.
In 2022, Huscher et al. introduced a novel pancreaticojejunostomy (PJ) technique using an intraductal coronary artery stent as a structural fulcrum to enhance anastomotic stability in high-risk patients. Despite encouraging early results, the technique remained confined to a single center without external validation. The present study addresses this gap by evaluating its multicenter feasibility and reproducibility. This retrospective observational study included all consecutive patients undergoing PJ with intraductal stenting according to Huscher’s technique between 2022 and 2024 across three institutions. Only newly treated patients were included. Seventy-three patients underwent the open approach using a coronary artery stent, and seven underwent robotic PJ using a self-expanding metallic stent. Outcomes were analyzed descriptively and compared for contextual consistency with the original 2022 pilot series. Among 80 consecutive patients, 82.5% were classified as high risk for postoperative pancreatic fistula (POPF), and the remaining 17.5% as intermediate risk, according to a-FRS criteria. Clinically relevant POPF occurred in 5% of cases (4/80), with no mortality and only one late stent displacement. Stent positioning was systematically verified by CT on postoperative days 5 and 30. When compared with the original pilot series, outcomes were consistent, suggesting cross-institutional reproducibility. Robotic adaptation proved feasible, although limited by small sample size. This is the first multicenter experience validating Huscher’s intraductal-stent-based PJ technique. The open approach demonstrated consistent results across independent surgeons and centers, while the robotic application should be considered a feasibility finding. These results support further investigation through larger prospective studies.
Background: Neurofibromatosis type 1 (NF1) is an autosomal dominant genetic disorder. It primarily involves cutaneous, neurological and skeletal systems, with vascular involvement being relatively uncommon. Among the reported vascular manifestations, stenoses, aneurysms, and arteriovenous fistulas (AVFs) are infrequent but potentially serious complications. Their manifestation with cerebral steal syndrome is of even lower prevalence. Case Description: We report a case of a 52-year-old female with known NF1 who presented with progressive headache, transient episodes of right-sided weakness and speech problems. Total selective digital subtractive cerebral angiography revealed AVF of the left occipital region, primarily supplied by a large longitudinal defect in the V4 segment of the left vertebral artery, with additional feeders from the left occipital and ascending cervical arteries. A vertebrobasilar "steal" phenomenon was observed, with retrograde filling of the basilar circulation via the carotid system. Given the progressive symptoms and hemodynamic compromise, the patient underwent endovascular embolization using balloon catheters, detachable coils and Onyx 18 liquid embolic agent. The procedure was uneventful, with successful devascularization of the AVF and improved perfusion to the surrounding tissue. Post procedural recovery was favorable, with disappearing of existed symptoms. Conclusions: AVFs are rare in patients with NF1, and their pathogenesis is not fully understood. Steal syndrome occurs when high flow vascular malformations divert blood from surrounding brain parenchyma, leading to ischemic symptoms. Our case contributes to the limited literature on NF1-associated AVF and supports early neurovascular imaging in symptomatic individuals. Awareness of this rare vascular complication in NF1 is crucial for timely diagnosis and management.
Artificial intelligence (AI) continues to rapidly transform the practice of medicine, with clinicians increasingly adopting data-driven decision-making aids and diagnostic support tools. Orthopaedic physicians are well poised to harness the capabilities of AI, with an abundance of quantifiable imaging, biomechanical data, and structured clinical parameters lending themselves to algorithmic interpretation and automation. Namely, AI-augmented vision systems may increase the breadth of information readily available to clinicians, whereas smart exam rooms and automated clinical summaries may soon streamline clinical workflows to decrease administrative burden and allow more time for direct patient care. Personalised education materials and visual aids may improve patient understanding and compliance, with the aim of optimising patient outcomes. Generative medical and orthopaedic event models may soon alter decision-making heuristics and improve patient counselling. While the widespread adaptation of AI into clinical practices is not without limitations, physicians will likely come to share an increasingly symbiotic relationship with these platforms throughout their continued evolution. Accordingly, it is imperative that current and future orthopaedic practitioners become well-versed in harnessing the capabilities of AI and continue to identify new avenues for such technologies to benefit clinicians and patients alike. As such, the current manuscript provides a narrative review of the potential future applications of AI within orthopaedic practices by exploring current and developing technologies and detailing how the continued integration of AI-powered systems may serve to revolutionise the delivery of orthopaedic care. LEVEL OF EVIDENCE: Level V.
Objectif Étudier la possibilité d’identifier la tumeur primitive de métastases cérébrales en IRM grâce à l’intelligence artificielle. Méthodes 962 métastases ont été individualisées à partir des IRM cérébrales de 200 patients. Les tumeurs primitives ont été classées en 6 catégories. Une classification en 2 catégories, comprenant principalement les adénocarcinomes et les tumeurs non adénocarcinomateuses, a également été utilisée. Quatre séquences étaient disponibles pour chaque patient : T1, T1 avec injection de produit de contraste, T2 et FLAIR. Dans un premier temps, un réseau de neurones convolutif (CNN) bidimensionnel a été utilisé. Dans un second temps, des données radiomiques ont été extraites du volume entier de chaque métastase et utilisées pour entraîner le modèle par des techniques d’apprentissage automatique : régression logistique et random forests. Résultats Le CNN bidimensionnel n’a pas permis de classer les métastases selon les catégories de tumeurs primitives. La régression logistique, quant à elle, a montré une précision de 0,5774 et un score F1 de 0,6120 pour la tâche de classification binaire. Les random forests ont quant à elles affiché une précision de 0,6012 et un score F1 de 0,4925 pour la tâche de classification à six classes, et une précision de 0,7024 et un score F1 de 0,7283 pour la tâche de classification binaire. Conclusion Ces résultats tendent à démontrer qu'il est possible de différencier les métastases d'adénocarcinome de celles de tumeurs non-adénocarcinomateuses. Limites L'utilisation de réseaux de neurones convolutifs 3D (CNN 3D) n'a pas été possible en raison de ressources de calcul limitées. Les analyses anatomopathologiques détaillées n'étaient pas disponibles. Dans le cadre de la classification binaire, il est probable que certaines métastases de tumeurs non-adénocarcinomateuses aient été incluses dans la classe des adénocarcinomes. Les performances de nos modèles auraient probablement été meilleures avec le type histologique précis.
Portal vein aneurysm (PVA) is a rare vascular condition, particularly when located in the retropancreatic space, where it may mimic pancreatic pathology or cause compressive symptoms. We report the case of a 69-year-old woman presenting with chronic postprandial pain and severe weight loss due to a 37 mm extrahepatic PVA compressing the neck of the pancreas and the main pancreatic duct. After failure of conservative management consisting of clinical observation and analgesic therapy, the patient underwent a totally laparoscopic central pancreatectomy, complete aneurysm excision, and portal vein reconstruction using a bovine pericardial patch. All steps, including vascular repair and pancreatojejunostomy with intraductal stenting, were performed intracorporeally. Postoperative recovery was uneventful, and long-term follow-up with contrast-enhanced 3D imaging confirmed vascular patency and complete symptom resolution. To our knowledge, this is the first video-documented case of a laparoscopic central pancreatectomy for retropancreatic PVA with biological venous reconstruction, highlighting a feasible and safe minimally invasive approach to complex vascular-pancreatic pathology.
