{"title":"[The key role of standardized structured reporting in pathology. Producing high-quality data today and preparing tomorrow's digital pathology].","authors":"Jean-Pierre Bellocq, Dominique Fétique","doi":"10.1016/j.annpat.2025.08.004","DOIUrl":"https://doi.org/10.1016/j.annpat.2025.08.004","url":null,"abstract":"<p><p>The pathology report is a critical source of medical information. It must evolve from a traditionally narrative format, regardless of its qualitative value, toward a communication medium based on standardised and structured data. Already useful in current practice, the standardised structured reporting (SSR) will become a cornerstone of digital pathology, in synergy with digital imaging and artificial intelligence. The reliability of the SSR and the precision of the data it provides, correlated with multimodal sources of clinical, radiological, or biological information, will contribute to the foundations of highly performant digital solutions. Despite its considerable potential, the SSR remains underutilised. Although the concept was favourably received by the specialty twenty years ago, the SSR has often been described as insufficiently scalable, poorly ergonomic, and a source of time loss in daily practice. Once challenged by the capacity of natural language processing (NLP) to structure retrospectively narrative reports, the SSR ultimately emerges as the most reliable data source for effective patient care and high-performing research. New tools developed within the framework of the national Impulsioninitiative should help overcome these reluctances and open concrete perspectives for pathologists from now on.</p>","PeriodicalId":50969,"journal":{"name":"Annales De Pathologie","volume":" ","pages":""},"PeriodicalIF":0.5,"publicationDate":"2025-10-07","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145253735","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Arnaud Uguen, Marion Le Rochais, Glen Le Flahec, Amelie Bourhis, Anabelle Remoué, Claire Bocciarelli, Virginie Conan-Charlet, Isabelle Quintin-Roué, Laurent Doucet, Pascale Marcorelles
{"title":"[Implementation of free and in-house artificial intelligence tools: Experience and examples in a digital pathology workflow].","authors":"Arnaud Uguen, Marion Le Rochais, Glen Le Flahec, Amelie Bourhis, Anabelle Remoué, Claire Bocciarelli, Virginie Conan-Charlet, Isabelle Quintin-Roué, Laurent Doucet, Pascale Marcorelles","doi":"10.1016/j.annpat.2025.08.005","DOIUrl":"https://doi.org/10.1016/j.annpat.2025.08.005","url":null,"abstract":"<p><p>Digital pathology and microscopic image analysis using artificial intelligence (AI) will revolutionize practice and diagnosis in pathology. The implementation and use of AI models in clinical workflows within digital pathology raise numerous questions regarding the role of these new tools. Alongside commercial software solutions, the use of free software and \"in-house\" models can be an attractive strategy, promoting the adoption of these new technologies by pathologists while addressing their diagnostic needs without constraints related to data flow or funding. The appropriate use of the right diagnostic model - subject to the continuous scrutiny of the pathologist in charge - is key to the integration of AI tools into clinical practice. This article presents an experience in implementing free and \"in-house\" AI tools based on the use of the QuPath software and its extensions within a digital pathology workflow.</p>","PeriodicalId":50969,"journal":{"name":"Annales De Pathologie","volume":" ","pages":""},"PeriodicalIF":0.5,"publicationDate":"2025-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145139379","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"[When the neuroendocrine phenotype is misleading: A hepatic tumor with a deceptive profile].","authors":"Nissrine Berry, Philipe Baltzinger, Aline Baltrès, Marie-Pierre Chenard, Antonin Fattori","doi":"10.1016/j.annpat.2025.09.001","DOIUrl":"https://doi.org/10.1016/j.annpat.2025.09.001","url":null,"abstract":"<p><p>We report the case of a 45-year-old woman in whom a solitary 5.5cm hepatic tumor was discovered during oncologic surveillance for a papillary thyroid carcinoma diagnosed ten years earlier. Biopsy revealed a tumor cell proliferation with \"endocrinoid\" morphology and convincing immunohistochemical expression of neuroendocrine markers, initially suggesting a well-differentiated grade 3 neuroendocrine tumor. FDG-PET/CT demonstrated isolated hypermetabolic activity in the liver lesion, with no corresponding uptake on DOTATOC-PET. Following neoadjuvant chemotherapy, the patient underwent segmental liver resection. Histopathological examination of the resected specimen showed a proliferation of monomorphic cells with ovoid nuclei, arranged in a tubulo-solid architecture, with focal areas reminiscent of a \"thyroid-like\" pattern. Tumor cells exhibited heterogeneous expression of neuroendocrine markers and strong, diffuse positivity for alpha-inhibin. RNA sequencing identified a NIPBL::NACC1 fusion transcript, leading to a revised diagnosis of hepatic carcinoma with NIPBL::NACC1 fusion. This recently described and rare hepatic tumor is challenging to diagnose on biopsy. Histologically, it is characterized by a monomorphic ovoid cell proliferation with a tubulo-solid growth pattern and focal thyroid-like morphology. Neuroendocrine marker expression is variable, but strong and diffuse alpha-inhibin staining is a consistent feature.</p>","PeriodicalId":50969,"journal":{"name":"Annales De Pathologie","volume":" ","pages":""},"PeriodicalIF":0.5,"publicationDate":"2025-09-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145139369","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Cas no 6. Infection placentaire à cytomégalovirus","authors":"Marie-Hélène Saint-Frison","doi":"10.1016/j.annpat.2025.01.004","DOIUrl":"10.1016/j.annpat.2025.01.004","url":null,"abstract":"","PeriodicalId":50969,"journal":{"name":"Annales De Pathologie","volume":"45 5","pages":"Pages 417-419"},"PeriodicalIF":0.5,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143069379","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
I. Hadhri, A. Bdioui, S. Moussa, A. Mahmoudi, Z. Eleuch, S. Mestiri, S. Hmissa
{"title":"Application de l’Intelligence Artificielle dans les pays à ressources limitées : exemple du laboratoire d’anatomie et de cytologie pathologiques de Sahloul, Sousse, Tunisie","authors":"I. Hadhri, A. Bdioui, S. Moussa, A. Mahmoudi, Z. Eleuch, S. Mestiri, S. Hmissa","doi":"10.1016/j.annpat.2025.06.012","DOIUrl":"10.1016/j.annpat.2025.06.012","url":null,"abstract":"<div><h3>Introduction</h3><div>Les avancées parallèles de la médecine et de l’informatique ont introduit des techniques révolutionnaires dans la localisation et le traitement des pathologies graves. Cependant, l’évaluation du pronostic reste une tâche difficile malgré la précision des équipements médicaux. La solution qui se dessine de plus en plus est l’implémentation d’une alternative informatique, à savoir l’intelligence artificielle, pour la stadification du cancer et l’évaluation du pronostic. Ce travail a ainsi comme objectif de mettre en évidence l’apport de l’intelligence artificielle dans l’anatomie pathologique.</div></div><div><h3>Matériels et méthodes</h3><div>Les données ont été collectées et traitées au département de pathologie de l’hôpital Sahloul, à Sousse, en Tunisie. À partir de lames de tumeurs neuroendocrines pancréatiques, de mélanomes uvéaux et de carcinomes colorectaux, nous avons pris des photos microscopiques : 1500 à 3000 photos de chaque type de tumeur à fort grossissement. La réalisation de notre modèle s’est déroulée en quatre étapes : traitement des données, création, entraînement, et évaluation du modèle.</div></div><div><h3>Résultats</h3><div>À travers une interface, l’application doit accomplir la tâche du médecin : calculer la moyenne des 10 plus grands nucléoles pour les cas de mélanome uvéal, estimer le ki67 pour les tumeurs neuroendocrines et évaluer le pourcentage de lymphocytes dans le stroma du carcinome colorectal. Un système de gan (réseaux génératifs antagonistes) a été utilisé. Ils présentent une classe de réseaux de neurones développée par Ian Goodfellow en 2014. Ils consistent en deux réseaux concurrents : un générateur et un discriminateur. Le générateur crée des échantillons de données (par exemple, des images) à partir de bruit aléatoire, tandis que le discriminateur tente de distinguer entre les échantillons réels et ceux générés. Ces réseaux sont entraînés simultanément : le générateur essaie de tromper le discriminateur, et le discriminateur essaie de devenir meilleur à détecter les faux échantillons. Ce jeu en boucle permet au générateur de produire des données de plus en plus réalistes. Les résultats obtenus ont montré une précision très proche de 1, surpassant les autres outils existants. Les résultats de ce travail peuvent servir de base à des recherches ultérieures, telles que l’application des étapes de traitement et des algorithmes de traitement d’images à d’autres types de tumeurs. De plus, le modèle peut être entraîné avec des données provenant d’autres tumeurs.</div></div><div><h3>Conclusion</h3><div>L’intelligence artificielle s’est révélée être un outil très utile et facilitateur dans ce contexte. Elle nous a permis de calculer avec précision un paramètre important et d’améliorer la prise en charge des patients. L’entraînement des méthodes d’IA et la validation des modèles d’IA à l’aide de vastes ensembles de données avant d’appliquer ces méthodes aux données personnelles peuvent ré","PeriodicalId":50969,"journal":{"name":"Annales De Pathologie","volume":"45 5","pages":"Page 458"},"PeriodicalIF":0.5,"publicationDate":"2025-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145050292","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号
