Imaging Study of Ageing: Genes, Brain and Behaviour” (PISA) seeks to characterise the phenotype and natural history of healthy adult Australians at high future risk of Alzheimer’s disease (AD). In particular, we are recruiting midlife and older Australians with high and low genetic risk of dementia to discover biological markers of early neuropathology, identify modifiable risk factors, and establish the very earliest phenotypic and neuronal signs of disease onset. PISA utilises genetic prediction to recruit and enrich a prospective cohort and follow them longitudinally. Online surveys and cognitive testing are used to characterise an Australia-wide sample currently totalling over 3800 participants. Participants from a defined at-risk cohort and positive controls (clinical cohort of patients with mild cognitive impairment or early AD) are invited for onsite visits for detailed functional, structural and molecular neuroimaging, lifestyle monitoring, detailed neurocognitive testing, plus blood sample donation. This paper describes recruitment of the PISA cohort, study methodology and baseline demographics.

Quadrantanopia caused by inadvertent severing of Meyer's Loop of the optic radiation is a well‐recognised complication of temporal lobectomy for conditions such as epilepsy. Dissection studies indicate that the anterior extent of Meyer's Loop varies considerably between individuals. Quantifying this for individual patients is thus an important step to improve the safety profile of temporal lobectomies. Previous attempts to delineate Meyer's Loop using diffusion MRI tractography have had difficulty estimating its full anterior extent, required manual ROI placement, and/or relied on advanced diffusion sequences that cannot be acquired routinely in most clinics. Here we present CONSULT: a pipeline that can delineate the optic radiation from raw DICOM data in a completely automated way via a combination of robust pre‐processing, segmentation, and alignment stages, plus simple improvements that bolster the efficiency and reliability of standard tractography. We tested CONSULT on 696 scans of predominantly healthy participants (539 unique brains), including both advanced acquisitions and simpler acquisitions that could be acquired in clinically acceptable timeframes. Delineations completed without error in 99.4% of the scans. The distance between Meyer's Loop and the temporal pole closely matched both averages and ranges reported in dissection studies for all tested sequences. Median scan‐rescan error of this distance was 1 mm. When tested on two participants with considerable pathology, delineations were successful and realistic. Through this, we demonstrate not only how to identify Meyer's Loop with clinically feasible sequences, but also that this can be achieved without fundamental changes to tractography algorithms or complex post‐processing methods., Quadrantanopia caused by inadvertent severing of Meyer's Loop of the optic radiation is a well‐recognised complication of temporal lobectomy. We demonstrated a fully automated pipeline that could delineate this structure reliably and realistically on more than 500 unique brains, including both advanced and more clinically‐accessible acquisitions. Results were in line with historical dissections studies and median scan‐rescan error was 1 mm.

Objective:To determine the effect of Aβ level on progression risk to MCI or dementia and longitudinal cognitive change in cognitively normal (CN) older individuals.Methods:All CN from the Australian Imaging Biomarkers and Lifestyle study (AIBL) with Aβ PET and ≥3 years follow-up were included (n=534; age 72±6 yrs; 27% Aβ positive; follow-up 5.3±1.7 yrs). Aβ level was divided using the standardised 0-100 Centiloid scale: 100 CL very high, noting >25 CL approximates a positive scan. Cox proportional hazards analysis and linear mixed effect models were used to assess risk of progression and cognitive decline.Results:Aβ levels in 63% were negative, 10% uncertain, 10% moderate, 14% high and 3% very high. Fifty-seven (11%) progressed to MCI or dementia. Compared to negative Aβ, the hazard ratio for progression for moderate Aβ was 3.2 (95% CI 1.3-7.6; pConclusion:The risk of MCI or dementia over 5 years in older CN is related to Aβ level on PET, 5% if negative vs 25% if positive but ranging from 12% if 26-50 CL to 28% if 51-100 CL and 50% if >100 CL. This information may be useful for dementia risk counselling and aid design of preclinical AD trials.

The study of neurodegenerative diseases relies on the reconstruction and analysis of the brain cortex from magnetic resonance imaging (MRI). Traditional frameworks for this task like FreeSurfer demand lengthy runtimes, while its accelerated variant FastSurfer still relies on a voxel-wise segmentation which is limited by its resolution to capture narrow continuous objects as cortical surfaces. Having these limitations in mind, we propose DeepCSR, a 3D deep learning framework for cortical surface reconstruction from MRI. Towards this end, we train a neural network model with hypercolumn features to predict implicit surface representations for points in a brain template space. After training, the cortical surface at a desired level of detail is obtained by evaluating surface representations at specific coordinates, and subsequently applying a topology correction algorithm and an isosurface extraction method. Thanks to the continuous nature of this approach and the efficacy of its hypercolumn features scheme, DeepCSR efficiently reconstructs cortical surfaces at high resolution capturing fine details in the cortical folding. Moreover, DeepCSR is as accurate, more precise, and faster than the widely used FreeSurfer toolbox and its deep learning powered variant FastSurfer on reconstructing cortical surfaces from MRI which should facilitate large-scale medical studies and new healthcare applications., Accepted in 2021 IEEE Winter Conference on Applications of Computer Vision (WACV)

Background Magnetic resonance (MR) T2 and T2* mapping sequences allow in vivo quantification of biochemical characteristics within joint cartilage of relevance to clinical assessment of conditions such as hip osteoarthritis (OA). Purpose To evaluate an automated immediate reliability analysis of T2 and T2* mapping from MR examinations of hip joint cartilage using a bone and cartilage segmentation pipeline based around focused shape modelling. Study type Technical validation. Subjects 17 asymptomatic volunteers (M: F 7:10, aged 22–47 years, mass 50–90 kg, height 163-189 cm) underwent unilateral hip joint MR examinations. Automated analysis of cartilage T2 and T2* data immediate reliability was evaluated in 9 subjects (M: F 4: 5) for each sequence. Field strength/sequence A 3 T MR system with a body matrix flex-coil was used to acquire images with the following sequences: T2 weighted 3D-trueFast Imaging with Steady-State Precession (water excitation; 10.18 ms repetition time (TR); 4.3 ms echo time (TE); Voxel Size (VS): 0.625 × 0.625 × 0.65 mm; 160 mm field of view (FOV); Flip Angle (FA): 30 degrees; Pixel Bandwidth (PB): 140 Hz/pixel); a multi-echo spin echo (MESE) T2 mapping sequence (TR/TE: 2080/18–90 ms (5 echoes); VS: 4 × 0.78 × 0.78 mm; FOV: 200 mm; FA: 180 degrees; PB: 230 Hz/pixel) and a MESE T2* mapping sequence (TR/TE: 873/3.82–19.1 ms (5 echoes); VS: 3 × 0.625 × 0.625 mm; FOV: 160 mm; FA: 25 degrees; PB: 250 Hz/pixel). Assessment Automated cartilage segmentation and quantitative analysis provided T2 and T2* data from test-retest MR examinations to assess immediate reliability. Statistical tests Coefficient of variation (CV) and intraclass correlations (ICC2, 1) to analyse automated T2 and T2* mapping reliability focusing on the clinically important superior cartilage regions of the hip joint. Results Comparisons between test-retest T2 and (T2*) data revealed mean CV's of 3.385% (1.25%), mean ICC2, 1′s of 0.871 (0.984) and median mean differences of −1.139ms (+0.195ms). Conclusion The T2 and T2* times from automated analyses of hip cartilage from test-retest MR examinations had high (T2) and excellent (T2*) immediate reliability.

This study aimed to identify which MRI and clinical assessments, alone or in combination, from (i) early (32 weeks postmenstrual age, PMA), (ii) term equivalent age (TEA) and (iii) 3 months corrected age (CA) are associated with motor or cognitive outcomes at 2 years CA in infants born31 weeks gestation.Prospective cohort study of 98 infants who underwent early and TEA MRI (n = 59 males; median birth gestational age 28 + 5 weeks). Hammersmith Neonatal Neurological Examination (HNNE), NICU Neonatal Neurobehavioural Scale and General Movements Assessment (GMs) were performed early and at TEA. Premie-Neuro was performed early and GMs, Test of Infant Motor Performance and visual assessment were performed at TEA and 3 months CA. Neurodevelopmental outcomes were determined using Bayley Scales of Infant and Toddler Development 3rd edition.The best combined motor outcome model included 3-month GMs (β = -11.41; 95% CI = -17.34, -5.49), TEA MRI deep grey matter score (β = -6.23; 95% CI = -9.47, -2.99) and early HNNE reflexes (β = 3.51; 95% CI = 0.86, 6.16). Combined cognitive model included 3-month GMs (β = -10.01; 95% CI = -15.90, -4.12) and TEA HNNE score (β = 1.33; 95% CI = 0.57, 2.08).Early neonatal neurological assessment improves associations with motor outcomes when combined with term MRI and 3-month GMs. Term neurological assessment combined with 3-month GMs improves associations with cognitive outcomes.We present associations between 32- and 40-week MRI, comprehensive clinical assessments and later 2-year motor and cognitive outcomes for children born31 weeks gestation. MRI and clinical assessment of motor, neurological and neurobehavioural function earlier than term equivalent age in very preterm infants is safe and becoming more available in clinical settings. Most of these children are discharged from hospital before term age and so completing assessments prior to discharge can assist with follow up. MRI and neurological assessment prior to term equivalent age while the child is still in hospital can provide earlier identification of children at highest risk of adverse outcomes and guide follow-up screening and intervention services.

Current pharmacologic therapies for patients with osteoarthritis are suboptimal.To determine the efficacy ofRandomized, double-blind, placebo-controlled trial. (Australian New Zealand Clinical Trials Registry: ACTRN12618000080224).Single-center study with patients from southern Tasmania, Australia.70 participants with symptomatic knee osteoarthritis and ultrasonography-defined effusion-synovitis.2 capsules of CL (The 2 primary outcomes were changes in knee pain on a visual analogue scale (VAS) and effusion-synovitis volume on magnetic resonance imaging (MRI). The key secondary outcomes were change in Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) pain and cartilage composition values. Outcomes were assessed over 12 weeks.CL improved VAS pain compared with placebo by -9.1 mm (95% CI, -17.8 to -0.4 mm [Modest sample size and short duration.CL was more effective than placebo for knee pain but did not affect knee effusion-synovitis or cartilage composition. Multicenter trials with larger sample sizes are needed to assess the clinical significance of these findings.University of Tasmania and Natural Remedies Private Limited.

Introduction Relationships between brain atrophy patterns of typical aging and Alzheimer's disease (AD), white matter disease, cognition, and AD neuropathology were investigated via machine learning in a large harmonized magnetic resonance imaging database (11 studies; 10,216 subjects). Methods Three brain signatures were calculated: Brain-age, AD-like neurodegeneration, and white matter hyperintensities (WMHs). Brain Charts measured and displayed the relationships of these signatures to cognition and molecular biomarkers of AD. Results WMHs were associated with advanced brain aging, AD-like atrophy, poorer cognition, and AD neuropathology in mild cognitive impairment (MCI)/AD and cognitively normal (CN) subjects. High WMH volume was associated with brain aging and cognitive decline occurring in an ≈10-year period in CN subjects. WMHs were associated with doubling the likelihood of amyloid beta (Aβ) positivity after age 65. Brain aging, AD-like atrophy, and WMHs were better predictors of cognition than chronological age in MCI/AD. Discussion A Brain Chart quantifying brain-aging trajectories was established, enabling the systematic evaluation of individuals' brain-aging patterns relative to this large consortium.

Brain morphometry from magnetic resonance imaging (MRI) is a consolidated biomarker for many neurodegenerative diseases. Recent advances in this domain indicate that deep convolutional neural networks can infer morphometric measurements within a few seconds. Nevertheless, the accuracy of the devised model for insightful bio-markers (mean curvature and thickness) remains unsatisfactory. In this paper, we propose a more accurate and efficient neural network model for brain morphometry named HerstonNet. More specifically, we develop a 3D ResNet-based neural network to learn rich features directly from MRI, design a multi-scale regression scheme by predicting morphometric measures at feature maps of different resolutions, and leverage a robust optimization method to avoid poor quality minima and reduce the prediction variance. As a result, HerstonNet improves the existing approach by 24.30% in terms of intraclass correlation coefficient (agreement measure) to FreeSurfer silver-standards while maintaining a competitive run-time.

Finite element analysis (FEA) provides a powerful approach for estimating the in-vivo loading characteristics of the hip joint during various locomotory and functional activities. However, time-consuming procedures, such as the generation of high-quality FE meshes and setup of FE simulation, typically make the method impractical for rapid applications which could be used in clinical routine. Alternatively, discrete element analysis (DEA) has been developed to quantify mechanical conditions of the hip joint in a fraction of time compared to FEA. Although DEA has proven effective in the estimation of contact stresses and areas in various complex applications, it has not yet been well characterised by its ability to evaluate contact mechanics for the hip joint during gait cycle loading using data from several individuals. The objective of this work was to compare DEA modelling against well-established FEA for analysing contact mechanics of the hip joint during walking gait. Subject-specific models were generated from magnetic resonance images of the hip joints in five asymptomatic subjects. The DEA and FEA models were then simulated for 13 loading time-points extracted from a full gait cycle. Computationally, DEA was substantially more efficient compared to FEA (simulation times of seconds vs. hours). The DEA and FEA methods had similar predictions for contact pressure distribution for the hip joint during normal walking. In all 13 simulated loading time-points across five subjects, the maximum difference in average contact pressures between DEA and FEA was within ±0.06 MPa. Furthermore, the difference in contact area ratio computed using DEA and FEA was less than ±6%.

1AbstractQuadrantanopia caused by inadvertent severing of Meyer’s Loop of the optic radiation is a well-recognised complication of temporal lobectomy for conditions such as epilepsy. Dissection studies indicate that the anterior extent of Meyer’s Loop varies considerably between individuals. Quantifying this for individual patients is thus an important step to improve the safety profile of temporal lobectomies. Previous attempts to delineate Meyer’s Loop using diffusion MRI tractography have had difficulty estimating its full anterior extent, required manual ROI placement, and/or relied on advanced diffusion sequences that cannot be acquired routinely in most clinics. Here we present CONSULT – a pipeline that can delineate the optic radiation from raw DICOM data in a completely automated way via a combination of robust preprocessing, segmentation, and alignment stages, plus simple improvements that bolster the efficiency and reliability of standard tractography. We tested CONSULT on 694 scans of predominantly healthy participants (538 unique brains), including both advanced acquisitions and simpler acquisitions that could be acquired in clinically acceptable timeframes. Delineations completed without error in 99.4% of the scans. The distance between Meyer’s Loop and the temporal pole closely matched both averages and ranges reported in dissection studies for all tested sequences. Median scan-rescan error of this distance was 1mm. When tested on two participants with considerable pathology, delineations were successful and realistic. Through this, we demonstrate not only how to identify Meyer’s Loop with clinically accessible sequences, but also that this can be achieved without fundamental changes to tractography algorithms or complex post-processing methods.HighlightsWe propose a fully automated means of delineating the optic radiation using diffusion MRI from DICOM dataAnatomical measurements from tractography of over 500 brains align well with previous dissection studiesThe proposed pipeline works well with clinically accessible and advanced acquisitionsMedian scan-rescan error was 1mmPlausible tractography was generated when pathology was present

Deep learning has emerged as a powerful approach to constructing imaging signatures of normal brain ageing as well as of various neuropathological processes associated with brain diseases. In particular, MRI-derived brain age has been used as a comprehensive biomarker of brain health that can identify both advanced and resilient ageing individuals via deviations from typical brain ageing. Imaging signatures of various brain diseases, including schizophrenia and Alzheimer’s disease, have also been identified using machine learning. Prior efforts to derive these indices have been hampered by the need for sophisticated and not easily reproducible processing steps, by insufficiently powered or diversified samples from which typical brain ageing trajectories were derived, and by limited reproducibility across populations and MRI scanners. Herein, we develop and test a sophisticated deep brain network (DeepBrainNet) using a large (n = 11 729) set of MRI scans from a highly diversified cohort spanning different studies, scanners, ages and geographic locations around the world. Tests using both cross-validation and a separate replication cohort of 2739 individuals indicate that DeepBrainNet obtains robust brain-age estimates from these diverse datasets without the need for specialized image data preparation and processing. Furthermore, we show evidence that moderately fit brain ageing models may provide brain age estimates that are most discriminant of individuals with pathologies. This is not unexpected as tightly-fitting brain age models naturally produce brain-age estimates that offer little information beyond age, and loosely fitting models may contain a lot of noise. Our results offer some experimental evidence against commonly pursued tightly-fitting models. We show that the moderately fitting brain age models obtain significantly higher differentiation compared to tightly-fitting models in two of the four disease groups tested. Critically, we demonstrate that leveraging DeepBrainNet, along with transfer learning, allows us to construct more accurate classifiers of several brain diseases, compared to directly training classifiers on patient versus healthy control datasets or using common imaging databases such as ImageNet. We, therefore, derive a domain-specific deep network likely to reduce the need for application-specific adaptation and tuning of generic deep learning networks. We made the DeepBrainNet model freely available to the community for MRI-based evaluation of brain health in the general population and over the lifespan.

High resolution three-dimensional (3D) magnetic resonance (MR) images are well suited for automated cartilage segmentation in the human knee joint. However, volumetric scans such as 3D Double-Echo Steady-State (DESS) images are not routinely acquired in clinical practice which limits opportunities for reliable cartilage segmentation using (fully) automated algorithms. In this work, a method for generating synthetic 3D MR (syn3D-DESS) images with better contrast and higher spatial resolution from routine, low resolution, two-dimensional (2D) Turbo-Spin Echo (TSE) clinical knee scans is proposed.A UNet convolutional neural network is employed for synthesizing enhanced artificial MR images suitable for automated knee cartilage segmentation. Training of the model was performed on a large, publically available dataset from the OAI, consisting of 578 MR examinations of knee joints from 102 healthy individuals and patients with knee osteoarthritis.The generated synthetic images have higher spatial resolution and better tissue contrast than the original 2D TSE, which allow high quality automated 3D segmentations of the cartilage. The proposed approach was evaluated on a separate set of MR images from 88 subjects with manual cartilage segmentations. It provided a significant improvement in automated segmentation of knee cartilages when using the syn3D-DESS images compared to the original 2D TSE images.The proposed method can successfully synthesize 3D DESS images from 2D TSE images to provide images suitable for automated cartilage segmentation.

This prospective cohort study, “Prospective Imaging Study of Ageing: Genes, Brain and Behaviour” (PISA) seeks to characterise the phenotype and natural history of healthy adult Australians at high future risk of Alzheimer’s disease (AD). In particular, we are recruiting mid-life Australians with high and low genetic risk of dementia to discover biological markers of early neuropathology, identify modifiable risk factors, and establish the very earliest phenotypic and neuronal signs of disease onset. PISA utilises genetic prediction to recruit and enrich a prospective cohort and follow them longitudinally. Online surveys and cognitive testing are used to characterise an Australia-wide sample currently totalling nearly 3,000 participants. Participants from a defined at-risk cohort and positive controls (clinical cohort of patients with mild cognitive impairment or early AD) are invited for onsite visits for lifestyle monitoring, detailed neurocognitive testing, blood sample donation, plus functional, structural and molecular neuroimaging. This paper describes recruitment of the PISA cohort, study methodology and baseline demographics.Author ApprovalAll authors have seen and approved this manuscript.

IntroductionInfants born very preterm are at risk of adverse neurodevelopmental outcomes, including cognitive deficits, motor impairments and cerebral palsy. Earlier identification enables targeted early interventions to be implemented with the aim of improving outcomes.Methods and analysisProtocol for 6-year follow-up of two cohorts of infants born AimsExamine the ability of early neonatal MRI, EEG and concurrent clinical measures at 32 weeks PMA to predict motor, cognitive, language, academic achievement and mental health outcomes at 6 years CA.Determine if early brain abnormalities persist and are evident on brain MRI at 6 years CA and the relationship to EEG and concurrent motor, cognitive, language, academic achievement and mental health outcomes.Ethics and disseminationEthical approval has been obtained from Human Research Ethics Committees at Children’s Health Queensland (HREC/19/QCHQ/49800) and The University of Queensland (2019000426). Study findings will be presented at national and international conferences and published in peer-reviewed journals.Trial registration numberACTRN12619000155190p.Web address of trialhttp://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?ACTRN=12619000155190p

HDR-MRI is a sophisticated non-linear image fusion technique for MRI which enhances image quality by fusing multiple contrast types into a single composite image. It offers improved outcomes in automatic segmentation and potentially in diagnostic power, but the existing technique is slow and requires accurate image co-registration in order to function reliably. In this work, a lightweight fully convolutional neural network architecture is developed with the goal of approximating HDR-MRI images in real-time. The resulting Contrast Enhancement Network (CEN) is capable of performing near-perfect ( $\text{SSIM} =0.98$ ) 2D approximations of HDR-MRI in 10ms and full 3D approximations in 1s, running two orders of magnitude faster than the original implementation. It is also able to perform the approximation ( $\text{SSIM} =0.93$ ) with only two of the three contrasts required to generate the original HDR-MRI image, while requiring no image co-registration.

Background Heme and iron homeostasis is perturbed in Alzheimer’s disease (AD); therefore, the aim of the study was to examine the levels and association of heme with iron-binding plasma proteins in cognitively normal (CN), mild cognitive impairment (MCI), and AD individuals from the Australian Imaging, Biomarker and Lifestyle Flagship Study of Ageing (AIBL) and Kerr Anglican Retirement Village Initiative in Ageing Health (KARVIAH) cohorts. Methods Non-targeted proteomic analysis by high-resolution mass spectrometry was performed to quantify relative protein abundances in plasma samples from 144 CN individuals from the AIBL and 94 CN from KARVIAH cohorts and 21 MCI and 25 AD from AIBL cohort. ANCOVA models were utilized to assess the differences in plasma proteins implicated in heme/iron metabolism, while multiple regression modeling (and partial correlation) was performed to examine the association between heme and iron proteins, structural neuroimaging, and cognitive measures. Results Of the plasma proteins implicated in iron and heme metabolism, hemoglobin subunit β (p = 0.001) was significantly increased in AD compared to CN individuals. Multiple regression modeling adjusted for age, sex, APOEε4 genotype, and disease status in the AIBL cohort revealed lower levels of transferrin but higher levels of hemopexin associated with augmented brain amyloid deposition. Meanwhile, transferrin was positively associated with hippocampal volume and MMSE performance, and hemopexin was negatively associated with CDR scores. Partial correlation analysis revealed lack of significant associations between heme/iron proteins in the CN individuals progressing to cognitive impairment. Conclusions In conclusion, heme and iron dyshomeostasis appears to be a feature of AD. The causal relationship between heme/iron metabolism and AD warrants further investigation.

To develop a local contrast-enhancing and feature-preserving high dynamic range (HDR) image processing algorithm for multichannel and multisequence MR images of multiple body regions and tissues, and to evaluate its performance for structure visualization, bias field (correction) mitigation, and automated tissue segmentation. A multiscale-shape and detail-enhancement HDR-MRI algorithm is applied to data sets of multichannel and multisequence MR images of the brain, knee, breast, and hip. In multisequence 3T hip images, agreement between automatic cartilage segmentations and corresponding synthesized HDR-MRI series were computed for mean voxel overlap established from manual segmentations for a series of cases. Qualitative comparisons between the developed HDR-MRI and standard synthesis methods were performed on multichannel 7T brain and knee data, and multisequence 3T breast and knee data. The synthesized HDR-MRI series provided excellent enhancement of fine-scale structure from multiple scales and contrasts, while substantially reducing bias field effects in 7T brain gradient echo, T1 and T2 breast images and 7T knee multichannel images. Evaluation of the HDR-MRI approach on 3T hip multisequence images showed superior outcomes for automatic cartilage segmentations with respect to manual segmentation, particularly around regions with hyperintense synovial fluid, across a set of 3D sequences. The successful combination of multichannel/sequence MR images into a single-fused HDR-MR image format provided consolidated visualization of tissues within 1 omnibus image, enhanced definition of thin, complex anatomical structures in the presence of variable or hyperintense signals, and improved tissue (cartilage) segmentation outcomes.

This study aimed to examine associations between structural MRI and concurrent motor, neurological and neurobehavioral measures at 30-32 weeks postmenstrual age (PMA; 'Early'), and at term equivalent age ('Term').In this prospective cohort study, infants underwent Early MRI (n = 119; 73 male; median 32 weeks 1 day PMA) and Term MRI (n = 102; 61 male; median 40 weeks 4 days PMA) at 3 T. Structural images were scored generating white matter (WM), cortical gray matter, deep gray matter, cerebellar and global brain abnormality scores. Clinical measures were General Movements Assessment (GMs), Hammersmith Neonatal Neurological Examination (HNNE) and NICU Neonatal Neurobehavioral Scale (NNNS). The Premie-Neuro was administered Early and the Test of Infant Motor Performance (TIMP) and a visual assessment at Term.Early MRI cerebellar scores were strongly associated with neurological components of HNNE (reflexes), NNNS (Hypertonicity), the Premie-Neuro neurological subscale (regression coefficient β = -0.06; 95% confidence interval CI = -0.09, -0.04; p .001) and cramped-synchronized GMs (β = 1.10; 95%CI = 0.57, 1.63; p .001). Term MRI WM and global scores were strongly associated with the TIMP (WM β = -1.02; 95%CI = -1.67, -0.36; p = .002; global β = -1.59; 95%CI = -2.62, -0.56; p = .001).Brain structure on Early and Term MRI was associated with concurrent motor, neurological and neurobehavioral function in very preterm infants.

Preterm birth causes significant disruption in ongoing brain development, frequently resulting in adverse neurodevelopmental outcomes. Brain imaging using diffusion MRI may provide valuable insight into microstructural properties of the developing brain. The aim of this study was to establish whether the recently introduced fixel-based analysis method, with its associated measures of fibre density (FD), fibre bundle cross-section (FC), and fibre density and bundle cross-section (FDC), is suitable for the investigation of the preterm infant brain at term equivalent age. High-angular resolution diffusion weighted images (HARDI) of 55 preterm-born infants and 20 term-born infants, scanned around term-equivalent age, were included in this study (3 T, 64 directions, b = 2000 s/mm2). Postmenstrual age at the time of MRI, and intracranial volume (FC and FDC only), were identified as confounding variables. Gestational age at birth was correlated with all fixel measures in the splenium of the corpus callosum. Compared to term-born infants, preterm infants showed reduced FD, FC, and FDC in a number of regions, including the corpus callosum, anterior commissure, cortico-spinal tract, optic radiations, and cingulum. Preterm infants with minimal macroscopic brain abnormality showed more extensive reductions than preterm infants without any macroscopic brain abnormality; however, little differences were observed between preterm infants with no and with minimal brain abnormality. FC showed significant reductions in preterm versus term infants outside regions identified with FD and FDC, highlighting the complementary role of these measures. Fixel-based analysis identified both microstructural and macrostructural abnormalities in preterm born infants, providing a more complete picture of early brain development than previous diffusion tensor imaging (DTI) based approaches., Highlights • Gestational age at birth associated with measurements in corpus callosum splenium. • Preterms without macroscopic brain abnormality show differences to term infants. • No differences between preterms with minimal versus without abnormality detected.

Background The objective of this study was to investigate the potential of unsegmented structural T1w MR images of adolescent brain for predicting uncorrected/actual fluid intelligence scores without any predefined feature extraction. We also examined whether prediction of uncorrected scores is simply a harder problem from both biological and technical point of view, than prediction of residualised scores. Methods ABCD (Adolescent Brain Cognitive Development) study data was used from 7709 children aged 9–10, including T1-weighted MRIs and fluid intelligence scores, with data split into training (n = 3739), validation (n = 415) and test (n = 3555) subsets. We developed several deep learning convolutional neural network (CNN) models for both actual and residualised fluid intelligence score prediction from the MR images. State of the art, conventional or reverse 2D/3D CNN architectures were developed to perform the regression task and optimised based on Pearson's correlation coefficient, r. The models were then compared with published results on the same dataset. Results Our proposed model achieved prediction accuracies of r = 0.18 (p Conclusion Our deep learning CNN was able to establish a weak but stable correlation between structural brain features and raw fluid intelligence. To improve neuroimaging-based fluid intelligence prediction performance, future studies will be required to explore ensembled regression strategies with multiple machine learning algorithms on multimodal MRIs.

Rationale and Objectives This study aimed to evaluate the accuracy of an automated method for segmentation and T2 mapping of the medial meniscus (MM) and lateral meniscus (LM) in clinical magnetic resonance images from patients with acute knee injury. Materials and Methods Eighty patients scheduled for surgery of an anterior cruciate ligament or meniscal injury underwent magnetic resonance imaging of the knee (multiplanar two-dimensional [2D] turbo spin echo [TSE] or three-dimensional [3D]-TSE examinations, T2 mapping). Each meniscus was automatically segmented from the 2D-TSE (composite volume) or 3D-TSE images, auto-partitioned into anterior, mid, and posterior regions, and co-registered onto the T2 maps. The Dice similarity index (spatial overlap) was calculated between automated and manual segmentations of 2D-TSE (15 patients), 3D-TSE (16 patients), and corresponding T2 maps (31 patients). Pearson and intraclass correlation coefficients (ICC) were calculated between automated and manual T2 values. T2 values were compared (Wilcoxon rank sum tests) between torn and non-torn menisci for the subset of patients with both manual and automated segmentations to compare statistical outcomes of both methods. Results The Dice similarity index values for the 2D-TSE, 3D-TSE, and T2 map volumes, respectively, were 76.4%, 84.3%, and 75.2% for the MM and 76.4%, 85.1%, and 76.1% for the LM. There were strong correlations between automated and manual T2 values (rMM = 0.95, ICCMM = 0.94; rLM = 0.97, ICCLM = 0.97). For both the manual and the automated methods, T2 values were significantly higher in torn than in non-torn MM for the full meniscus and its subregions (P Conclusions The present automated method offers a promising alternative to manual T2 mapping analyses of the menisci and a considerable advance for integration into clinical workflows.

See Derry and Kent (doi:10.1093/awx167) for a scientific commentary on this article.The large variance in cognitive deterioration in subjects who test positive for amyloid-β by positron emission tomography indicates that convergent pathologies, such as iron accumulation, might combine with amyloid-β to accelerate Alzheimer's disease progression. Here, we applied quantitative susceptibility mapping, a relatively new magnetic resonance imaging method sensitive to tissue iron, to assess the relationship between iron, amyloid-β load, and cognitive decline in 117 subjects who underwent baseline magnetic resonance imaging and amyloid-β positron emission tomography from the Australian Imaging, Biomarkers and Lifestyle study (AIBL). Cognitive function data were collected every 18 months for up to 6 years from 100 volunteers who were either cognitively normal (n = 64) or diagnosed with mild cognitive impairment (n = 17) or Alzheimer's disease (n = 19). Among participants with amyloid pathology (n = 45), higher hippocampal quantitative susceptibility mapping levels predicted accelerated deterioration in composite cognition tests for episodic memory [β(standard error) = -0.169 (0.034), P = 9.2 × 10-7], executive function [β(standard error) = -0.139 (0.048), P = 0.004), and attention [β(standard error) = -0.074 (0.029), P = 0.012]. Deteriorating performance in a composite of language tests was predicted by higher quantitative susceptibility mapping levels in temporal lobe [β(standard error) = -0.104 (0.05), P = 0.036] and frontal lobe [β(standard error) = -0.154 (0.055), P = 0.006]. These findings indicate that brain iron might combine with amyloid-β to accelerate clinical progression and that quantitative susceptibility mapping could be used in combination with amyloid-β positron emission tomography to stratify individuals at risk of decline.

Generative adversarial network (GAN) has been widely explored for cross-modality medical image synthesis. The existing GAN models usually adversarially learn a global sample space mapping from the source-modality to the target-modality and then indiscriminately apply this mapping to all samples in the whole space for prediction. However, due to the scarcity of training samples in contrast to the complicated nature of medical image synthesis, learning a single global sample space mapping that is “optimal” to all samples is very challenging, if not intractable. To address this issue, this paper proposes sample-adaptive GAN models, which not only cater for the global sample space mapping between the source- and the target-modalities but also explore the local space around each given sample to extract its unique characteristic. Specifically, the proposed sample-adaptive GANs decompose the entire learning model into two cooperative paths. The baseline path learns a common GAN model by fitting all the training samples as usual for the global sample space mapping. The new sample-adaptive path additionally models each sample by learning its relationship with its neighboring training samples and using the target-modality features of these training samples as auxiliary information for synthesis. Enhanced by this sample-adaptive path, the proposed sample-adaptive GANs are able to flexibly adjust themselves to different samples, and therefore optimize the synthesis performance. Our models have been verified on three cross-modality MR image synthesis tasks from two public datasets, and they significantly outperform the state-of-the-art methods in comparison. Moreover, the experiment also indicates that our sample-adaptive strategy could be utilized to improve various backbone GAN models. It complements the existing GANs models and can be readily integrated when needed.

Estimation of structural biomarkers and covariance networks from MRI have provided valuable insight into the morphological processes and organisation of the human brain. State-of-the-art analyses such as linear mixed effects (LME) models and pairwise descriptive correlation networks are usually performed independently, providing an incomplete picture of the relationships between the biomarkers and network organisation. Furthermore, descriptive network analyses do not generalise to the population level. In this work, we develop a Bayesian generative model based on wombling that allows joint statistical inference on biomarkers and connectivity covariance structure. The parameters of the wombling model were estimated via Markov chain Monte Carlo methods, which allow for simultaneous inference of the brain connectivity matrix and the association of participants’ biomarker covariates. To demonstrate the utility of wombling on real data, the method was used to characterise intrahemispheric cortical thickness and networks in a study cohort of subjects with Alzheimer’s disease (AD), mild-cognitive impairment and healthy ageing. The method was also compared with state-of-the-art alternatives. Our Bayesian modelling approach provided posterior probabilities for the connectivity matrix of the wombling model, accounting for the uncertainty for each connection. This provided superior inference in comparison with descriptive networks. On the study cohort, there was a loss of connectivity across diagnosis levels from healthy to Alzheimer’s disease for all network connections (posterior probability ≥ 0.7). In addition, we found that wombling and LME model approaches estimated that cortical thickness progressively decreased along the dementia pathway. The major advantage of the wombling approach was that spatial covariance among the regions and global cortical thickness estimates could be estimated. Joint modelling of biomarkers and covariance networks using our novel wombling approach allowed accurate identification of probabilistic networks and estimated biomarker changes that took into account spatial covariance. The wombling model provides a novel tool to address multiple brain features, such as morphological and connectivity changes facilitating a better understanding of disease pathology.