{"id":15649,"date":"2021-12-14T07:00:26","date_gmt":"2021-12-14T06:00:26","guid":{"rendered":"https:\/\/convergences.online\/hemato\/?p=15649"},"modified":"2024-12-11T14:22:04","modified_gmt":"2024-12-11T13:22:04","slug":"reconnaissance-des-cellules-de-la-moelle-osseuse-par-apprentissage-automatique","status":"publish","type":"post","link":"https:\/\/www.hematostat.net\/en\/reconnaissance-des-cellules-de-la-moelle-osseuse-par-apprentissage-automatique\/","title":{"rendered":"Reconnaissance des cellules de la moelle osseuse par apprentissage automatique"},"content":{"rendered":"

R\u00e9f. : HematoStat.net ; 2 (4) : R34<\/p>\n<\/p>\n

Matek C et al. Highly accurate differentiation of bone marrow cell morphologies using deep neural networks on a large image data set. Blood. 18 nov 2021;138(20):1917<\/em>\u2011<\/em>27.<\/em><\/p>\n

<\/em> <\/em><\/p>\n

R\u00e9sum\u00e9 de l’article<\/strong><\/p>\n

Les auteurs ont eu recours \u00e0 171 374 images cytologiques de frottis de moelle osseuse num\u00e9ris\u00e9es qui concernaient 945 patients suivis entre 2011 et 2013. Les images \u00e9taient annot\u00e9es par des cytologistes experts. Cette s\u00e9rie \u00e9tait ensuite utilis\u00e9e dans un r\u00e9seau dit de \u00ab neurones \u00e0 convolution \u00bb (technique d\u2019apprentissage automatique permettant de diff\u00e9rencier les images) afin d\u2019am\u00e9liorer l\u2019algorithme. Une validation externe \u00e9tait r\u00e9alis\u00e9e sur 627 images de cellules provenant de 10 patients.<\/p>\n

Dans nos pratiques<\/strong><\/p>\n

Encore largement perfectible, le mod\u00e8le pr\u00e9sent\u00e9 ici illustre tout le potentiel des techniques d\u2019apprentissage automatique pour la reconnaissance cellulaire en h\u00e9matologie, y compris dans des situations complexes (frottis m\u00e9dullaire). De l\u2019imagerie diagnostique \u00e0 l\u2019anatomopathologie, ces techniques ouvrent de nombreuses perspectives et constitueront une aide ind\u00e9niable pour am\u00e9liorer la qualit\u00e9 et la prise en charge des patients.<\/p>\n

Le regard du statisticien<\/strong><\/p>\n

Ce qu\u2019on appelle le \u2018deep learning\u2019 <\/em>aujourd\u2019hui est de plus en plus utilis\u00e9 dans la reconnaissance d\u2019images, ici le proc\u00e9d\u00e9 de \u2018neural network\u2019 <\/em>accompagn\u00e9 de cross-validations <\/em>et validation externe qui servent \u00e0 mesurer la performance de la technique pour la reconnaissance de types de cellules. Il existe d\u2019autres outils comme la morphom\u00e9trie g\u00e9om\u00e9trique qui pourrait \u00eatre une autre piste dans l\u2019analyse de formes.<\/p>\n<\/div><\/div><\/div><\/div><\/div>","protected":false},"excerpt":{"rendered":"","protected":false},"author":16,"featured_media":15650,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"_monsterinsights_skip_tracking":false,"_monsterinsights_sitenote_active":false,"_monsterinsights_sitenote_note":"","_monsterinsights_sitenote_category":0,"_uf_show_specific_survey":0,"_uf_disable_surveys":false,"footnotes":""},"categories":[36],"tags":[211,212,213,214,215,216],"ppma_author":[456],"class_list":["post-15649","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-revue-de-presse","tag-algorithme","tag-deep-learning","tag-imagerie","tag-morphometrie-geometrique","tag-neural-network","tag-technique","author-alexis-genthon"],"aioseo_notices":[],"authors":[{"term_id":456,"user_id":16,"is_guest":0,"slug":"alexis-genthon","display_name":"Alexis GENTHON","avatar_url":{"url":"https:\/\/www.hematostat.net\/wp-content\/uploads\/2023\/06\/Capture-decran-2021-05-07-a-14.25.02.png","url2x":"https:\/\/www.hematostat.net\/wp-content\/uploads\/2023\/06\/Capture-decran-2021-05-07-a-14.25.02.png"},"first_name":"","last_name":"","user_url":"","description":"H\u00e9matologue.\r\nCorrespondance : H\u00f4pital Saint-Antoine \r\nService h\u00e9matologie clinique\r\n184 rue du Faubourg Saint-Antoine 75012 Paris."}],"_links":{"self":[{"href":"https:\/\/www.hematostat.net\/en\/wp-json\/wp\/v2\/posts\/15649","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.hematostat.net\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.hematostat.net\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.hematostat.net\/en\/wp-json\/wp\/v2\/users\/16"}],"replies":[{"embeddable":true,"href":"https:\/\/www.hematostat.net\/en\/wp-json\/wp\/v2\/comments?post=15649"}],"version-history":[{"count":2,"href":"https:\/\/www.hematostat.net\/en\/wp-json\/wp\/v2\/posts\/15649\/revisions"}],"predecessor-version":[{"id":18203,"href":"https:\/\/www.hematostat.net\/en\/wp-json\/wp\/v2\/posts\/15649\/revisions\/18203"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.hematostat.net\/en\/wp-json\/"}],"wp:attachment":[{"href":"https:\/\/www.hematostat.net\/en\/wp-json\/wp\/v2\/media?parent=15649"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.hematostat.net\/en\/wp-json\/wp\/v2\/categories?post=15649"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.hematostat.net\/en\/wp-json\/wp\/v2\/tags?post=15649"},{"taxonomy":"author","embeddable":true,"href":"https:\/\/www.hematostat.net\/en\/wp-json\/wp\/v2\/ppma_author?post=15649"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}