A prospective study of MRI biomarkers in the brain and lower limb muscles for prediction of lower limb motor recovery following stroke.

The aim of this prospective observational longitudinal study was to explore and decipher the predictive value of prospective MRI biomarkers in the brain and lower limb muscles for 3-month lower limb motor recovery following stroke. In the brain, we measured the integrity of the corticospinal tract (fractional anisotropy/"FA"). In the muscles, we measured volume, fatty replacement (fat fraction analysis and proton spectroscopy) and oedema. Measurements were taken at two time points: (1) within 4 weeks of stroke (baseline measurement, clinical and imaging) and (2) 3 months following stroke (follow up measurement, clinical only). Clinical measurements consisted of assessments of functional ability and strength (Fugl-Meyer score, motor NIHSS, Functional Ambulation Category/"FAC", and muscle dynamometry). Twenty-three patients completed imaging and clinical assessments at baseline and follow-up; five patients had partial imaging assessment. The results provided some evidence that damage to the corticospinal tract would result in less motor recovery: recovery of the Fugl-Meyer score and dynamometric ankle plantarflexion, ankle dorsiflexion, and knee extension correlated positively and significantly with fractional anisotropy (0.406-0.457; p = 0.034-p = 0.016). However, fractional anisotropy demonstrated a negative correlation with recovery of the Functional Ambulation Category (-0.359, p = 0.046). For the muscle imaging, significant inverse correlation was observed between vastus lateralis fat fraction vs. NIHSS recovery (-0.401, p = 0.04), and a strong positive correlation was observed between ratio of intra- to extra-myocellular lipid concentrations and the recovery of knee flexion (0.709, p = 0.007). This study supports previous literature indicating a positive correlation between the integrity of the corticospinal tract and motor recovery post-stroke, expanding the limited available literature describing this relationship specifically for the lower limb. However, recovery of functional ambulation behaved differently to other clinical recovery markers by demonstrating an inverse relationship with corticospinal tract integrity. The study also introduces some muscle imaging biomarkers as potentially valuable in the prediction of 3-month lower limb motor recovery following stroke.

Broadening the Spectrum of SLC22A5 Phenotype: Primary Carnitine Deficiency Presenting with Focal Myoclonus.

Primary carnitine deficiency (PCD) is caused by pathogenic variants of the SLC22A5 gene, which encodes a transmembrane protein that functions as a high affinity carnitine transporter. Carnitine is essential for the transport of acyl-CoA, produced from fatty acids, into the mitochondria where they are oxidised to produce energy. We present the case history of an 8-year-old boy who presented with fever, lethargy, focal rhythmic (3 Hz) left wrist twitching, and severe encephalopathy. MRI brain showed basal ganglia involvement. Metabolic investigations revealed low serum carnitine; whole genome sequencing confirmed compound heterozygous SLC22A5 mutations. With carnitine replacement, intensive care support, and neurorehabilitation, he made a remarkable recovery, regaining independent breathing, speech, mobility, and hand use. Seizure presentation in PCD is rare and presentation with sustained focal myoclonus has not been previously reported. This case expands the known phenotype of PCD. Prompt carnitine replacement is imperative.

Applying machine learning classifiers to automate quality assessment of paediatric dynamic susceptibility contrast (DSC-) MRI data.

OBJECTIVE: Investigate the performance of qualitative review (QR) for assessing dynamic susceptibility contrast (DSC-) MRI data quality in paediatric normal brain and develop an automated alternative to QR. METHODS: 1027 signal-time courses were assessed by Reviewer 1 using QR. 243 were additionally assessed by Reviewer 2 and % disagreements and Cohen's κ (κ) were calculated. The signal drop-to-noise ratio (SDNR), root mean square error (RMSE), full width half maximum (FWHM) and percentage signal recovery (PSR) were calculated for the 1027 signal-time courses. Data quality thresholds for each measure were determined using QR results. The measures and QR results trained machine learning classifiers. Sensitivity, specificity, precision, classification error and area under the curve from a receiver operating characteristic curve were calculated for each threshold and classifier. RESULTS: Comparing reviewers gave 7% disagreements and κ = 0.83. Data quality thresholds of: 7.6 for SDNR; 0.019 for RMSE; 3 s and 19 s for FWHM; and 42.9 and 130.4% for PSR were produced. SDNR gave the best sensitivity, specificity, precision, classification error and area under the curve values of 0.86, 0.86, 0.93, 14.2% and 0.83. Random forest was the best machine learning classifier, giving sensitivity, specificity, precision, classification error and area under the curve of 0.94, 0.83, 0.93, 9.3% and 0.89. CONCLUSION: The reviewers showed good agreement. Machine learning classifiers trained on signal-time course measures and QR can assess quality. Combining multiple measures reduces misclassification. ADVANCES IN KNOWLEDGE: A new automated quality control method was developed, which trained machine learning classifiers using QR results.

A Primary Spinal Nongerminomatous Germ Cell Tumor With Multiple Intracranial Metastases in a Toddler.

Central nervous system germ cell tumors (CNS-GCTs) comprise 4% of all pediatric CNS tumors, with one third being nongerminomatous GCT (CNS-NG-GCT) type. The majority of these tumors arise in the intracranial compartment with 20% having drop metastases in the spine. We present a rare case of a 2-year-old boy with a primary intradural-extramedullary NG-GCT arising from the lumbosacral spine with a trifecta of unfavorable features, that is, young age, alpha-feto protein >1000 ng/mL, and disseminated disease within the cranium. Owing to his young age, he was treated with chemotherapy alone, avoiding radiation. His tumor marker (alpha-feto protein) declined from 8468 to 10 k-U/L over 8 weeks, and he remained in remission at the last follow-up. This atypical presentation of an intradural-extramedullary tumor with cranial dissemination in a childhood NG-GCT has yet to be described in the literature. Here we use this opportunity to highlight the treatment strategies and challenges in this unique clinical case.

Imaging response assessment for CNS germ cell tumours: consensus recommendations from the European Society for Paediatric Oncology Brain Tumour Group and North American Children's Oncology Group.

Homogeneous and common objective disease assessments and standardised response criteria are important for better international clinical trials for CNS germ cell tumours. Currently, European protocols differ from those of North America (the USA and Canada) in terms of criteria to assess radiological disease response. An international working group of the European Society for Paediatric Oncology Brain Tumour Group and North American Children's Oncology Group was therefore established to review existing literature and current practices, identify major challenges regarding imaging assessment, and develop consensus recommendations for imaging response assessment for patients with CNS germ cell tumours. New clinical imaging standards were defined for the most common sites of CNS germ cell tumour and for the definition of locoregional extension. These new standards will allow the evaluation of response to therapy in patients with CNS germ cell tumours to be more consistent, and facilitate direct comparison of treatment outcomes across international studies.

Dynamic susceptibility-contrast magnetic resonance imaging with contrast agent leakage correction aids in predicting grade in pediatric brain tumours: a multicenter study.

BACKGROUND: Relative cerebral blood volume (rCBV) measured using dynamic susceptibility-contrast MRI can differentiate between low- and high-grade pediatric brain tumors. Multicenter studies are required for translation into clinical practice. OBJECTIVE: We compared leakage-corrected dynamic susceptibility-contrast MRI perfusion parameters acquired at multiple centers in low- and high-grade pediatric brain tumors. MATERIALS AND METHODS: Eighty-five pediatric patients underwent pre-treatment dynamic susceptibility-contrast MRI scans at four centers. MRI protocols were variable. We analyzed data using the Boxerman leakage-correction method producing pixel-by-pixel estimates of leakage-uncorrected (rCBVuncorr) and corrected (rCBVcorr) relative cerebral blood volume, and the leakage parameter, K2. Histological diagnoses were obtained. Tumors were classified by high-grade tumor. We compared whole-tumor median perfusion parameters between low- and high-grade tumors and across tumor types. RESULTS: Forty tumors were classified as low grade, 45 as high grade. Mean whole-tumor median rCBVuncorr was higher in high-grade tumors than low-grade tumors (mean ± standard deviation [SD] = 2.37±2.61 vs. -0.14±5.55; P<0.01). Average median rCBV increased following leakage correction (2.54±1.63 vs. 1.68±1.36; P=0.010), remaining higher in high-grade tumors than low grade-tumors. Low-grade tumors, particularly pilocytic astrocytomas, showed T1-dominant leakage effects; high-grade tumors showed T2*-dominance (mean K2=0.017±0.049 vs. 0.002±0.017). Parameters varied with tumor type but not center. Median rCBVuncorr was higher (mean = 1.49 vs. 0.49; P=0.015) and K2 lower (mean = 0.005 vs. 0.016; P=0.013) in children who received a pre-bolus of contrast agent compared to those who did not. Leakage correction removed the difference. CONCLUSION: Dynamic susceptibility-contrast MRI acquired at multiple centers helped distinguish between children's brain tumors. Relative cerebral blood volume was significantly higher in high-grade compared to low-grade tumors and differed among common tumor types. Vessel leakage correction is required to provide accurate rCBV, particularly in low-grade enhancing tumors.

Settling moths are the vital component of pollination in Himalayan ecosystem of North-East India, pollen transfer network approach revealed.

Majority of the pollination related studies are based on the diurnal pollinators, and the nocturnal pollinators received less scientific attention. We reveal the significance of settling moths in pollination of angiosperm families in Himalayan ecosystem of North-East India. The refined and novel method of pollen extraction from the proboscides provides a more robust assessment of the pollen carrying capacity. The study is based on one of the largest data sets (140 pollen transporter moth species (PTMS)), with interpretation based on seasonal as well as altitudinal data. In the present study about 65% moths (91 species) carried sufficient quantities of pollen grains to be considered as potential pollinators (PPMS). Teliphasa sp. (Crambidae) and Cuculia sp. (Noctuidae) are found to carry the highest quantity of pollen. We found pollen grains of 21 plant families and the abundant pollen are from Betulaceae, Fabaceae, Rosaceae and Ericaceae. Species composition of PTMS and PPMS in pre-monsoon, monsoon, and post-monsoon revealed the dominance of Geometridae. Maximum diversity of PTMS and PPMS is found from 2000 to 2500 m altitude. The nocturnal pollen transfer network matrices exhibited high degree of selectivity (H2' = 0.86).

Ground Target Tracking Using an Airborne Angle-Only Sensor with Terrain Uncertainty and Sensor Biases.

Airborne angle-only sensors can be used to track stationary or mobile ground targets. In order to make the problem observable in 3-dimensions (3-D), the height of the target (i.e., the height of the terrain) from the sea-level is needed to be known. In most of the existing works, the terrain height is assumed to be known accurately. However, the terrain height is usually obtained from Digital Terrain Elevation Data (DTED), which has different resolution levels. Ignoring the terrain height uncertainty in a tracking algorithm will lead to a bias in the estimated states. In addition to the terrain uncertainty, another common source of uncertainty in angle-only sensors is the sensor biases. Both these uncertainties must be handled properly to obtain better tracking accuracy. In this paper, we propose algorithms to estimate the sensor biases with the target(s) of opportunity and algorithms to track targets with terrain and sensor bias uncertainties. Sensor bias uncertainties can be reduced by estimating the biases using the measurements from the target(s) of opportunity with known horizontal positions. This step can be an optional step in an angle-only tracking problem. In this work, we have proposed algorithms to pick optimal targets of opportunity to obtain better bias estimation and algorithms to estimate the biases with the selected target(s) of opportunity. Finally, we provide a filtering framework to track the targets with terrain and bias uncertainties. The Posterior Cramer-Rao Lower Bound (PCRLB), which provides the lower bound on achievable estimation error, is derived for the single target filtering with an angle-only sensor with terrain uncertainty and measurement biases. The effectiveness of the proposed algorithms is verified by Monte Carlo simulations. The simulation results show that sensor biases can be estimated accurately using the target(s) of opportunity and the tracking accuracies of the targets can be improved significantly using the proposed algorithms when the terrain and bias uncertainties are present.

Metabolite selection for machine learning in childhood brain tumour classification.

MRS can provide high accuracy in the diagnosis of childhood brain tumours when combined with machine learning. A feature selection method such as principal component analysis is commonly used to reduce the dimensionality of metabolite profiles prior to classification. However, an alternative approach of identifying the optimal set of metabolites has not been fully evaluated, possibly due to the challenges of defining this for a multi-class problem. This study aims to investigate metabolite selection from in vivo MRS for childhood brain tumour classification. Multi-site 1.5 T and 3 T cohorts of patients with a brain tumour and histological diagnosis of ependymoma, medulloblastoma and pilocytic astrocytoma were retrospectively evaluated. Dimensionality reduction was undertaken by selecting metabolite concentrations through multi-class receiver operating characteristics and compared with principal component analysis. Classification accuracy was determined through leave-one-out and k-fold cross-validation. Metabolites identified as crucial in tumour classification include myo-inositol (P < 0.05, AUC=0.81±0.01 ), total lipids and macromolecules at 0.9 ppm (P < 0.05, AUC=0.78±0.01 ) and total creatine (P < 0.05, AUC=0.77±0.01 ) for the 1.5 T cohort, and glycine (P < 0.05, AUC=0.79±0.01 ), total N-acetylaspartate (P < 0.05, AUC=0.79±0.01 ) and total choline (P < 0.05, AUC=0.75±0.01 ) for the 3 T cohort. Compared with the principal components, the selected metabolites were able to provide significantly improved discrimination between the tumours through most classifiers (P < 0.05). The highest balanced classification accuracy determined through leave-one-out cross-validation was 85% for 1.5 T 1 H-MRS through support vector machine and 75% for 3 T 1 H-MRS through linear discriminant analysis after oversampling the minority. The study suggests that a group of crucial metabolites helps to achieve better discrimination between childhood brain tumours.

A Novel Case of Homozygous Interferon Alpha/Beta Receptor Alpha Chain (IFNAR1) Deficiency With Hemophagocytic Lymphohistiocytosis.

We present a case of complete deficiency of the interferon alpha/beta receptor alpha chain (IFNAR1) in a child with fatal systemic hyperinflammation, apparently provoked by live-attenuated viral vaccination. Such pathologic hyperinflammation, fulfilling criteria for hemophagocytic lymphohistiocytosis, is an emerging phenotype accompanying inborn errors of type I interferon immunity.

Combining multi-site magnetic resonance imaging with machine learning predicts survival in pediatric brain tumors.

Brain tumors represent the highest cause of mortality in the pediatric oncological population. Diagnosis is commonly performed with magnetic resonance imaging. Survival biomarkers are challenging to identify due to the relatively low numbers of individual tumor types. 69 children with biopsy-confirmed brain tumors were recruited into this study. All participants had perfusion and diffusion weighted imaging performed at diagnosis. Imaging data were processed using conventional methods, and a Bayesian survival analysis performed. Unsupervised and supervised machine learning were performed with the survival features, to determine novel sub-groups related to survival. Sub-group analysis was undertaken to understand differences in imaging features. Survival analysis showed that a combination of diffusion and perfusion imaging were able to determine two novel sub-groups of brain tumors with different survival characteristics (p < 0.01), which were subsequently classified with high accuracy (98%) by a neural network. Analysis of high-grade tumors showed a marked difference in survival (p = 0.029) between the two clusters with high risk and low risk imaging features. This study has developed a novel model of survival for pediatric brain tumors. Tumor perfusion plays a key role in determining survival and should be considered as a high priority for future imaging protocols.

Classification of paediatric brain tumours by diffusion weighted imaging and machine learning.

To determine if apparent diffusion coefficients (ADC) can discriminate between posterior fossa brain tumours on a multicentre basis. A total of 124 paediatric patients with posterior fossa tumours (including 55 Medulloblastomas, 36 Pilocytic Astrocytomas and 26 Ependymomas) were scanned using diffusion weighted imaging across 12 different hospitals using a total of 18 different scanners. Apparent diffusion coefficient maps were produced and histogram data was extracted from tumour regions of interest. Total histograms and histogram metrics (mean, variance, skew, kurtosis and 10th, 20th and 50th quantiles) were used as data input for classifiers with accuracy determined by tenfold cross validation. Mean ADC values from the tumour regions of interest differed between tumour types, (ANOVA P < 0.001). A cut off value for mean ADC between Ependymomas and Medulloblastomas was found to be of 0.984 × 10-3 mm2 s-1 with sensitivity 80.8% and specificity 80.0%. Overall classification for the ADC histogram metrics were 85% using Naïve Bayes and 84% for Random Forest classifiers. The most commonly occurring posterior fossa paediatric brain tumours can be classified using Apparent Diffusion Coefficient histogram values to a high accuracy on a multicentre basis.

Case report: MRI findings of acute uremic encephalopathy in a 1-year-old boy.

We present a 1-year-old boy who presented to the emergency department with a 7-day history of diarrhoea and vomiting. The initial renal function profile demonstrated a urea of 55 mmol l-1 (normal range between 5 and 20 mmol l-1), creatinine 695 micromol/L (normal range between 62-106 micromol/L) and potassium 9.1 mmol l-1 (normal range between 3.5-5.0 mmol l-1), with a profound metabolic acidosis. Upon examination, there were no significant findings, specifically no neurological abnormality. He was prescribed back-to-back Salbutamol nebulisers, to increase the shift of extracellular potassium into the intracellular space, followed by i.v. calcium gluconate, with some improvement in potassium levels. A further 5 mmol of sodium bicarbonate was given, as well as a stat dose of 1 mg/kg furosemide, and per rectal calcium resonium. He was then commenced on an infusion with 10% dextrose with insulin. He was subsequently found to be in urinary retention and a catheter was inserted, which drained 1700 ml. A subsequent renal function profile, 24 hours after admission, demonstrated improvement with urea 39 mmol l-1, creatinine 300 micromol/L and potassium 3.0 mEq/L.

Complement factor I deficiency: A potentially treatable cause of fulminant cerebral inflammation.

OBJECTIVE: To raise awareness of complement factor I (CFI) deficiency as a potentially treatable cause of severe cerebral inflammation. METHODS: Case report with neuroradiology, neuropathology, and functional data describing the mutation with review of literature. RESULTS: We present a case of acute, fulminant, destructive cerebral edema in a previously well 11-year-old, demonstrating massive activation of complement pathways on neuropathology and compound heterozygote status for 2 pathogenic mutations in CFI which result in normal levels but completely abrogate function. CONCLUSIONS: Our case adds to a very small number of extant reports of this phenomenon associated with a spectrum of inflammatory histopathologies including hemorrhagic leukoencephalopathy and clinical presentations resembling severe acute disseminated encephalomyelitis. CFI deficiency can result in uncontrolled activation of the complement pathways in the brain resulting in devastating cerebral inflammation. The deficit is latent, but the catastrophic dysregulation of the complement system may be the result of a C3 acute phase response. Diagnoses to date have been retrospective. Diagnosis requires a high index of suspicion and clinician awareness of the limitations of first-line clinical tests of complement activity and activation. Simple measurement of circulating CFI levels, as here, may fail to diagnose functional deficiency with absent CFI activity. These diagnostic challenges may mean that the CFI deficiency is being systematically under-recognized as a cause of fulminant cerebral inflammation. Complement inhibitory therapies (such as eculizumab) offer new potential treatment, underlining the importance of prompt recognition, and real-time whole exome sequencing may play an important future role.

Distinguishing between paediatric brain tumour types using multi-parametric magnetic resonance imaging and machine learning: A multi-site study.

The imaging and subsequent accurate diagnosis of paediatric brain tumours presents a radiological challenge, with magnetic resonance imaging playing a key role in providing tumour specific imaging information. Diffusion weighted and perfusion imaging are commonly used to aid the non-invasive diagnosis of children's brain tumours, but are usually evaluated by expert qualitative review. Quantitative studies are mainly single centre and single modality. The aim of this work was to combine multi-centre diffusion and perfusion imaging, with machine learning, to develop machine learning based classifiers to discriminate between three common paediatric tumour types. The results show that diffusion and perfusion weighted imaging of both the tumour and whole brain provide significant features which differ between tumour types, and that combining these features gives the optimal machine learning classifier with >80% predictive precision. This work represents a step forward to aid in the non-invasive diagnosis of paediatric brain tumours, using advanced clinical imaging.

Microbleeds in dementia with Lewy bodies.

INTRODUCTION: Microbleeds are associated with the development of dementia in older people and are common in Alzheimer's disease (AD). Their prevalence and clinical importance in dementia with Lewy bodies (DLB) is unclear. The objective of this study was to compare the rates of microbleeds in DLB with those in AD and healthy older people, and investigate associations between microbleeds and amyloid deposition, vascular risk and disease severity in DLB. METHODS: DLB (n = 30), AD (n = 18) and control (n = 20) participants underwent clinical assessment at baseline and 1 year in this longitudinal observational study. 3T MRI (including T2* susceptibility weighted imaging) and florbetapir PET were carried out at baseline. Microbleeds were rated visually and a standardised uptake value ratio (SUVR) was calculated from florbetapir PET scans. RESULTS: 40% of DLB subjects had microbleeds compared with 50% of AD and 15% of controls. Compared to DLB without microbleeds, those with microbleeds had higher systolic BP (156 ± 26 v. 135 ± 19 mmHg; p = 0.03), but did not have greater levels of vascular disease or amyloid deposition (SUVR 1.25 ± 0.24 v. 1.25 ± 0.22; p = 0.33). There was evidence of less severe dementia in DLB participants with microbleeds, but these differences may have been driven by a shorter disease duration in those with microbleeds. CONCLUSION: The presence of microbleeds in DLB is associated with higher blood pressure, but not with other measures of vascular disease or amyloid deposition. The relationship between microbleeds and clinical presentation remains unclear.

Asymmetrical Bioimpedance in the Anterior Circulation for Urgent Stratification of suspected Stroke (ABACUS Stroke): study protocol for a diagnostic accuracy study.

BACKGROUND: Early identification and treatment of stroke improve outcome. Ischaemic stroke due to large vessel occlusion (LVO) benefits from time-critical thrombectomy but this is only available in highly specialised healthcare services. Cerebral Bioimpedance Asymmetry (CBA) measurement obtained with the portable and rapid Cerebrotech Visor™ System device may be able to identify certain types of stroke including LVO. This test could be deployed pre-hospital and used to immediately direct patients to the most appropriate healthcare service for treatment. This study is evaluating the diagnostic accuracy of CBA measurements obtained from a real-world population of suspected stroke. METHODS: Study design: Prospective observational cohort study.Setting: A hyperacute stroke unit and neuroscience centre in North East England.Participants: Adults with a paramedic assigned diagnosis of suspected stroke arriving at hospital within 6 hours of symptom onset.Index Test: Cerebral Bioimpedance Asymmetry measurement performed using the Cerebrotech Visor™ System. Measurement values produce continuous data (range 0 -100); pre-defined threshold for positive state ≥ 10.Reference Standard Tests: Standard CT brain +/- CT/MR angiography, and expert clinician opinion will establish the following clinical outcomes which constitute the suspected stroke population: ischaemic stroke +/- large vessel occlusion; symptomatic severe anterior vessel stenosis; large (≥60ml) and small (<60mls) vessel intracerebral haemorrhage; transient ischaemic attack; stroke mimic conditions; prior territorial stroke.Analyses: Sensitivity, specificity, negative and positive predictive values, area under the Receiver Operating Characteristic curve for identification of i) "complex stroke" (ischaemic stroke with large vessel occlusion or symptomatic severe anterior vessel stenosis or intracerebral haemorrhage ≥60ml or prior territorial stroke) and ii) ischaemic stroke with large vessel occlusion in isolation.Sample size: 124 participants. DISCUSSION: The results from this study will determine how accurately CBA measurement using the Cerebrotech Visor™ System can identify key stroke types within the suspected stroke population. Acceptable diagnostic performance would be an important step forwards for access to time-critical treatments. TRIAL REGISTRATION: Registered with ISRCTN (identifier: ISRCTN79169844) on 06/08/2018.

Unusual cause of treatable Parkinsonism.

A 78-year-old woman presented with features of bilateral, asymmetric Parkinsonism for 1 year, with prominent difficulties with continence, swallowing and apathy. Brain imaging showed evidence of chronic venous sinus thrombosis with dilated serpiginous vessels over the brain surface in keeping with a dural arteriovenous fistula, together with high T2 signal on MRI in the basal ganglia. Having responded only modestly to levodopa, she received 6 months of anticoagulation followed by embolisation of the dural arteriovenous fistula, with good response. Cerebral dural arteriovenous fistula is a rare, structural cause of atypical Parkinsonism.

Effects of antiplatelet therapy on stroke risk by brain imaging features of intracerebral haemorrhage and cerebral small vessel diseases: subgroup analyses of the RESTART randomised, open-label trial.

BACKGROUND: Findings from the RESTART trial suggest that starting antiplatelet therapy might reduce the risk of recurrent symptomatic intracerebral haemorrhage compared with avoiding antiplatelet therapy. Brain imaging features of intracerebral haemorrhage and cerebral small vessel diseases (such as cerebral microbleeds) are associated with greater risks of recurrent intracerebral haemorrhage. We did subgroup analyses of the RESTART trial to explore whether these brain imaging features modify the effects of antiplatelet therapy. METHODS: RESTART was a prospective, randomised, open-label, blinded-endpoint, parallel-group trial at 122 hospitals in the UK that assessed whether starting antiplatelet therapy might reduce the risk of recurrent symptomatic intracerebral haemorrhage compared with avoiding antiplatelet therapy. For this prespecified subgroup analysis, consultant neuroradiologists masked to treatment allocation reviewed brain CT or MRI scans performed before randomisation to confirm participant eligibility and rate features of the intracerebral haemorrhage and surrounding brain. We followed participants for primary (recurrent symptomatic intracerebral haemorrhage) and secondary (ischaemic stroke) outcomes for up to 5 years (reported elsewhere). For this report, we analysed eligible participants with intracerebral haemorrhage according to their treatment allocation in primary subgroup analyses of cerebral microbleeds on MRI and in exploratory subgroup analyses of other features on CT or MRI. The trial is registered with the ISRCTN registry, number ISRCTN71907627. FINDINGS: Between May 22, 2013, and May 31, 2018, 537 participants were enrolled, of whom 525 (98%) had intracerebral haemorrhage: 507 (97%) were diagnosed on CT (252 assigned to start antiplatelet therapy and 255 assigned to avoid antiplatelet therapy, of whom one withdrew and was not analysed) and 254 (48%) underwent the required brain MRI protocol (122 in the start antiplatelet therapy group and 132 in the avoid antiplatelet therapy group). There were no clinically or statistically significant hazards of antiplatelet therapy on recurrent intracerebral haemorrhage in primary subgroup analyses of cerebral microbleed presence (2 or more) versus absence (0 or 1) (adjusted hazard ratio [HR] 0·30 [95% CI 0·08-1·13] vs 0·77 [0·13-4·61]; pinteraction=0·41), cerebral microbleed number 0-1 versus 2-4 versus 5 or more (HR 0·77 [0·13-4·62] vs 0·32 [0·03-3·66] vs 0·33 [0·07-1·60]; pinteraction=0·75), or cerebral microbleed strictly lobar versus other location (HR 0·52 [0·004-6·79] vs 0·37 [0·09-1·28]; pinteraction=0·85). There was no evidence of heterogeneity in the effects of antiplatelet therapy in any exploratory subgroup analyses (all pinteraction>0·05). INTERPRETATION: Our findings exclude all but a very modest harmful effect of antiplatelet therapy on recurrent intracerebral haemorrhage in the presence of cerebral microbleeds. Further randomised trials are needed to replicate these findings and investigate them with greater precision. FUNDING: British Heart Foundation.