公开(公告)号：EP4398132A1

公开(公告)日：2024-07-10

申请号：EP23382001.8

申请日：2023-01-03

申请人： Centre De Visió Per Computador ,  Fundació Institut d'Investigació en Ciències de la Salut Germans Trias i Pujol (IGTP) ,  Universitat Autónoma de Barcelona

发明人： GIL RESINA, Debora ,  SANCHEZ RAMOS, Carles ,  ROSELL GRATACÓS, Antoni ,  BAEZA MENA, Sonia Margarita ,  TORRES, Guillermo Eduardo

IPC分类号： G06F18/2413 ,  A61B5/00 ,  G06T7/00 ,  G16H50/20

CPC分类号： G06V2201/03120220101 ,  G06F18/24133 ,  G06T7/0012 ,  G16H50/20 ,  G16H30/40 ,  G06V10/82

摘要： A method, system and computer programs for characterization of pathologies are provided. The method comprises accessing or receiving image data and clinical and demographic data and/or genomic information, the image data including different types of scans of a subject, and the clinical and demographic data including attributes of a group of subjects; computing radiomic features using a feature extraction algorithm and embedding the computed radiomic features and the clinical and demographic data and/or genomic information in an abstract space representing the lesion; performing a characterization of the lesion at different levels by determining categorical and quantitative variables associated to each level using the computed radiomic features, providing a multi-diagnostic and/or -prognostic characterization of the lesion as a result; and transforming the provided multi-diagnostic and/or -prognostic characterization of the lesion into radiological words or annotations that quantitatively and/or qualitatively describe the lesion using deep learning or machine learning algorithms.