Design and implementation of predictive models based on radiomics to assess response to immunotherapy in non-small-cell lung cancer

• M. Corral Bolaños ^[1] ; B. Farina ^[1] ; A.D. Ramos Guerra ^[1] ; C. Palacios Miras ^[2] ; G. Gallardo Madueño ^[3] ; A. Muñoz-Barrutia ^[4] ; G. R. Peces-Barba ^[2] ; L. M. Seijo ^[5] ; J. Corral ^[5] ; I. Gil Bazo ^[6] ; M. Dómine Gómez ^[2] ; M. J. Ledesma-Carbayo ^[1]
  1. [1] Universidad Politécnica de Madrid

     Universidad Politécnica de Madrid

     Madrid, España

     
  2. [2] Fundación Jiménez Díaz

     Fundación Jiménez Díaz

     Madrid, España

     
  3. [3] Universidad de Navarra

     Universidad de Navarra

     Pamplona, España

     
  4. [4] Universidad Carlos III de Madrid

     Universidad Carlos III de Madrid

     Madrid, España

     
  5. [5] Clínica Universitaria de Navarra

     Clínica Universitaria de Navarra

     Pamplona, España

     
  6. [6] Instituto de Investigación Sanitaria de Navarra

     Instituto de Investigación Sanitaria de Navarra

     Pamplona, España

     

  Mostrar afiliaciones +
• Localización: XXXVIII Congreso Anual de la Sociedad Española de Ingeniería Biomédica. CASEIB 2020: Libro de actas / Roberto Hornero Sánchez (ed. lit.), Jesús Poza Crespo (ed. lit.), Carlos Gómez Peña (ed. lit.), María García Gadañón (ed. lit.), 2020, ISBN 978-84-09-25491-0, págs. 181-184
• Idioma: inglés
• Texto completo no disponible (Saber más ...)
• Resumen

  • Lung cancer is the leading cause of cancer-related deaths in Europe. Immunotherapy treatments have been proved as the new standard of care for stage III-IV non-small cell lung cancer patients. However, the treatments vary in success, and there is not a reliable biomarker. This retrospective project aimed to develop a predictive model based on radiomics through machine learning or deep learning techniques to assess the response to the treatment, understood as the progression (or not) of the disease. Then, the study was complemented with an analysis of the progression-free survival time and an attempt of association with biological data. We used the basal computed tomography images of the primary tumour lesions from a cohort with 84 patients with IV stage nonsmall- cell lung cancer. The best performance model reached an AUC of 0.80 – 90 % CI [0.62, 0.99]. Our results suggest that the radiomics models may be useful for patient classification