Lazy Resampling: Fast and information preserving preprocessing for deep learning.

BACKGROUND AND OBJECTIVE: Preprocessing of data is a vital step for almost all deep learning workflows. In computer vision, manipulation of data intensity and spatial properties can improve network stability and can provide an important source of generalisation for deep neural networks. Models are frequently trained with preprocessing pipelines composed of many stages, but these pipelines come with a drawback; each stage that resamples the data costs time, degrades image quality, and adds bias to the output. Long pipelines can also be complex to design, especially in medical imaging, where cropping data early can cause significant artifacts. METHODS: We present Lazy Resampling, a software that rephrases spatial preprocessing operations as a graphics pipeline. Rather than each transform individually modifying the data, the transforms generate transform descriptions that are composited together into a single resample operation wherever possible. This reduces pipeline execution time and, most importantly, limits signal degradation. It enables simpler pipeline design as crops and other operations become non-destructive. Lazy Resampling is designed in such a way that it provides the maximum benefit to users without requiring them to understand the underlying concepts or change the way that they build pipelines. RESULTS: We evaluate Lazy Resampling by comparing traditional pipelines and the corresponding lazy resampling pipeline for the following tasks on Medical Segmentation Decathlon datasets. We demonstrate lower information loss in lazy pipelines vs. traditional pipelines. We demonstrate that Lazy Resampling can avoid catastrophic loss of semantic segmentation label accuracy occurring in traditional pipelines when passing labels through a pipeline and then back through the inverted pipeline. Finally, we demonstrate statistically significant improvements when training UNets for semantic segmentation. CONCLUSION: Lazy Resampling reduces the loss of information that occurs when running processing pipelines that traditionally have multiple resampling steps and enables researchers to build simpler pipelines by making operations such as rotation and cropping effectively non-destructive. It makes it possible to invert labels back through a pipeline without catastrophic loss of accuracy. A reference implementation for Lazy Resampling can be found at https://github.com/KCL-BMEIS/LazyResampling. Lazy Resampling is being implemented as a core feature in MONAI, an open source python-based deep learning library for medical imaging, with a roadmap for a full integration.

Predicting 1, 2 and 3 year emergent referable diabetic retinopathy and maculopathy using deep learning.

BACKGROUND: Predicting diabetic retinopathy (DR) progression could enable individualised screening with prompt referral for high-risk individuals for sight-saving treatment, whilst reducing screening burden for low-risk individuals. We developed and validated deep learning systems (DLS) that predict 1, 2 and 3 year emergent referable DR and maculopathy using risk factor characteristics (tabular DLS), colour fundal photographs (image DLS) or both (multimodal DLS). METHODS: From 162,339 development-set eyes from south-east London (UK) diabetic eye screening programme (DESP), 110,837 had eligible longitudinal data, with the remaining 51,502 used for pretraining. Internal and external (Birmingham DESP, UK) test datasets included 27,996, and 6928 eyes respectively. RESULTS: Internal multimodal DLS emergent referable DR, maculopathy or either area-under-the receiver operating characteristic (AUROC) were 0.95 (95% CI: 0.92-0.98), 0.84 (0.82-0.86), 0.85 (0.83-0.87) for 1 year, 0.92 (0.87-0.96), 0.84 (0.82-0.87), 0.85 (0.82-0.87) for 2 years, and 0.85 (0.80-0.90), 0.79 (0.76-0.82), 0.79 (0.76-0.82) for 3 years. External multimodal DLS emergent referable DR, maculopathy or either AUROC were 0.93 (0.88-0.97), 0.85 (0.80-0.89), 0.85 (0.76-0.85) for 1 year, 0.93 (0.89-0.97), 0.79 (0.74-0.84), 0.80 (0.76-0.85) for 2 years, and 0.91 (0.84-0.98), 0.79 (0.74-0.83), 0.79 (0.74-0.84) for 3 years. CONCLUSIONS: Multimodal and image DLS performance is significantly better than tabular DLS at all intervals. DLS accurately predict 1, 2 and 3 year emergent referable DR and referable maculopathy using colour fundal photographs, with additional risk factor characteristics conferring improvements in prognostic performance. Proposed DLS are a step towards individualised risk-based screening, whereby AI-assistance allows high-risk individuals to be closely monitored while reducing screening burden for low-risk individuals.

Diabetic retinopathy (DR) is a disease where the light-sensing layer at the back of the eye (retina) becomes damaged by raised blood sugar levels. It affects around one in three of the 463 million people with diabetes worldwide and is a leading cause of acquired vision loss in working-age adults. In this study, we developed computer-based models to predict when DR would reach a stage where vision could be threatened up to 3-years in the future. Our study shows that this system can accurately predict sight-threatening DR in patients with diabetes. This could mean fewer unnecessary visits for individuals at low-risk of DR progression, but closer monitoring and potentially earlier treatment for individuals at high-risk of DR progression, which could reduce the risk of vision loss.

Computer-aided prognosis of tuberculous meningitis combining imaging and non-imaging data.

Tuberculous meningitis (TBM) is the most lethal form of tuberculosis. Clinical features, such as coma, can predict death, but they are insufficient for the accurate prognosis of other outcomes, especially when impacted by co-morbidities such as HIV infection. Brain magnetic resonance imaging (MRI) characterises the extent and severity of disease and may enable more accurate prediction of complications and poor outcomes. We analysed clinical and brain MRI data from a prospective longitudinal study of 216 adults with TBM; 73 (34%) were HIV-positive, a factor highly correlated with mortality. We implemented an end-to-end framework to model clinical and imaging features to predict disease progression. Our model used state-of-the-art machine learning models for automatic imaging feature encoding, and time-series models for forecasting, to predict TBM progression. The proposed approach is designed to be robust to missing data via a novel tailored model optimisation framework. Our model achieved a 60% balanced accuracy in predicting the prognosis of TBM patients over the six different classes. HIV status did not alter the performance of the models. Furthermore, our approach identified brain morphological lesions caused by TBM in both HIV and non-HIV-infected, associating lesions to the disease staging with an overall accuracy of 96%. These results suggest that the lesions caused by TBM are analogous in both populations, regardless of the severity of the disease. Lastly, our models correctly identified changes in disease symptomatology and severity in 80% of the cases. Our approach is the first attempt at predicting the prognosis of TBM by combining imaging and clinical data, via a machine learning model. The approach has the potential to accurately predict disease progression and enable timely clinical intervention.

Symptoms before and after COVID-19: a population and case-control study using prospective data.

BACKGROUND: Some individuals experience prolonged illness after acute coronavirus disease 2019 (COVID-19). We assessed whether pre-infection symptoms affected post-acute COVID illness duration. METHODS: Survival analysis was performed in adults (n=23 452) with community-managed severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection prospectively self-logging data through the ZOE COVID Symptom Study app, at least weekly, from 8 weeks before to 12 weeks after COVID-19 onset, conditioned on presence versus absence of baseline symptoms (4-8 weeks before COVID-19). A case-control study was performed in 1350 individuals with long illness (≥8 weeks, including 906 individuals (67.1%) with illness ≥12 weeks), matched 1:1 (for age, sex, body mass index, testing week, prior infection, vaccination, smoking, index of multiple deprivation) with 1350 individuals with short illness (<4 weeks). Baseline symptoms were compared between the two groups, and against post-COVID symptoms. RESULTS: Individuals reporting baseline symptoms had longer COVID-related symptom duration (median 15 days versus 10 days for individuals without baseline symptoms) with baseline fatigue nearly doubling duration. Two-thirds (910 (67.4%) of 1350) of individuals with long illness were asymptomatic beforehand. However, 440 (32.6%) had baseline symptoms, versus 255 (18.9%) of 1350 individuals with short illness (p<0.0001). Baseline symptoms doubled the odds ratio for long illness (2.14, 95% CI 1.78-2.57). Prior comorbidities were more common in individuals with long versus short illness. In individuals with long illness, baseline symptomatic (versus asymptomatic) individuals were more likely to be female, younger, and have prior comorbidities; and baseline and post-acute symptoms, and symptom burden, correlated strongly. CONCLUSIONS: Individuals experiencing symptoms before COVID-19 had longer illness duration and increased odds of long illness. However, many individuals with long illness were well before SARS-CoV-2 infection.

Radiogenomic biomarkers for immunotherapy in glioblastoma: A systematic review of magnetic resonance imaging studies.

Background: Immunotherapy is an effective "precision medicine" treatment for several cancers. Imaging signatures of the underlying genome (radiogenomics) in glioblastoma patients may serve as preoperative biomarkers of the tumor-host immune apparatus. Validated biomarkers would have the potential to stratify patients during immunotherapy clinical trials, and if trials are beneficial, facilitate personalized neo-adjuvant treatment. The increased use of whole genome sequencing data, and the advances in bioinformatics and machine learning make such developments plausible. We performed a systematic review to determine the extent of development and validation of immune-related radiogenomic biomarkers for glioblastoma. Methods: A systematic review was performed following PRISMA guidelines using the PubMed, Medline, and Embase databases. Qualitative analysis was performed by incorporating the QUADAS 2 tool and CLAIM checklist. PROSPERO registered: CRD42022340968. Extracted data were insufficiently homogenous to perform a meta-analysis. Results: Nine studies, all retrospective, were included. Biomarkers extracted from magnetic resonance imaging volumes of interest included apparent diffusion coefficient values, relative cerebral blood volume values, and image-derived features. These biomarkers correlated with genomic markers from tumor cells or immune cells or with patient survival. The majority of studies had a high risk of bias and applicability concerns regarding the index test performed. Conclusions: Radiogenomic immune biomarkers have the potential to provide early treatment options to patients with glioblastoma. Targeted immunotherapy, stratified by these biomarkers, has the potential to allow individualized neo-adjuvant precision treatment options in clinical trials. However, there are no prospective studies validating these biomarkers, and interpretation is limited due to study bias with little evidence of generalizability.

Glioblastoma and radiotherapy: A multicenter AI study for Survival Predictions from MRI (GRASP study).

BACKGROUND: The aim was to predict survival of glioblastoma at 8 months after radiotherapy (a period allowing for completing a typical course of adjuvant temozolomide), by applying deep learning to the first brain MRI after radiotherapy completion. METHODS: Retrospective and prospective data were collected from 206 consecutive glioblastoma, isocitrate dehydrogenase -wildtype patients diagnosed between March 2014 and February 2022 across 11 UK centers. Models were trained on 158 retrospective patients from 3 centers. Holdout test sets were retrospective (n = 19; internal validation), and prospective (n = 29; external validation from 8 distinct centers). Neural network branches for T2-weighted and contrast-enhanced T1-weighted inputs were concatenated to predict survival. A nonimaging branch (demographics/MGMT/treatment data) was also combined with the imaging model. We investigated the influence of individual MR sequences; nonimaging features; and weighted dense blocks pretrained for abnormality detection. RESULTS: The imaging model outperformed the nonimaging model in all test sets (area under the receiver-operating characteristic curve, AUC P = .038) and performed similarly to a combined imaging/nonimaging model (P > .05). Imaging, nonimaging, and combined models applied to amalgamated test sets gave AUCs of 0.93, 0.79, and 0.91. Initializing the imaging model with pretrained weights from 10 000s of brain MRIs improved performance considerably (amalgamated test sets without pretraining 0.64; P = .003). CONCLUSIONS: A deep learning model using MRI images after radiotherapy reliably and accurately determined survival of glioblastoma. The model serves as a prognostic biomarker identifying patients who will not survive beyond a typical course of adjuvant temozolomide, thereby stratifying patients into those who might require early second-line or clinical trial treatment.

SARS-CoV-2 infection following booster vaccination: Illness and symptom profile in a prospective, observational community-based case-control study.

BACKGROUND: Booster COVID-19 vaccines have shown efficacy in clinical trials and effectiveness in real-world data against symptomatic and severe illness. However, some people still become infected with SARS-CoV-2 following a third (booster) vaccination. This study describes the characteristics of SARS-CoV-2 illness following a third vaccination and assesses the risk of progression to symptomatic disease in SARS-CoV-2 infected individuals with time since vaccination. METHODS: This prospective, community-based, case-control study used data from UK-based, adult (≥18 years) users of the COVID Symptom Study mobile application, self-reporting a first positive COVID-19 test between June 1, 2021 and April 1, 2022. To describe the characteristics of SARS-CoV-2 illness following a third vaccination, we selected cases and controls who had received a third and second dose of monovalent vaccination against COVID-19, respectively, and reported a first positive SARS-CoV-2 test at least 7 days after most recent vaccination. Cases and controls were matched (1:1) based on age, sex, BMI, time between first vaccination and infection, and week of testing. We used logistic regression models (adjusted for age, sex, BMI, level of social deprivation and frailty) to analyse associations of disease severity, overall disease duration, and individual symptoms with booster vaccination status. To assess for potential waning of vaccine effectiveness, we compared disease severity, duration, and symptom profiles of individuals testing positive within 3 months of most recent vaccination (reference group) to profiles of individuals infected between 3 and 4, 4-5, and 5-6 months, for both third and second dose. All analyses were stratified by time period, based on the predominant SARS-CoV-2 variant at time of infection (Delta: June 1, 2021-27 Nov, 2021; Omicron: 20 Dec, 2021-Apr 1, 2022). FINDINGS: During the study period, 50,162 (Delta period) and 162,041 (Omicron) participants reported a positive SARS-CoV-2 test. During the Delta period, infection following three vaccination doses was associated with lower odds of long COVID (symptoms≥ 4 weeks) (OR=0.83, CI[0.50-1.36], p < 0.0001), hospitalisation (OR=0.55, CI[0.39-0.75], p < 0.0001) and severe symptoms (OR=0.36, CI[0.27-0.49], p < 0.0001), and higher odds of asymptomatic infection (OR=3.45, CI[2.86-4.16], p < 0.0001), compared to infection following only two vaccination doses. During the Omicron period, infection following three vaccination doses was associated with lower odds of severe symptoms (OR=0.48, CI[0.42-0.55], p < 0.0001). During the Delta period, infected individuals were less likely to report almost all individual symptoms after a third vaccination. During the Omicron period, individuals were less likely to report most symptoms after a third vaccination, except for upper respiratory symptoms e.g. sneezing (OR=1.40, CI[1.18-1.35], p < 0.0001), runny nose (OR=1.26, CI[1.18-1.35], p < 0.0001), sore throat (OR=1.17, CI[1.10-1.25], p < 0.0001), and hoarse voice (OR=1.13, CI[1.06-1.21], p < 0.0001), which were more likely to be reported. There was evidence of reduced vaccine effectiveness during both Delta and Omicron periods in those infected more than 3 months after their most recent vaccination, with increased reporting of severe symptoms, long duration illness, and most individual symptoms. INTERPRETATION: This study suggests that a third dose of monovalent vaccine may reduce symptoms, severity and duration of SARS-CoV-2 infection following vaccination. For Omicron variants, the third vaccination appears to reduce overall symptom burden but may increase upper respiratory symptoms, potentially due to immunological priming. There is evidence of waning vaccine effectiveness against progression to symptomatic and severe disease and long COVID after three months. Our findings support ongoing booster vaccination promotion amongst individuals at high risk from COVID-19, to reduce severe symptoms and duration of illness, and health system burden. Disseminating knowledge on expected symptoms following booster vaccination may encourage vaccine uptake.

The effects of COVID-19 on cognitive performance in a community-based cohort: a COVID symptom study biobank prospective cohort study.

Background: Cognitive impairment has been reported after many types of infection, including SARS-CoV-2. Whether deficits following SARS-CoV-2 improve over time is unclear. Studies to date have focused on hospitalised individuals with up to a year follow-up. The presence, magnitude, persistence and correlations of effects in community-based cases remain relatively unexplored. Methods: Cognitive performance (working memory, attention, reasoning, motor control) was assessed in a prospective cohort study of participants from the United Kingdom COVID Symptom Study Biobank between July 12, 2021 and August 27, 2021 (Round 1), and between April 28, 2022 and June 21, 2022 (Round 2). Participants, recruited from the COVID Symptom Study smartphone app, comprised individuals with and without SARS-CoV-2 infection and varying symptom duration. Effects of COVID-19 exposures on cognitive accuracy and reaction time scores were estimated using multivariable ordinary least squares linear regression models weighted for inverse probability of participation, adjusting for potential confounders and mediators. The role of ongoing symptoms after COVID-19 infection was examined stratifying for self-perceived recovery. Longitudinal analysis assessed change in cognitive performance between rounds. Findings: 3335 individuals completed Round 1, of whom 1768 also completed Round 2. At Round 1, individuals with previous positive SARS-CoV-2 tests had lower cognitive accuracy (N = 1737, ß = -0.14 standard deviations, SDs, 95% confidence intervals, CI: -0.21, -0.07) than negative controls. Deficits were largest for positive individuals with ≥12 weeks of symptoms (N = 495, ß = -0.22 SDs, 95% CI: -0.35, -0.09). Effects were comparable to hospital presentation during illness (N = 281, ß = -0.31 SDs, 95% CI: -0.44, -0.18), and 10 years age difference (60-70 years vs. 50-60 years, ß = -0.21 SDs, 95% CI: -0.30, -0.13) in the whole study population. Stratification by self-reported recovery revealed that deficits were only detectable in SARS-CoV-2 positive individuals who did not feel recovered from COVID-19, whereas individuals who reported full recovery showed no deficits. Longitudinal analysis showed no evidence of cognitive change over time, suggesting that cognitive deficits for affected individuals persisted at almost 2 years since initial infection. Interpretation: Cognitive deficits following SARS-CoV-2 infection were detectable nearly two years post infection, and largest for individuals with longer symptom durations, ongoing symptoms, and/or more severe infection. However, no such deficits were detected in individuals who reported full recovery from COVID-19. Further work is needed to monitor and develop understanding of recovery mechanisms for those with ongoing symptoms. Funding: Chronic Disease Research Foundation, Wellcome Trust, National Institute for Health and Care Research, Medical Research Council, British Heart Foundation, Alzheimer's Society, European Union, COVID-19 Driver Relief Fund, French National Research Agency.

Systematic Review of Artificial Intelligence for Abnormality Detection in High-volume Neuroimaging and Subgroup Meta-analysis for Intracranial Hemorrhage Detection.

PURPOSE: Most studies evaluating artificial intelligence (AI) models that detect abnormalities in neuroimaging are either tested on unrepresentative patient cohorts or are insufficiently well-validated, leading to poor generalisability to real-world tasks. The aim was to determine the diagnostic test accuracy and summarise the evidence supporting the use of AI models performing first-line, high-volume neuroimaging tasks. METHODS: Medline, Embase, Cochrane library and Web of Science were searched until September 2021 for studies that temporally or externally validated AI capable of detecting abnormalities in first-line computed tomography (CT) or magnetic resonance (MR) neuroimaging. A bivariate random effects model was used for meta-analysis where appropriate. This study was registered on PROSPERO as CRD42021269563. RESULTS: Out of 42,870 records screened, and 5734 potentially eligible full texts, only 16 studies were eligible for inclusion. Included studies were not compromised by unrepresentative datasets or inadequate validation methodology. Direct comparison with radiologists was available in 4/16 studies and 15/16 had a high risk of bias. Meta-analysis was only suitable for intracranial hemorrhage detection in CT imaging (10/16 studies), where AI systems had a pooled sensitivity and specificity 0.90 (95% confidence interval [CI] 0.85-0.94) and 0.90 (95% CI 0.83-0.95), respectively. Other AI studies using CT and MRI detected target conditions other than hemorrhage (2/16), or multiple target conditions (4/16). Only 3/16 studies implemented AI in clinical pathways, either for pre-read triage or as post-read discrepancy identifiers. CONCLUSION: The paucity of eligible studies reflects that most abnormality detection AI studies were not adequately validated in representative clinical cohorts. The few studies describing how abnormality detection AI could impact patients and clinicians did not explore the full ramifications of clinical implementation.

Effect of socioeconomic deprivation as determined by the English deprivation deciles on the progression of diabetic retinopathy and maculopathy: a multivariate case-control analysis of 88 910 patients attending a South-East London diabetic eye screening service.

PURPOSE: To determine associations between deprivation using the Index of Multiple Deprivation (IMD and individual IMD subdomains) with incident referable diabetic retinopathy/maculopathy (termed rDR). METHODS: Anonymised demographic and screening data collected by the South-East London Diabetic Eye Screening Programme were extracted from September 2013 to December 2019. Multivariable Cox proportional models were used to explore the association between the IMD, IMD subdomains and rDR. RESULTS: From 118 508 people with diabetes who attended during the study period, 88 910 (75%) were eligible. The mean (± SD) age was 59.6 (±14.7) years; 53.94% were male, 52.58% identified as white, 94.28% had type 2 diabetes and the average duration of diabetes was 5.81 (±6.9) years; rDR occurred in 7113 patients (8.00%). Known risk factors of younger age, Black ethnicity, type 2 diabetes, more severe baseline DR and diabetes duration conferred a higher risk of incident rDR. After adjusting for these known risk factors, the multivariable analysis did not show a significant association between IMD (decile 1 vs decile 10) and rDR (HR: 1.08, 95% CI: 0.87 to 1.34, p=0.511). However, high deprivation (decile 1) in three IMD subdomains was associated with rDR, namely living environment (HR: 1.64, 95% CI: 1.12 to 2.41, p=0.011), education skills (HR: 1.64, 95% CI: 1.12 to 2.41, p=0.011) and income (HR: 1.19, 95% CI: 1.02 to 1.38, p=0.024). CONCLUSION: IMD subdomains allow for the detection of associations between aspects of deprivation and rDR, which may be missed when using the aggregate IMD. The generalisation of these findings outside the UK population requires corroboration internationally.

Operationalizing the centiloid scale for [18F]florbetapir PET studies on PET/MRI.

INTRODUCTION: The Centiloid scale aims to harmonize amyloid beta (Aß) positron emission tomography (PET) measures across different analysis methods. As Centiloids were created using PET/computerized tomography (CT) data and are influenced by scanner differences, we investigated the Centiloid transformation with data from Insight 46 acquired with PET/magnetic resonanceimaging (MRI). METHODS: We transformed standardized uptake value ratios (SUVRs) from 432 florbetapir PET/MRI scans processed using whole cerebellum (WC) and white matter (WM) references, with and without partial volume correction. Gaussian-mixture-modelling-derived cutpoints for Aß PET positivity were converted. RESULTS: The Centiloid cutpoint was 14.2 for WC SUVRs. The relationship between WM and WC uptake differed between the calibration and testing datasets, producing implausibly low WM-based Centiloids. Linear adjustment produced a WM-based cutpoint of 18.1. DISCUSSION: Transformation of PET/MRI florbetapir data to Centiloids is valid. However, further understanding of the effects of acquisition or biological factors on the transformation using a WM reference is needed. HIGHLIGHTS: Centiloid conversion of amyloid beta positron emission tomography (PET) data aims to standardize results.Centiloid values can be influenced by differences in acquisition.We converted florbetapir PET/magnetic resonance imaging data from a large birth cohort.Whole cerebellum referenced values could be reliably transformed to Centiloids.White matter referenced values may be less generalizable between datasets.

Profiling post-COVID-19 condition across different variants of SARS-CoV-2: a prospective longitudinal study in unvaccinated wild-type, unvaccinated alpha-variant, and vaccinated delta-variant populations.

BACKGROUND: Self-reported symptom studies rapidly increased understanding of SARS-CoV-2 during the COVID-19 pandemic and enabled monitoring of long-term effects of COVID-19 outside hospital settings. Post-COVID-19 condition presents as heterogeneous profiles, which need characterisation to enable personalised patient care. We aimed to describe post-COVID-19 condition profiles by viral variant and vaccination status. METHODS: In this prospective longitudinal cohort study, we analysed data from UK-based adults (aged 18-100 years) who regularly provided health reports via the Covid Symptom Study smartphone app between March 24, 2020, and Dec 8, 2021. We included participants who reported feeling physically normal for at least 30 days before testing positive for SARS-CoV-2 who subsequently developed long COVID (ie, symptoms lasting longer than 28 days from the date of the initial positive test). We separately defined post-COVID-19 condition as symptoms that persisted for at least 84 days after the initial positive test. We did unsupervised clustering analysis of time-series data to identify distinct symptom profiles for vaccinated and unvaccinated people with post-COVID-19 condition after infection with the wild-type, alpha (B.1.1.7), or delta (B.1.617.2 and AY.x) variants of SARS-CoV-2. Clusters were then characterised on the basis of symptom prevalence, duration, demography, and previous comorbidities. We also used an additional testing sample with additional data from the Covid Symptom Study Biobank (collected between October, 2020, and April, 2021) to investigate the effects of the identified symptom clusters of post-COVID-19 condition on the lives of affected people. FINDINGS: We included 9804 people from the COVID Symptom Study with long COVID, 1513 (15%) of whom developed post-COVID-19 condition. Sample sizes were sufficient only for analyses of the unvaccinated wild-type, unvaccinated alpha variant, and vaccinated delta variant groups. We identified distinct profiles of symptoms for post-COVID-19 condition within and across variants: four endotypes were identified for infections due to the wild-type variant (in unvaccinated people), seven for the alpha variant (in unvaccinated people), and five for the delta variant (in vaccinated people). Across all variants, we identified a cardiorespiratory cluster of symptoms, a central neurological cluster, and a multi-organ systemic inflammatory cluster. These three main clusers were confirmed in a testing sample. Gastrointestinal symptoms clustered in no more than two specific phenotypes per viral variant. INTERPRETATION: Our unsupervised analysis identified different profiles of post-COVID-19 condition, characterised by differing symptom combinations, durations, and functional outcomes. Our classification could be useful for understanding the distinct mechanisms of post-COVID-19 condition, as well as for identification of subgroups of individuals who might be at risk of prolonged debilitation. FUNDING: UK Government Department of Health and Social Care, Chronic Disease Research Foundation, The Wellcome Trust, UK Engineering and Physical Sciences Research Council, UK Research and Innovation London Medical Imaging & Artificial Intelligence Centre for Value-Based Healthcare, UK National Institute for Health Research, UK Medical Research Council, British Heart Foundation, UK Alzheimer's Society, and ZOE.

Learn2Reg: Comprehensive Multi-Task Medical Image Registration Challenge, Dataset and Evaluation in the Era of Deep Learning.

Image registration is a fundamental medical image analysis task, and a wide variety of approaches have been proposed. However, only a few studies have comprehensively compared medical image registration approaches on a wide range of clinically relevant tasks. This limits the development of registration methods, the adoption of research advances into practice, and a fair benchmark across competing approaches. The Learn2Reg challenge addresses these limitations by providing a multi-task medical image registration data set for comprehensive characterisation of deformable registration algorithms. A continuous evaluation will be possible at https://learn2reg.grand-challenge.org. Learn2Reg covers a wide range of anatomies (brain, abdomen, and thorax), modalities (ultrasound, CT, MR), availability of annotations, as well as intra- and inter-patient registration evaluation. We established an easily accessible framework for training and validation of 3D registration methods, which enabled the compilation of results of over 65 individual method submissions from more than 20 unique teams. We used a complementary set of metrics, including robustness, accuracy, plausibility, and runtime, enabling unique insight into the current state-of-the-art of medical image registration. This paper describes datasets, tasks, evaluation methods and results of the challenge, as well as results of further analysis of transferability to new datasets, the importance of label supervision, and resulting bias. While no single approach worked best across all tasks, many methodological aspects could be identified that push the performance of medical image registration to new state-of-the-art performance. Furthermore, we demystified the common belief that conventional registration methods have to be much slower than deep-learning-based methods.

Detection of cerebral aneurysms using artificial intelligence: a systematic review and meta-analysis.

BACKGROUND: Subarachnoid hemorrhage from cerebral aneurysm rupture is a major cause of morbidity and mortality. Early aneurysm identification, aided by automated systems, may improve patient outcomes. Therefore, a systematic review and meta-analysis of the diagnostic accuracy of artificial intelligence (AI) algorithms in detecting cerebral aneurysms using CT, MRI or DSA was performed. METHODS: MEDLINE, Embase, Cochrane Library and Web of Science were searched until August 2021. Eligibility criteria included studies using fully automated algorithms to detect cerebral aneurysms using MRI, CT or DSA. Following Preferred Reporting Items for Systematic Reviews and Meta-Analysis: Diagnostic Test Accuracy (PRISMA-DTA), articles were assessed using Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2). Meta-analysis included a bivariate random-effect model to determine pooled sensitivity, specificity, and area under the receiver operator characteristic curve (ROC-AUC). PROSPERO: CRD42021278454. RESULTS: 43 studies were included, and 41/43 (95%) were retrospective. 34/43 (79%) used AI as a standalone tool, while 9/43 (21%) used AI assisting a reader. 23/43 (53%) used deep learning. Most studies had high bias risk and applicability concerns, limiting conclusions. Six studies in the standalone AI meta-analysis gave (pooled) 91.2% (95% CI 82.2% to 95.8%) sensitivity; 16.5% (95% CI 9.4% to 27.1%) false-positive rate (1-specificity); 0.936 ROC-AUC. Five reader-assistive AI studies gave (pooled) 90.3% (95% CI 88.0% - 92.2%) sensitivity; 7.9% (95% CI 3.5% to 16.8%) false-positive rate; 0.910 ROC-AUC. CONCLUSION: AI has the potential to support clinicians in detecting cerebral aneurysms. Interpretation is limited due to high risk of bias and poor generalizability. Multicenter, prospective studies are required to assess AI in clinical practice.

Adulthood cognitive trajectories over 26 years and brain health at 70 years of age: findings from the 1946 British Birth Cohort.

Few studies can address how adulthood cognitive trajectories relate to brain health in 70-year-olds. Participants (n = 468, 49% female) from the 1946 British birth cohort underwent 18F-Florbetapir PET/MRI. Cognitive function was measured in childhood (age 8 years) and across adulthood (ages 43, 53, 60-64 and 69 years) and was examined in relation to brain health markers of ß-amyloid (Aß) status, whole brain and hippocampal volume, and white matter hyperintensity volume (WMHV). Taking into account key contributors of adult cognitive decline including childhood cognition, those with greater Aß and WMHV at age 70 years had greater decline in word-list learning memory in the preceding 26 years, particularly after age 60. In contrast, those with smaller whole brain and hippocampal volume at age 70 years had greater decline in processing search speed, subtly manifest from age 50 years. Subtle changes in memory and processing speed spanning 26 years of adulthood were associated with markers of brain health at 70 years of age, consistent with detectable prodromal cognitive effects in early older age.

CrossMoDA 2021 challenge: Benchmark of cross-modality domain adaptation techniques for vestibular schwannoma and cochlea segmentation.

Domain Adaptation (DA) has recently been of strong interest in the medical imaging community. While a large variety of DA techniques have been proposed for image segmentation, most of these techniques have been validated either on private datasets or on small publicly available datasets. Moreover, these datasets mostly addressed single-class problems. To tackle these limitations, the Cross-Modality Domain Adaptation (crossMoDA) challenge was organised in conjunction with the 24th International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI 2021). CrossMoDA is the first large and multi-class benchmark for unsupervised cross-modality Domain Adaptation. The goal of the challenge is to segment two key brain structures involved in the follow-up and treatment planning of vestibular schwannoma (VS): the VS and the cochleas. Currently, the diagnosis and surveillance in patients with VS are commonly performed using contrast-enhanced T1 (ceT1) MR imaging. However, there is growing interest in using non-contrast imaging sequences such as high-resolution T2 (hrT2) imaging. For this reason, we established an unsupervised cross-modality segmentation benchmark. The training dataset provides annotated ceT1 scans (N=105) and unpaired non-annotated hrT2 scans (N=105). The aim was to automatically perform unilateral VS and bilateral cochlea segmentation on hrT2 scans as provided in the testing set (N=137). This problem is particularly challenging given the large intensity distribution gap across the modalities and the small volume of the structures. A total of 55 teams from 16 countries submitted predictions to the validation leaderboard. Among them, 16 teams from 9 different countries submitted their algorithm for the evaluation phase. The level of performance reached by the top-performing teams is strikingly high (best median Dice score - VS: 88.4%; Cochleas: 85.7%) and close to full supervision (median Dice score - VS: 92.5%; Cochleas: 87.7%). All top-performing methods made use of an image-to-image translation approach to transform the source-domain images into pseudo-target-domain images. A segmentation network was then trained using these generated images and the manual annotations provided for the source image.

Toward semi-automatic biologically effective dose treatment plan optimisation for Gamma Knife radiosurgery.

Objective.Dose-rate effects in Gamma Knife radiosurgery treatments can lead to varying biologically effective dose (BED) levels for the same physical dose. The non-convex BED model depends on the delivery sequence and creates a non-trivial treatment planning problem. We investigate the feasibility of employing inverse planning methods to generate treatment plans exhibiting desirable BED characteristics using the per iso-centre beam-on times and delivery sequence.Approach.We implement two dedicated optimisation algorithms. One approach relies on mixed-integer linear programming (MILP) using a purposely developed convex underestimator for the BED to mitigate local minima issues at the cost of computational complexity. The second approach (local optimisation) is faster and potentially usable in a clinical setting but more prone to local minima issues. It sequentially executes the beam-on time (quasi-Newton method) and sequence optimisation (local search algorithm). We investigate the trade-off between time to convergence and solution quality by evaluating the resulting treatment plans' objective function values and clinical parameters. We also study the treatment time dependence of the initial and optimised plans using BED95(BED delivered to 95% of the target volume) values.Main results.When optimising the beam-on times and delivery sequence, the local optimisation approach converges several orders of magnitude faster than the MILP approach (minutes versus hours-days) while typically reaching within 1.2% (0.02-2.08%) of the final objective function value. The quality parameters of the resulting treatment plans show no meaningful difference between the local and MILP optimisation approaches. The presented optimisation approaches remove the treatment time dependence observed in the original treatment plans, and the chosen objectives successfully promote more conformal treatments.Significance.We demonstrate the feasibility of using an inverse planning approach within a reasonable time frame to ensure BED-based objectives are achieved across varying treatment times and highlight the prospect of further improvements in treatment plan quality.

Automated image curation in diabetic retinopathy screening using deep learning.

Diabetic retinopathy (DR) screening images are heterogeneous and contain undesirable non-retinal, incorrect field and ungradable samples which require curation, a laborious task to perform manually. We developed and validated single and multi-output laterality, retinal presence, retinal field and gradability classification deep learning (DL) models for automated curation. The internal dataset comprised of 7743 images from DR screening (UK) with 1479 external test images (Portugal and Paraguay). Internal vs external multi-output laterality AUROC were right (0.994 vs 0.905), left (0.994 vs 0.911) and unidentifiable (0.996 vs 0.680). Retinal presence AUROC were (1.000 vs 1.000). Retinal field AUROC were macula (0.994 vs 0.955), nasal (0.995 vs 0.962) and other retinal field (0.997 vs 0.944). Gradability AUROC were (0.985 vs 0.918). DL effectively detects laterality, retinal presence, retinal field and gradability of DR screening images with generalisation between centres and populations. DL models could be used for automated image curation within DR screening.

Post-vaccination infection rates and modification of COVID-19 symptoms in vaccinated UK school-aged children and adolescents: A prospective longitudinal cohort study.

Background: We aimed to explore the effectiveness of one-dose BNT162b2 vaccination upon SARS-CoV-2 infection, its effect on COVID-19 presentation, and post-vaccination symptoms in children and adolescents (CA) in the UK during periods of Delta and Omicron variant predominance. Methods: In this prospective longitudinal cohort study, we analysed data from 115,775 CA aged 12-17 years, proxy-reported through the Covid Symptom Study (CSS) smartphone application. We calculated post-vaccination infection risk after one dose of BNT162b2, and described the illness profile of CA with post-vaccination SARS-CoV-2 infection, compared to unvaccinated CA, and post-vaccination side-effects. Findings: Between August 5, 2021 and February 14, 2022, 25,971 UK CA aged 12-17 years received one dose of BNT162b2 vaccine. The probability of testing positive for infection diverged soon after vaccination, and was lower in CA with prior SARS-CoV-2 infection. Vaccination reduced proxy-reported infection risk (-80·4% (95% CI -0·82 -0·78) and -53·7% (95% CI -0·62 -0·43) at 14-30 days with Delta and Omicron variants respectively, and -61·5% (95% CI -0·74 -0·44) and -63·7% (95% CI -0·68 -0.59) after 61-90 days). Vaccinated CA who contracted SARS-CoV-2 during the Delta period had milder disease than unvaccinated CA; during the Omicron period this was only evident in children aged 12-15 years. Overall disease profile was similar in both vaccinated and unvaccinated CA. Post-vaccination local side-effects were common, systemic side-effects were uncommon, and both resolved within few days (3 days in most cases). Interpretation: One dose of BNT162b2 vaccine reduced risk of SARS-CoV-2 infection for at least 90 days in CA aged 12-17 years. Vaccine protection varied for SARS-CoV-2 variant type (lower for Omicron than Delta variant), and was enhanced by pre-vaccination SARS-CoV-2 infection. Severity of COVID-19 presentation after vaccination was generally milder, although unvaccinated CA also had generally mild disease. Overall, vaccination was well-tolerated. Funding: UK Government Department of Health and Social Care, Chronic Disease Research Foundation, The Wellcome Trust, UK Engineering and Physical Sciences Research Council, UK Research and Innovation London Medical Imaging & Artificial Intelligence Centre for Value Based Healthcare, UK National Institute for Health Research, UK Medical Research Council, British Heart Foundation and Alzheimer's Society, and ZOE Limited.