Introduction:Earlier intervention in glaucoma has been suggested to slow disease progression and preserve visual function and quality of life. Consequently, minimally invasive glaucoma surgery (MIGS) is increasingly used in mild-to-moderate glaucoma. Although numerous techniques and devices are available, their comparative efficacy remains debated: most procedures lower intraocular pressure to the mid-teens in primary open-angle glaucoma, but without clear evidence of superiority. As these surgeries are now offered to younger, otherwise healthy patients, safety has therefore become a central criterion in technique selection. Methods:This PRISMA-based systematic review analyzed safety outcomes from peer-reviewed studies of the main MIGS procedures published between 2014 and 2024. Five databases were searched using current and historical device names. Non-clinical studies, case series, and nonstandard techniques were excluded from quantitative analysis but retained qualitatively to capture rare events. A total of 401 studies, representing 39,381 eyes and 68,917 eye-years of follow-up, were included. Highest reported and weighted mean complication rates were calculated by procedure type. Results:Safety profiles varied. Trabecular bypass implants and ab interno canaloplasty were associated with low rates of serious adverse events and minimal anatomical disruption. Suprachoroidal devices carried higher risks of hypotony, inflammation, and malposition. Across all categories, chronic changes to angle anatomy and occasionnal re-interventions highlighted the potential for long-term sequelae, with endothelial cell loss emerging as a key concern for certain procedures.However, heterogeneity in definitions and reporting limited comparability. Common events such as hyphema and IOP spikes were inconsistently documented, while late complications like endothelial cell loss or peripheral anterior synechiae were often overlooked. This underreporting risks conflicting conclusions and undermines comparisons. Conclusion:For patients with mild, stable disease or those undergoing opportunistic combined surgery, tissue-sparing procedures with the lowest observed complication rates may be preferable. Standardized definitions, long-term follow-up, and harmonized safety reporting are becoming essential.
As AI systems are increasingly used to draft radiology reports, reliably evaluating their clinical quality remains a critical challenge. Large language model (LLM)-based metrics are now the best-correlated with radiologist judgment, yet they output a single opaque score that neither a clinician nor a model builder can easily interpret or audit. We introduce RadMatch, a multi-stage, LLM-based metric that decomposes report comparison into a structured finding-level matching with significance-aware scoring and error characterization across seven clinical attribute dimensions (status, location, severity, morphology, certainty, longitudinal comparison, and measurement). The main score is the actionable-error count, both interpretable and auditable. Candidate findings are graded correct, partial, or incorrect, and unmatched findings are counted as missed or hallucinated. Triage and actionable safety recall/precision and per-subset views add complementary, deployment-oriented lenses. Across two expert benchmarks, RadMatch is the most clinically aligned metric, matching inter-radiologist agreement on ReXVal and more than doubling the best prior metric on the harder RadEvalExpert. Relying only on few-shot prompting, it is designed to extend to other modalities and anatomies. We will release RadMatch as open-source code with an interactive dashboard for inspecting results.
Background Approximately 15 million people suffer from strokes each year and accurate prognostication is crucial for care management. This study evaluated the performance of the Stroke Prognosis in Intensive CarE (SPICE) score to predict functional outcomes at 1 year in patients who underwent mechanical ventilation following a stroke.Methods Ancillary analysis of the SPICE prospective multicentre cohort study conducted from 2017 to 2019 in 33 intensive care units (ICUs) of the Greater Paris area. All patients requiring mechanical ventilation on ICU admission and within 7 days of a non-traumatic stroke were included. The primary outcome was poor functional prognosis at 1 year (severe disability or death: modified Rankin Scale score of 4–6). Variables independently associated with the outcome were used to compute a prognostic score (SPICE score).Results Retained variables were stroke subtype, time between stroke diagnosis and initiation of mechanical ventilation, age, the Charlson Comorbidity Index and the Glasgow Coma Score (GCS) at ICU admission. The median SPICE score of the 364 patients of the cohort was 15 (IQR 11–19) and demonstrated acceptable predictive power (area under the receiver operating characteristic curve (AUROC) of 0.752) for poor outcome and outperformed GCS (AUROC 0.658) and Sequential Organ Failure Assessment (AUROC 0.655) scores. A threshold of >18 points on the SPICE score had a sensitivity of 35.7% and a specificity of 95% for poor outcome prediction, yielding a positive predictive value of 93.5% (95% CI 86.5% to 97.6%) and a positive likelihood ratio of 7.16 (95% CI 3.22 to 15.9).Conclusions For mechanically ventilated patients after a stroke, the SPICE score showed adequate predictive power for poor functional outcome at 1 year. Validation of the SPICE score in an external cohort is needed.
Introduction L’ataxie spinocérébelleuse SCA27B est une cause fréquente de syndrome cérébelleux de début tardif idiopathique. La triade est : âge de début ≥50 ans, ataxie épisodique puis chronique et nystagmus battant vers le bas. Objectifs Nous avons étudié l’apport de la vidéo-oculographie (VOG) dans le diagnostic de SCA27B et la réponse au traitement par 4-aminopyridique en vie réelle. Méthodes Nous avons rétrospectivement analysé les patients adressés pour suspicion d’ataxie en VOG, à la fondation Rothschild, en 2023, et sélectionné les patients ayant une ILOCA (idiopathic late onset cerebellar ataxia). Nous avons comparé sur le plan clinique, paraclinique, video-oculographique et radiologique les patients mutés (≥200 triplets) versus non mutés. Les patients mutés ont été traités par 4-aminopyridine et réévalués à 2 mois puis 8 mois cliniquement (échelle SARA) et en VOG. Résultats Sur 103 patients examinés en VOG pour suspicion d’ataxie, 35 patients avaient un ILOCA, parmi eux 34 ont été testé et 12 avaient une mutation SCA27B (35 %). L’âge moyen était de 68 ans, 50 % de femmes et score SARA médian 6. Le downbeat nystagmus (DBN) était fortement évocateur d’une mutation SCA27B (92 % des patients mutés versus 33 %, p=0,003), souvent associé à un gaze-evoked nystagmus GEN (75 % vs.19 %, p=0,005). L’IRM n’a pas retrouvé de différence significative entre les deux groupes. Sur les 9 patients traités, tous ont rapporté une amélioration de l’équilibre, corroboré par l’amélioration du score clinique SARA (médiane de 5,5 à 3) après 2 mois de traitement (p=0,015) et 8 mois. Discussion SCA27B représente un tiers de patients adressé pour ILOCA. Le DBN souvent associé à un GEN est très évocateur de cette mutation. Conclusion La VOG est un examen clef pour détecter le DBN qui est fortement associé à un SCA27B. Un traitement par 4-aminopyridine est efficace sur les symptômes.
Abstract Background and aims In acute stroke with large vessel occlusion (AIS-LVO), preventive strategies to mitigate hemorrhagic transformation (HT) following endovascular therapy (EVT) are needed. Inter-hospital transfer for EVT offers a unique therapeutic window during which preventive treatments could be administered, allowing sufficient exposure before EVT. We aimed to determine whether blood-brain barrier (BBB) disruption assessed prior to transfer is associated with 24-hour HT. Methods Multicenter retrospective study including AIS-LVO patients transferred from primary to comprehensive stroke centers for EVT assessment, with baseline MR-perfusion imaging performed before transfer. BBB disruption was quantified as the percentage signal change due to gadolinium leakage on perfusion source images. Mean permeability derangement (MPD) was defined as the average of voxels within the ischemic core exhibiting permeability values greater than a prespecified threshold. The primary outcome was any HT on 24-hour follow-up imaging. Results A total of 289 patients were included (median age 74 years, NIHSS 13, core volume 17 mL). HT occurred in 129 patients (45%), including HI1-2 in 24% and PH1-2 in 20%. Median MPD was higher in patients with HT than in those without (5.1% vs 3.0%; p<0.001). In multivariable analysis, elevated MPD was independently associated with HT after adjustment for core volume, glucose, occlusion site, and onset-to-imaging time (aOR 1.21; 95%CI 1.10-1.33; p<0.001). Similar associations were observed for parenchymal hematoma as a secondary outcome. Conclusions BBB disruption on pre-transfer perfusion MRI is strongly associated with 24-hour HT in AIS-LVO patients. Targeted interventions aimed at stabilizing the BBB during transfer warrant investigation in clinical trials. Conflict of interest All authors : nothing to disclose Figure 1 - belongs to Conclusions
