The correction carries its own digital object identifier, 10.1007/s40846-026-01048-4, permanently anchoring the notice to the scholarly record, while the underlying research remains citable at 10.1007/s40846-024-00922-3 as volume 44, pages 799 to 808, of the journal. Springer’s version of record for the correction is dated 27 August 2026, and the document participates in Crossmark, the cross-publisher initiative that flags readers whenever a paper they are viewing has been updated. What the notice does not do is explain how the error arose. It does not say whether the wrong image file was uploaded during production, whether a panel was mislabeled, or whether the mistake was caught by the authors, a reader or the editorial office. It simply presents the correct figure and confirms that the original article has been corrected. Typically rendered as a ranked bar chart, the figure shows at a glance which variables the model leans on most — precisely why its accuracy matters.

Behind the notice stands a research team that spans two complementary sides of the neurovascular problem. Mariyam Argymbay, Shams Khan, Noman Ahmad and Yasin Mamatjan are based in the Faculty of Science at Thompson Rivers University in Kamloops, British Columbia, with Mamatjan serving as corresponding author. Mira Salih is affiliated with the Brain Aneurysm Institute at Harvard Medical School and Beth Israel Deaconess Medical Center in Boston, a clinical environment devoted to the vascular pathologies that can precipitate devastating brain events. The pairing is telling. Stroke risk assessment is not purely a software exercise; it demands fluency in the epidemiology of hypertension, atrial fibrillation, diabetes and the other conditions that precede cerebrovascular accidents, and it demands a sense of how probabilistic information lands on an actual patient. A collaboration that joins a Canadian computing and biomedical engineering group with a Harvard-affiliated aneurysm research institute is exactly the kind of coalition this problem tends to attract.

The system the team describes is, at its core, a machine-learning pipeline wearing two hats. The first hat is predictive. Like clinical risk models before it, a recommender system for stroke risk assessment ingests patient variables — the kinds of features that dominate stroke epidemiology, such as age, blood pressure, diabetes status, cardiac rhythm abnormalities, smoking history and prior vascular events — and produces an estimate of an individual’s probability of stroke. Systems of this type are usually validated retrospectively, trained and tested on recorded patient data with performance summarized by standard metrics, before anyone contemplates prospective use. The second hat is prescriptive. Where classical risk scores stop at a number, a recommender maps that number onto actions: which screenings, interventions or lifestyle changes are most relevant for a person at a given level of risk. In engineering terms, the recommendation layer is a decision-support component that converts a calibrated probability into prioritized, personalized guidance — conceptually closer to how streaming platforms convert viewing histories into watchlists, except the stakes are measured in neurons rather than evenings.

The Strokebot is the conversational face of that machinery — a chatbot integrated directly into the risk-assessment workflow rather than bolted on afterward. Health chatbots of this kind typically conduct structured dialogue to gather or confirm risk-relevant information, translate an abstract risk score into plain language, answer follow-up questions and steer users toward appropriate care, including education about the sudden facial drooping, arm weakness and speech difficulty that mark stroke’s warning signs. The design logic is friction reduction. A risk model locked behind a dashboard helps experts; a risk model that talks helps everyone else. Integration also matters for data flow, because a conversational agent that feeds the underlying recommender can, in principle, keep the model’s inputs current and its recommendations aligned with what the user has actually been told. No credible chatbot claims diagnostic authority; the goal is triage and engagement rather than replacement of physicians, and responsible implementations keep a human clinician firmly in the loop.

The corrected Figure 3 concerns the system’s third role, and arguably its most important one: self-explanation. SHAP — SHapley Additive exPlanations — imports a concept from cooperative game theory devised by economist Lloyd Shapley in the 1950s, work later honored with a Nobel Memorial Prize. Shapley’s question was how to divide a game’s payout fairly among players whose contributions differ. SHAP recasts a machine-learning prediction as exactly that game: each input feature is a player, the prediction is the payout, and a feature’s Shapley value is its average marginal contribution to the prediction, computed across all possible orderings of the players. The result is additive and locally faithful — the prediction equals a baseline value plus the sum of every feature’s contribution — which is why SHAP has become one of the most widely used tools for opening up otherwise opaque models such as gradient-boosted tree ensembles and neural networks. Exact Shapley computation grows combinatorially with feature count, so practical implementations rely on model-structure shortcuts and careful sampling to make the arithmetic tractable at real-world scale.

When Shapley values are computed for every individual in a dataset, their absolute magnitudes can be averaged into a single global picture of what the model relies on most. That averaged, ranked summary is what Figure 3 presents: a SHAP-based global risk factor ranking showing which inputs the stroke model weights most heavily across the population it learned from. For clinicians, such a chart functions as a contract. If the model promotes a biologically implausible factor to the top, or buries blood pressure beneath noise variables, the discrepancy is a red flag visible before the system ever reaches a patient. If the ranking instead tracks established stroke epidemiology, it builds confidence that the algorithm has learned medicine rather than artifacts. This is why an incorrectly published ranking figure is not a cosmetic problem. It is a misdelivery of the model’s most consequential self-description, read by anyone skimming the paper for the one picture that summarizes a thousand lines of code.

The timeline is also instructive. Roughly twenty months separate the original publication in December 2024 from the correction in August 2026, an interval that reflects the ordinary rhythms of post-publication scrutiny rather than scandal. Corrections are among the most common documents in scientific publishing, and the infrastructure surrounding them — persistent identifiers, Crossmark badges, version-of-record timestamps — exists precisely so that an updated figure can supersede a faulty one without erasing the historical trail. The original article’s page now leads readers to the corrected version, preserving the citation trail while ensuring the fixed figure is what most visitors encounter. The alternative, silently swapping an image inside a published paper, would corrode the very trust that identifiers and archives are built to protect. In fast-moving fields where machine-learning health papers accumulate citations quickly, a DOI-anchored correction ensures that anyone citing, reproducing or deploying the work meets the amended version first. The machinery worked as designed: slowly, visibly and on the record.

The broader stakes are difficult to overstate. Stroke remains one of the world’s leading causes of death and long-term disability, and widely cited global estimates put new cases at well over ten million each year, with projections suggesting the burden will climb as populations age. The encouraging corollary, reinforced by decades of epidemiological research, is that the large majority of stroke risk is tied to detectable, modifiable factors — with elevated blood pressure consistently emerging as the single most powerful one — which is why tools that can find at-risk individuals early and talk them toward prevention hold such appeal for strained health systems. Global prevention campaigns have drilled the same message for years: control hypertension, treat atrial fibrillation with anticoagulation where indicated, manage diabetes and cholesterol, quit smoking, keep moving. An explainable model that reproduces those priorities and personalizes them to an individual’s profile could extend their reach. But deployment hinges on credibility, and credibility requires that the model’s published explanation be exactly what its authors intended.

Figure 3 now reads as its authors intended, and a correction notice of a few hundred words has quietly done its job. The episode is a useful reminder that in medical artificial intelligence, the explanation is part of the intervention. A Strokebot can only be as trustworthy as the risk model beneath it, and the risk model can only be as trustworthy as the published evidence of how it weighs the world. When that evidence appears in error, the whole chain of trust wobbles; when it is corrected, one link at a time and on the record, the chain holds. Science’s smallest genre, the erratum, rarely goes viral. But it is where the discipline does its most honest bookkeeping — and in this case, it is where a machine’s account of stroke risk was set right.

Takayasu arteritis, often called “pulseless disease,” primarily affects the aorta and its major branches. Inflammation can thicken the arterial wall, reduce the diameter of the vessel, and eventually restrict blood flow to the brain, arms, kidneys, or other organs. Neurological complications may include transient ischemic attacks, strokes, dizziness, visual disturbances, and fainting. Because symptoms and laboratory markers do not always reflect the full extent of vascular injury, clinicians increasingly rely on imaging to identify active inflammation and structural narrowing. The new work explores an unusual imaging target: the amygdala, a small almond-shaped structure located in the medial temporal lobe and involved in emotional processing, stress responses, autonomic regulation, and communication with the immune system.

The investigators analyzed data from 303 people with Takayasu arteritis who underwent whole-body ¹⁸F-fluorodeoxyglucose positron emission tomography/computed tomography, commonly known as ¹⁸F-FDG PET/CT. The radioactive glucose analogue is taken up by metabolically active cells, allowing PET to visualize tissues with increased glucose consumption. In large-vessel vasculitis, inflammatory cells in the arterial wall can accumulate FDG and produce a measurable signal. The researchers also quantified FDG uptake in the amygdala and bone marrow, as well as in affected vessel walls, while collecting clinical information, blood-test results, and vascular imaging findings. Participants were followed for a median of 27 months, during which cerebrovascular and other adverse events were recorded.

The principal result was not uniform across the entire cohort. When all participants were analyzed together, amygdalar standardized uptake values, or SUVs, were not significantly associated with cerebrovascular events. SUV is a semi-quantitative measure that estimates how much tracer has accumulated in a region after accounting for factors such as injected dose and body size. SUVmax represents the highest measured activity within a region of interest, whereas SUVmean reflects the average activity. This distinction matters because a single intense voxel can influence SUVmax, while SUVmean may provide a broader estimate of regional metabolic activity. In the overall study population, neither measurement consistently separated patients who experienced cerebrovascular events from those who remained event-free.

A clearer pattern emerged in the treatment-naïve subgroup. Patients who had suffered cerebrovascular events showed lower amygdalar activity than those without such events. Mean amygdalar SUVmax was 9.3 compared with 10.3 in event-free patients, while mean SUVmean was 6.6 compared with 7.4. The differences were statistically significant, with p values of 0.011 and 0.003, respectively. When the researchers divided patients according to amygdalar metabolic activity, 24.3 percent of people in the low-SUV group had experienced cerebrovascular events, compared with 15.9 percent in the higher-SUV group. The low-activity group also had higher immunoglobulin G and immunoglobulin A levels and lower lymphocyte counts, suggesting that reduced amygdalar uptake may coexist with distinctive systemic immune features.

The relationship became especially notable when the researchers examined structural disease in the arteries supplying the head and neck. Higher amygdalar SUVmax was identified as an independent protective factor against combined carotid and vertebral artery stenosis. The reported odds ratio was 0.876, with a p value of 0.032. An odds ratio below one indicates that, within the statistical model, increasing amygdalar activity was associated with lower odds of the outcome after accounting for other evaluated factors. The carotid arteries deliver blood to much of the brain’s anterior circulation, while the vertebral arteries contribute to the posterior circulation. Narrowing in both systems can substantially reduce cerebral blood flow and increase the risk of ischemic injury.

Follow-up findings provided additional support for the signal, although they also illustrated the complexity of the biology. Patients who later experienced cerebrovascular events had a significantly lower amygdalar SUVmax than a group described as having new-onset symptoms without the same event outcome: 8.2 compared with 10.4. This observation raises the possibility that amygdalar metabolic activity could reflect a brain-body state associated with vascular vulnerability before or during clinically important disease. However, PET uptake is not a direct measurement of stress, emotion, or immune control. It can be influenced by age, medication, glucose levels, scanner characteristics, image-processing methods, brain structure, and other medical conditions. The amygdala is also small, making accurate measurement vulnerable to partial-volume effects, in which limited spatial resolution causes activity from neighboring tissues to blend into the region of interest.

The authors’ interpretation builds on a growing body of research concerning the brain’s role in cardiovascular and immune regulation. Earlier studies in other populations have linked resting amygdalar activity with cardiovascular events, while experimental work has shown that stress-related neural circuits can influence the hypothalamic-pituitary-adrenal axis, sympathetic nervous system, bone marrow activity, and inflammatory signaling. The amygdala communicates with regions that regulate autonomic output and endocrine responses, and these pathways can affect circulating immune cells and the behavior of inflammatory tissues. In Takayasu arteritis, such neuroimmune interactions could theoretically alter the inflammatory environment surrounding the aorta and its branches. The present study does not prove this mechanism, but it adds a new imaging-based association to the emerging concept that vascular inflammation may be shaped by both immune processes and neural activity.

The findings should therefore be viewed as a potential biomarker discovery rather than a clinical test ready for routine use. The study was observational, and its results cannot establish whether reduced amygdalar activity contributes to arterial stenosis, results from chronic vascular disease, or reflects another factor shared by patients with worse outcomes. The absence of a significant association in the full cohort also suggests that treatment exposure and disease history may modify the relationship. In addition, the reported associations came from a single clinical cohort and require confirmation in independent populations using standardized PET acquisition and analysis. Future studies could combine serial brain PET, vascular imaging, inflammatory biomarkers, autonomic measurements, psychological assessments, and long-term clinical follow-up. If the association is reproduced, amygdalar metabolism might eventually help identify patients who need closer neurological surveillance, more detailed carotid and vertebral imaging, or intensified prevention strategies.

For now, the study offers a striking shift in perspective on Takayasu arteritis. The disease is traditionally assessed through arterial anatomy, blood-flow measurements, laboratory inflammation markers, and metabolic activity within the vessel wall. The new results suggest that the brain itself may contain information about the risk of vascular complications. A low amygdalar PET signal cannot yet predict an individual stroke, and it should not replace established clinical evaluation. Nevertheless, the work highlights how a scan originally used to map glucose metabolism can reveal connections between emotional-neural circuitry, systemic immunity, and arterial injury. As researchers continue to decode these pathways, the amygdala may become an important part of the story of how large-vessel inflammation affects the whole body.

Subject of Research: Takayasu arteritis, amygdalar metabolism, cerebrovascular events, and carotid-vertebral artery stenosis

Article Title: Amygdalar metabolic activity associated with cerebrovascular events and carotid-vertebral artery stenosis in takayasu arteritis

Article References: Ma L, Wu B, Wu S, et al. “Amygdalar metabolic activity associated with cerebrovascular events and carotid-vertebral artery stenosis in takayasu arteritis.” European Journal of Nuclear Medicine and Molecular Imaging (2026). References include Tawakol A, Ishai A, Takx RA, et al. “Relation between resting amygdalar activity and cardiovascular events: a longitudinal and cohort study.” The Lancet. 2017;389:834–845.

Image Credits: AI Generated

DOI: 10.1007/s00259-026-08090-z

Keywords: Takayasu arteritis; amygdala; ¹⁸F-FDG PET/CT; cerebrovascular events; carotid artery stenosis; vertebral artery stenosis; neuroimmune interaction; vascular inflammation; brain metabolism; nuclear medicine imaging

Patients classified as having mild carotid stenosis—less than 50% narrowing—have conventionally been considered low risk for recurrent ischemic events and thus managed conservatively with medical therapy alone. Yet clinical experience reveals a troubling paradox: despite optimal best medical therapy (BMT), a considerable subset of these patients continue to suffer strokes or transient ischemic episodes, signaling that traditional metrics may inadequately capture true vulnerability. This perplexing clinical scenario has galvanized researchers at Toyama University in Japan to investigate novel predictors of stroke risk beyond luminal stenosis severity.

Led by Lecturer Dr. Daina Kashiwazaki and neurosurgeon Dr. Satoshi Kuroda, the “Mild but Unstable Stenosis of Internal Carotid Artery” (MUSIC) study enrolled 124 patients across multiple centers who had suffered cerebrovascular or retinal ischemic events ipsilateral to mild carotid stenosis. The study, published in the prestigious Journal of Neurosurgery in early 2025, set out to rigorously characterize the clinical features, detailed radiological plaque characteristics, and longitudinal treatment outcomes in this underappreciated patient cohort. Importantly, all participants received best medical therapy tailored to their risk factors, while approximately half also underwent surgical intervention via carotid endarterectomy (CEA) or carotid artery stenting (CAS).

What sets the MUSIC study apart is its intensive focus on plaque composition, rather than solely the degree of arterial narrowing, as a determinant of future ischemic events. Advanced neuroimaging techniques identified a strikingly high prevalence of unstable plaque features in the cohort: around 81% exhibited radiologically unstable plaques, and nearly 60% harbored intraplaque hemorrhage (IPH)—a pathologic hallmark indicative of plaque vulnerability and active disease. IPH, characterized by bleeding within the atherosclerotic lesion, is known to accelerate plaque progression, increase inflammation, and dramatically elevate the risk of rupture and thrombosis.

The prognostic implications of these findings were profound. Patients with IPH faced a significantly elevated risk of ipsilateral ischemic stroke and other cerebrovascular endpoints, including ocular ischemic symptoms and progression necessitating surgical intervention. Critically, the incidence of ischemic stroke over two years was substantially lower—just 1.7%—in patients who received surgical plaque removal via CEA compared to 15.1% in those managed with best medical therapy alone. These data unequivocally position plaque instability and hemorrhagic components as pivotal risk modulators, overshadowing luminal stenosis percentage in symptomatic mild carotid disease.

Such insights disrupt existing clinical guidelines that traditionally reserve CEA for patients with moderate to severe stenosis, generally excluding those with mild disease. The MUSIC study’s findings advocate for a paradigm shift towards personalized stroke prevention strategies that incorporate detailed plaque imaging to stratify risk and guide treatment. By identifying high-risk patients with unstable plaque features early, clinicians could intervene surgically before disabling ischemic events occur. This more nuanced approach holds the promise to transform prophylactic stroke care in millions worldwide.

Another remarkable facet highlighted by the study is the observation that nearly half of the participants were already receiving antithrombotic agents prior to enrollment but continued to experience ischemic events. This suggests that conservative medical management alone may be insufficient, or that therapeutic resistance may exist in this subgroup with unstable mild carotid stenosis. The identification of such resistance underscores the urgent need for alternative strategies, including surgical plaque removal, to avert catastrophic outcomes.

From a pathophysiological perspective, the study reinforces the concept that the composition and biological activity of atherosclerotic plaques—rather than static luminal narrowing—drive the natural history of ischemic stroke risk. Intraplaque hemorrhage triggers a vicious cycle of inflammation, neovascular proliferation, and plaque destabilization, which culminate in embolic strokes even in arterial segments with less than 50% stenosis. These dynamic plaque processes cannot be captured by traditional angiographic stenosis assessment, necessitating advanced imaging modalities such as high-resolution magnetic resonance imaging (MRI) for adequate risk appraisal.

The potential clinical repercussions of adopting plaque composition evaluation are sweeping. Preoperative assessment of IPH and other high-risk plaque morphologies would enable neurosurgeons and neurologists to tailor individualized treatment plans, selectively recommending carotid endarterectomy for those who stand to benefit most while sparing low-risk patients from unnecessary surgery. This approach promises to enhance patient safety, improve functional outcomes, and optimize resource allocation in vascular neurosurgery.

Looking ahead, the MUSIC investigators envision integrating plaque characterization into routine clinical algorithms for all individuals presenting with symptomatic mild carotid stenosis. This forward-thinking strategy departs from the traditional stenosis-threshold model and advocates for a more sophisticated, biology-driven framework that addresses the heterogeneity of atherosclerotic disease. Through wider adoption, such paradigm shifts could ultimately reduce the global burden of stroke by intercepting high-risk patients before irreversible neurological damage occurs.

While the study’s prospective design and multicenter enrollment strengthen its validity, further research is warranted to corroborate these findings across diverse populations and to refine imaging protocols and treatment thresholds. Moreover, investigating adjunctive pharmacotherapies aimed specifically at stabilizing vulnerable plaques represents a promising avenue to complement surgical interventions. Ultimately, multidisciplinary collaboration combining advanced neuroimaging, vascular neurosurgery, and precision medicine will be vital to translating these insights into clinical practice.

In conclusion, the MUSIC study from Toyama University decisively challenges the prevailing clinical mindset that views symptomatic mild carotid stenosis as inherently low risk. By illuminating the critical role of plaque instability and intraplaque hemorrhage in precipitating recurrent ischemic events, this research heralds a new era in stroke prevention. The shift towards incorporating plaque composition evaluation into patient assessment heralds a future where individualized, mechanism-based interventions supplant crude stenosis metrics, offering new hope for patients previously deemed low risk yet suffering devastating strokes. This transformative work underscores the pressing necessity for personalized approaches in vascular neurology and neurosurgery and exemplifies the impact of targeted research in improving global health outcomes.

—

Subject of Research: People

Article Title: Clinical features, radiological findings, and outcome in patients with symptomatic mild carotid stenosis: a MUSIC study

News Publication Date: February 21, 2025

Web References: https://doi.org/10.3171/2024.10.JNS241185

References: Kashiwazaki D, Kuroda S, et al. Clinical features, radiological findings, and outcome in patients with symptomatic mild carotid stenosis: a MUSIC study. Journal of Neurosurgery. Published online February 21, 2025. DOI: 10.3171/2024.10.JNS241185

Image Credits: Lecturer Daina Kashiwazaki from Toyama University, Japan

Keywords: Carotid artery, Medical treatments, Risk factors, Ischemia, Atherosclerosis, Bleeding