Costs and health effects of CT perfusion-based selection for endovascular thrombectomy within 6 hours of stroke onset: a model-based health economic evaluation.

BACKGROUND: Although CT perfusion (CTP) is often incorporated in acute stroke workflows, it remains largely unclear what the associated costs and health implications are in the long run of CTP-based patient selection for endovascular treatment (EVT) in patients presenting within 6 hours after symptom onset with a large vessel occlusion. METHODS: Patients with a large vessel occlusion were included from a Dutch nationwide cohort (n=703) if CTP imaging was performed before EVT within 6 hours after stroke onset. Simulated cost and health effects during 5 and 10 years follow-up were compared between CTP based patient selection for EVT and providing EVT to all patients. Outcome measures were the net monetary benefit at a willingness-to-pay of €80 000 per quality-adjusted life year, incremental cost-effectiveness ratio), difference in costs from a healthcare payer perspective (ΔCosts) and quality-adjusted life years (ΔQALY) per 1000 patients for 1000 model iterations as outcomes. RESULTS: Compared with treating all patients, CTP-based selection for EVT at the optimised ischaemic core volume (ICV≥110 mL) or core-penumbra mismatch ratio (MMR≤1.4) thresholds resulted in losses of health (median ΔQALYs for ICV≥110 mL: -3.3 (IQR: -5.9 to -1.1), for MMR≤1.4: 0.0 (IQR: -1.3 to 0.0)) with median ΔCosts for ICV≥110 mL of -€348 966 (IQR: -€712 406 to -€51 158) and for MMR≤1.4 of €266 513 (IQR: €229 403 to €380 110)) per 1000 patients. Sensitivity analyses did not yield any scenarios for CTP-based selection of patients for EVT that were cost-effective for improving health, including patients aged ≥80 years CONCLUSION: In EVT-eligible patients presenting within 6 hours after symptom onset, excluding patients based on CTP parameters was not cost-effective and could potentially harm patients.

Classic "PCH" Genes are a Rare Cause of Radiologic Pontocerebellar Hypoplasia.

The term Pontocerebellar Hypoplasia (PCH) was initially used to designate a heterogeneous group of fetal-onset genetic neurodegenerative disorders. As a descriptive term, PCH refers to pons and cerebellum of reduced volume. In addition to the classic PCH types described in OMIM, many other disorders can result in a similar imaging appearance. This study aims to review imaging, clinical and genetic features and underlying etiologies of a cohort of children with PCH on imaging. We systematically reviewed brain images and clinical charts of 38 patients with radiologic evidence of PCH. Our cohort included 21 males and 17 females, with ages ranging between 8 days to 15 years. All individuals had pons and cerebellar vermis hypoplasia, and 63% had cerebellar hemisphere hypoplasia. Supratentorial anomalies were found in 71%. An underlying etiology was identified in 68% and included chromosomal (21%), monogenic (34%) and acquired (13%) causes. Only one patient had pathogenic variants in an OMIM listed PCH gene. Outcomes were poor regardless of etiology, though no one had regression. Approximately one third of patients deceased at a median age of 8 months. All individuals had global developmental delay, 50% were non-verbal, 64% were non-ambulatory and 45% required gastrostomy feeding. This cohort demonstrates that radiologic PCH has heterogenous etiologies and the "classic" OMIM-listed PCH genes underlie only a minority of cases. Broad genetic testing, including chromosomal microarray and exome or multigene panels, is recommended in individuals with PCH-like imaging appearance. Our results strongly suggest that the term PCH should be used to designate radiologic findings, and not to imply neurogenerative disorders.

Intraoperative magnetic resonance imaging in glioma surgery: a single-center experience.

PURPOSE: To investigate the effect of intraoperative magnetic resonance imaging (Io MRI) on overall and progression-free survival (OS and PFS), on the extent of resection (EOR) in patients with glioma, and impact of the radiological diagnosis on the decision to continue the surgery when a residual mass was detected on Io MRI. METHODS: The study comprised 153 glioma patients who received surgical treatment between 2013 and 2023. One-hundred twenty-five of them had Io MRI guidance during surgery. The remainder 28 patients constituted the control group who did not undergo Io MRI. All patients' age at surgery, gender, initial radiological diagnosis, primary tumor localization, EOR, last histopathological diagnosis, and the follow-up periods were recorded. RESULTS: The rate of tumor recurrence in Io MRI cases was significantly lower compared to the cases in the control group (p < .0001). It was decided to continue the operation in 45 Io MRI applied cases. This raised the gross total resection (GTR) rate from 33.6% to 49.6% in the Io MRI group. The frequency of GTR was significantly higher in patients with an initial radiological diagnosis of low grade glioma than those with high grade glioma. The shortest OS was seen in occipital gliomas. CONCLUSION: In this study, the convenience provided by the high-field MRI device was explored and proven both in reducing the tumor burden, increasing the PFS, and providing the surgeon with a maximal resection in the first operation.

Can large language models pass official high-grade exams of the European Society of Neuroradiology courses? A direct comparison between OpenAI chatGPT 3.5, OpenAI GPT4 and Google Bard.

We compared different LLMs, notably chatGPT, GPT4, and Google Bard and we tested whether their performance differs in subspeciality domains, in executing examinations from four different courses of the European Society of Neuroradiology (ESNR) notably anatomy/embryology, neuro-oncology, head and neck and pediatrics. Written exams of ESNR were used as input data, related to anatomy/embryology (30 questions), neuro-oncology (50 questions), head and neck (50 questions), and pediatrics (50 questions). All exams together, and each exam separately were introduced to the three LLMs: chatGPT 3.5, GPT4, and Google Bard. Statistical analyses included a group-wise Friedman test followed by a pair-wise Wilcoxon test with multiple comparison corrections. Overall, there was a significant difference between the 3 LLMs (p < 0.0001), with GPT4 having the highest accuracy (70%), followed by chatGPT 3.5 (54%) and Google Bard (36%). The pair-wise comparison showed significant differences between chatGPT vs GPT 4 (p < 0.0001), chatGPT vs Bard (p < 0. 0023), and GPT4 vs Bard (p < 0.0001). Analyses per subspecialty showed the highest difference between the best LLM (GPT4, 70%) versus the worst LLM (Google Bard, 24%) in the head and neck exam, while the difference was least pronounced in neuro-oncology (GPT4, 62% vs Google Bard, 48%). We observed significant differences in the performance of the three different LLMs in the running of official exams organized by ESNR. Overall GPT 4 performed best, and Google Bard performed worst. This difference varied depending on subspeciality and was most pronounced in head and neck subspeciality.

Go Green in Neuroradiology: towards reducing the environmental impact of its practice.

Raising public awareness about the relevance of supporting sustainable practices is required owing to the phenomena of global warming caused by the rising production of greenhouse gases. The healthcare sector generates a relevant proportion of the total carbon emissions in developed countries, and radiology is estimated to be a major contributor to this carbon footprint. Neuroradiology markedly contributes to this negative environmental effect, as this radiological subspecialty generates a high proportion of diagnostic and interventional imaging procedures, the majority of them requiring high energy-intensive equipment. Therefore, neuroradiologists and neuroradiological departments are especially responsible for implementing decisions and initiatives able to reduce the unfavourable environmental effects of their activities, by focusing on four strategic pillars-reducing energy, water, and helium use; properly recycling and/or disposing of waste and residues (including contrast media); encouraging environmentally friendly behaviour; and reducing the effects of ionizing radiation on the environment. The purpose of this article is to alert neuroradiologists about their environmental responsibilities and to analyse the most productive strategic axes, goals, and lines of action that contribute to reducing the environmental impact associated with their professional activities.

Automatic generation of conclusions from neuroradiology MRI reports through natural language processing.

PURPOSE: The conclusion section of a radiology report is crucial for summarizing the primary radiological findings in natural language and essential for communicating results to clinicians. However, creating these summaries is time-consuming, repetitive, and prone to variability and errors among different radiologists. To address these issues, we evaluated a fine-tuned Text-To-Text Transfer Transformer (T5) model for abstractive summarization to automatically generate conclusions for neuroradiology MRI reports in a low-resource language. METHODS: We retrospectively applied our method to a dataset of 232,425 neuroradiology MRI reports in Spanish. We compared various pre-trained T5 models, including multilingual T5 and those newly adapted for Spanish. For precise evaluation, we employed BLEU, METEOR, ROUGE-L, CIDEr, and cosine similarity metrics alongside expert radiologist assessments. RESULTS: The findings are promising, with the models specifically fine-tuned for neuroradiology MRI achieving scores of 0.46, 0.28, 0.52, 2.45, and 0.87 in the BLEU-1, METEOR, ROUGE-L, CIDEr, and cosine similarity metrics, respectively. In the radiological experts' evaluation, they found that in 75% of the cases evaluated, the conclusions generated by the system were as good as or even better than the manually generated conclusions. CONCLUSION: The methods demonstrate the potential and effectiveness of customizing state-of-the-art pre-trained models for neuroradiology, yielding automatic MRI report conclusions that nearly match expert quality. Furthermore, these results underscore the importance of designing and pre-training a dedicated language model for radiology report summarization.

Widespread haemorrhages in infants post-shunting (WHIPS): clinical features, risk factors and neuroimaging characteristics of a rare and under-recognised phenomenon.

PURPOSE: Infants undergoing CSF shunting procedures face a rare complication which we propose to rename "Widespread Haemorrhages in Infants Post-Shunting" (WHIPS) to better capture this unique phenomenon specific to infants undergoing CSF diversion. Our objective is to analyse the risk factors for WHIPS development and provide a detailed neuroradiological description of these haemorrhages. MATERIALS AND METHODS: A radiology information system (RIS) was searched using the search terms "shunt" and/or "catheter" and/or "drain" and/or "ventriculoperitoneal" and/or "VP" between September 2008 to January 2021 for patients < 12 months of age. Clinical data was compiled for each patient meeting the inclusion criteria. Included cases were reviewed by three radiologists for the presence of WHIPS with calculation of the bifrontal ratio and documenting haemorrhage number, morphology, location and lobar distribution. RESULTS: 51 patients met inclusion criteria, 8 WHIPS patients and 43 controls. There was a statistically significant correlation between a larger post-op head circumference and WHIPS (p = 0.04). WHIPS was associated with post-haemorrhagic hydrocephalus and post-infectious hydrocephalus (p = 0.009). WHIPS were identified in the cortico-subcortical regions, periventricular white matter, and deep white matter. Haemorrhages were either punctate, ovoid or confluent. Haemorrhages ranged from single to innumerable. CONCLUSIONS: WHIPS represent a rare and under-recognised complication of CSF shunting unique to the infantile population. We postulate deep and superficial medullary venous haemorrhage as an underlying mechanism related to disordered intracranial hydrodynamics which are exacerbated in the infantile population due to underdeveloped arachnoid granulations and a compliant skull.

A Simplified One-Size-Fits-All Approach to Carotid Stenting.

BACKGROUND: Carotid artery stenosis causes up to 20% of ischemic strokes. Stenting is used as an alternative to endarterectomy in symptomatic patients. Each commercially available stent offers numerous stent diameters/lengths. Most centers thus carefully match each individual stenosis to a specific stent length/diameter stent size. However, this process can be time-consuming and costly while the relative benefit of a custom stent sizing versus one-size-fits-all approach has not been well evaluated yet. We hypothesized that a 'one-size-fits-all' default approach to carotid stenting results in comparable results to a customized approach. METHODS: We conducted a descriptive retrospective cohort study on 154 patients who presented to our academic carotid revascularization clinic with symptomatic carotid artery stenosis who underwent carotid artery stenting for peri- and postprocedural carotid artery stenting complications. The primary outcomes were periprocedural (within 24 hours of the procedure) or postprocedural (within 30 days of the procedure) TIA, stroke, or death. The secondary outcome was the estimated degree of stenosis on follow-up ultrasound performed within 6 months of the procedure. RESULTS: The complication rate within the first 24 hours was 4.5% while that during the first 30 days postprocedure was 6.5%. Age over 80 and degree of stenosis on postprocedural cerebral angiogram were associated with an increased risk of complications. Severe restenosis was reported in 16.8% of patients within 6 months postprocedure. CONCLUSION: Our study suggests that using a simplified, one-size-fits-all, approach to carotid stenting results in safe and effective outcomes, suggesting an alternative to simplify a complex medical procedure.

Spontaneous regression of an epidermoid cyst in a pediatric patient-Case report and review of the literature.

Epidermoid cysts are infrequent, benign, slow-growing, space-occupying lesions that account for 0.5-1.8% of primary intracranial tumors. We report the case of a 17-month-old child who presented in 2015 for one episode of pallor associated with hypotonia. Epilepsy was excluded and MRI was recommended. The MRI was performed and there were no focal parenchymal lesions, but it showed an extra-axial ovoid lesion with imaging characteristics consistent with epidermoid cyst. Follow-up MRI at one year was performed and it showed minimal increased in dimensions of the cyst, without changes into the signal of the lesion. Another MRI was performed 7 years after and it showed complete resolution of the cyst. Six months afterwards, another MRI was performed and it confirmed the complete regression of the cyst, without any extra-axial masses reported. The patient did not present any neurological anomalies. No follow-up MRI was recommended afterwards. Spontaneous regression of epidermoid cyst in pediatric population is an extremely rare event, but it should be taken into account when the patient shows no symptoms. This is the third case of spontaneous regression of an epidermoid cyst reported in pediatric patients, and the first one in the temporal region. Careful follow-up and watchful waiting could be an option to surgical treatment in epidermoid cysts.

Multisystem factors contributing to redundant intracranial vascular imaging in the ED.

PURPOSE: To evaluate the multisystem factors contributing to redundant neurovascular orders in the ED. METHODS: This was an IRB-approved, retrospective study, performed at a single institution examining a 5-year history of redundant CTA/MRA head and neck (HN) exams performed in the ED for patients with no documented clinical change in mental status/neurological exam necessitating additional imaging. Factors contributing to redundant ordering including provider experience, synchronous order placement, and radiologist recommendations were examined. Additionally, the impact of duplicative imaging in terms of medical cost and ED length of stay was evaluated. RESULTS: 250 patients met inclusion criteria with both CTA/MRA of the HN performed during a single ED encounter (total 500 exams). 190 (76%) redundant exams were not recommended by a radiologist and contributed to an added ED length of stay of 3.6 h on average. Provider experience was not a significant contributing factor. 60 (24%) of redundant exams were recommended by a radiologist and were most frequently CTAs needed to clarify an area of artifact/high-grade stenosis/occlusion on a primary MRA exam. CONCLUSION: Evaluation of contributing factors to redundant CTA/MRA HN exams ordering has highlighted multiple associated factors including provider experience, recommendations by radiologists for clarification of MRA findings, as well as systems processes related to synchronous CTA/MRA order placement.

Assessment of post-thrombectomy brain hemorrhage in acute ischemic stroke with dual-energy CT: how reliable is it in clinical practice?

PURPOSE: Acute ischemic stroke is currently among the main causes of mortality in Western countries. The current guidelines suggest different flowcharts of diagnostic work-up and treatment modalities, including endovascular thrombectomy. Immediately after intra-arterial recanalization, a brain CT scan is usually performed to assess for the presence of peri-procedural complications; in this setting, it is very hard, if possible, to differentiate blood from iodinated contrast material, which is normally present in ischemic tissue because of BBB disruption. Dual-energy CT may be used for this purpose, exploiting its ability to discriminate different materials. MATERIALS AND METHODS: We retrospectively studied 44 patients with acute ischemic stroke who were treated with endovascular recanalization at San Giovanni Bosco Hospital in Turin and were then scanned with DECT technology. Subsequent scan was used as standard, since iodine from contrast staining is usually reabsorbed in 24 h and blood persists longer. A χ2 test of independence was performed to examine the relationship between blood detected by DECT scan after the endovascular procedure and the presence of blood in the same areas on the following scans, with a significant result: χ2 (1, N = 37) = 10.7086, p = 0.0010. RESULTS: Patients with blood detected on DECT scans had a double chance of having hemorrhagic infarction in follow-up scans, (RR 2.02). The sensitivity and specificity of DECT were respectively 70% and 90%, with an overall diagnostic accuracy of 76% and a positive and negative predictive value, respectively, of 95% and 53%. CONCLUSION: Dual-energy CT scan after endovascular recanalization in ischemic stroke identifies early hemorrhagic infarction with excellent specificity and good overall diagnostic accuracy, representing a reliable diagnostic tool in everyday clinical practice.

Artificial intelligence applied in acute ischemic stroke: from child to elderly.

This review will summarize artificial intelligence developments in acute ischemic stroke in recent years and forecasts for the future. Stroke is a major healthcare concern due to its effects on the patient's quality of life and its dependence on the timing of the identification as well as the treatment. In recent years, attention increased on the use of artificial intelligence (AI) systems to help categorize, prognosis, and to channel these patients toward the right therapeutic procedure. Machine learning (ML) and in particular deep learning (DL) systems using convoluted neural networks (CNN) are becoming increasingly popular. Various studies over the years evaluated the use of these methods of analysis and prediction in the assessment of stroke patients, and at the same time, several applications and software have been developed to support the neuroradiologists and the stroke team to improve patient outcomes.

Multicenter comparison using two AI stroke CT perfusion software packages for determining thrombectomy eligibility.

BACKGROUND: Stroke AI platforms assess infarcted core and potentially salvageable tissue (penumbra) to identify patients suitable for mechanical thrombectomy. Few studies have compared outputs of these platforms, and none have been multicenter or considered NIHSS or scanner/protocol differences. Our objective was to compare volume estimates and thrombectomy eligibility from two widely used CT perfusion (CTP) packages, Viz.ai and RAPID.AI, in a large multicenter cohort. METHODS: We analyzed CTP data of acute stroke patients with large vessel occlusion (LVO) from four institutions. Core and penumbra volumes were estimated by each software and DEFUSE-3 thrombectomy eligibility assessed. Results between software packages were compared and categorized by NIHSS score, scanner manufacturer/model, and institution. RESULTS: Primary analysis of 362 cases found statistically significant differences in both software's volume estimations, with subgroup analysis showing these differences were driven by results from a single scanner model, the Canon Aquilion One. Viz.ai provided larger estimates with mean differences of 8cc and 18cc for core and penumbra, respectively (p<0.001). NIHSS subgroup analysis also showed systematically larger Viz.ai volumes (p<0.001). Despite volume differences, a significant difference in thrombectomy eligibility was not found. Additional subgroup analysis showed significant differences in penumbra volume for the Phillips Ingenuity scanner, and thrombectomy eligibility for the Canon Aquilion One scanner at one center (7 % increased eligibility with Viz.ai, p=0.03). CONCLUSIONS: Despite systematic differences in core and penumbra volume estimates between Viz.ai and RAPID.AI, DEFUSE-3 eligibility was not statistically different in primary or NIHSS subgroup analysis. A DEFUSE-3 eligibility difference, however, was seen on one scanner at one institution, suggesting scanner model and local CTP protocols can influence performance and cause discrepancies in thrombectomy eligibility. We thus recommend centers discuss optimal scanning protocols with software vendors and scanner manufacturers to maximize CTP accuracy.

Influence of Time of Admission on Endovascular Thrombectomy (EVT): Comparison of Outcomes During Business Hours Versus Off-Business Hours.

Purpose:To investigate the differences in endovascular thrombectomy (EVT) outcomes of patients treated for acute ischaemic stroke (AIS) during business versus off-business hours. Methods: A single-centre retrospective cohort study of patients with AIS treated with EVT from February 1, 2015, to May 31, 2021, was performed at a comprehensive stroke centre (CSC). Patients were divided into business (Monday to Friday, 8 AM-5 PM) versus off-business hours groups. The primary outcome was functional neurological disability, scored using the modified Rankin Scale (mRS) at 90 days. Secondary outcomes included the rate of successful reperfusion and procedural workflow time delays. Differences in proportions were assessed using Fisher's exact and Chi-Square tests as appropriate. For continuous variables, differences in medians between groups were assessed using Mann-Whitney U tests. Results: A total of 676 patients were included, with 399 patients (59%) comprising the off-business-hour group. No significant differences were seen in age, sex, ASPECTS score, or NIHSS at arrival. Off-business hours strokes had a longer delay between CSC arrival to groin puncture (minutes: 81 vs 44, P < .0001) and between imaging to groin puncture (minutes: 67 vs 32, P < .0001) compared to the business hours strokes. There were no differences in the rate of successful reperfusion (mTICI ≥2b) between groups (82% vs 83%, P = .61). At 90 days, 65% of patients in both groups had an mRS ≤2 (P = .91). Conclusion: Despite workflow delays in initiating EVT during off-business hours, there were no differences in the rate of successful reperfusion or functional outcomes.

Imaging Features of Immunodeficiency-Associated Primary CNS Lymphoma: A Review.

In the immunocompromised setting, there are distinct radiologic findings of primary central nervous system lymphoma (PCNSL), including necrotic ring-enhancing lesions, increased propensity for intralesional haemorrhage, and multiplicity. In this clinical context, advanced imaging with MR perfusion, spectroscopy, and diffusion-weighted imaging can be used to increase accuracy in the diagnosis of lymphoma over mimics such as high-grade glioma, metastases, or infection. This review summarizes the histology and pathophysiology of PCNSL in immunodeficient hosts, which provide a basis for its imaging appearances, prognosis, and treatment. This discussion is important for the general radiologist as the incidence of immunodeficiency-related PCNSL may be increasing.

Acute bilateral sensorineural hearing loss as presentation of leptomeningeal metastases.

We describe a rare occurrence of bilateral acute severe sensorineural hearing loss in a middle-aged man that heralded the diagnosis of metastatic gastric cancer.

Spontaneous intracranial hypotension.

Spontaneous intracranial hypotension (SIH) is a highly disabling but treatable secondary cause of headache. Recent progress in neuroradiological techniques has catalysed understanding of its pathophysiological basis and clinical diagnosis, and facilitated the development of more effective investigation and treatment methods. A UK-based specialist interest group recently produced the first multidisciplinary consensus guideline for the diagnosis and treatment of SIH. Here, we summarise a practical approach to its clinical and radiological diagnosis, symptomatic and non-targeted interventional treatment, radiological identification of leak site and targeted treatment of the leak once it has been localised.

External validation of the Brain Tumour Reporting and Data System (BT-RADS) in the multidisciplinary managementof post-treatment gliomas.

Purpose: To independently and externally validate the Brain Tumour Reporting and Data System (BT-RADS) for post-treatment gliomas and assess interobserver variability. Material and methods: In this retrospective observational study, consecutive MRIs of 100 post-treatment glioma patients were reviewed by two independent radiologists (RD1 and RD2) and assigned a BT-RADS score. Inter-observer agreement statistics were determined by kappa statistics. The BT-RADS-linked management recommendations per score were compared with the multidisciplinary meeting (MDM) decisions. Results: The overall agreement rate between RD1 and RD2 was 62.7% (κ = 0.67). The agreement rate between RD1 and consensus was 83.3% (κ = 0.85), while the agreement between RD2 and consensus was 69.3% (κ = 0.79). Among the radiologists, agreement was highest for score 2 and lowest for score 3b. There was a 97.9% agreement between BT-RADS-linked management recommendations and MDM decisions. Conclusions: BT-RADS scoring led to improved consistency, and standardised language in the structured MRI reporting of post-treatment brain tumours. It demonstrated good overall agreement among the reporting radiologists at both extremes; however, variation rates increased in the middle part of the spectrum. The interpretation categories linked to management decisions showed a near-perfect match with MDM decisions.

Added diagnostic yield of temporal artery magnetic resonance angiography in the evaluation of giant cell arteritis.

OBJECTIVES: Temporal artery magnetic resonance angiography (TAMRA) is a useful tool to investigate possible diagnoses of GCA. As acquired images also reveal other local structures, they may assist in finding alternative diagnoses when assessing for possible GCA. We sought to assess the utility of TAMRA in identifying other significant abnormalities either associated with a diagnosis of GCA or potentially mimicking a clinical presentation of GCA. METHODS: A retrospective cohort study was undertaken at St Joseph's Healthcare in Hamilton, Ontario, Canada between February 2007 and April 2020 and included patients who underwent TAMRA for a possible diagnosis of GCA. Patient demographics, diagnosis and imaging findings were extracted, and descriptive analysis of findings was performed. RESULTS: We included 340 individuals who underwent TAMRA for assessment of a potential diagnosis of GCA and had clinical information available; there were 126 (37.1%) diagnoses of GCA. Fourteen (4.1%) patients had findings on TAMRA that demonstrated an alternative diagnosis, findings were predominantly in the temporomandibular joint, orbit and meninges. Eighteen (14.3%) patients with GCA had intracranial vascular changes that were demonstrative of intracranial vasculitis; one stroke was attributed to intracranial GCA. CONCLUSIONS: TAMRA has proven utility in diagnosing GCA, and these data suggest that it also has utility in identifying alternative diagnoses to rule out the disease. Intracranial vasculitis was also seen in 14.3% of patients; the clinical impact of these findings is currently poorly understood and requires further study.

Brain age gap in neuromyelitis optica spectrum disorders and multiple sclerosis.

OBJECTIVE: To evaluate the clinical significance of deep learning-derived brain age prediction in neuromyelitis optica spectrum disorder (NMOSD) relative to relapsing-remitting multiple sclerosis (RRMS). METHODS: This cohort study used data retrospectively collected from 6 tertiary neurological centres in China between 2009 and 2018. In total, 199 patients with NMOSD and 200 patients with RRMS were studied alongside 269 healthy controls. Clinical follow-up was available in 85 patients with NMOSD and 124 patients with RRMS (mean duration NMOSD=5.8±1.9 (1.9-9.9) years, RRMS=5.2±1.7 (1.5-9.2) years). Deep learning was used to learn 'brain age' from MRI scans in the healthy controls and estimate the brain age gap (BAG) in patients. RESULTS: A significantly higher BAG was found in the NMOSD (5.4±8.2 years) and RRMS (13.0±14.7 years) groups compared with healthy controls. A higher baseline disability score and advanced brain volume loss were associated with increased BAG in both patient groups. A longer disease duration was associated with increased BAG in RRMS. BAG significantly predicted Expanded Disability Status Scale worsening in patients with NMOSD and RRMS. CONCLUSIONS: There is a clear BAG in NMOSD, although smaller than in RRMS. The BAG is a clinically relevant MRI marker in NMOSD and RRMS.