Deep learning for lung cancer on computed tomography: early detection and prognostic prediction


    215 Downloads (Pure)


    Lung cancer is one of the most fatal cancers in the world, the leading cause of death among both men and women. The five-year survival rate for lung cancer patients is only between 10 and 20%. However, the mortality rate can be reduced if lung cancer is diagnosed at an early stage and treated promptly. Screening trials have been established in many countries to improve early detetion of lung cancer, but it results in numerous scans that need to be evaluated, which is labor-intensive. On the other hand, when lung cancer is diagnosed at an early stage in screening, the clinical response after the treatment can vary between patients. Therefore, strong needs exist for accurate early detection and prognostic prediction of lung cancer.
    Deep learning recently has achieved great success in medical image analysis, especially for lung cancer. The results described in this thesis show that combining clinical procedures, deep learning techniques are feasible to assist radiologists with pulmonary nodule detection and rule out most negative scans in lung cancer screening. Besides, by integrating clinical factors and imaging features, deep learning can identify high mortality risk lung cancer patients who could benefit from adjuvant chemotherapy. With the implementation of lung cancer screening programs, more imaging and clinical data will be available, which enables deep learning to further boost the efficiency of screening procedures and lower the lung cancer mortality in the future.
    Originele taal-2English
    KwalificatieDoctor of Philosophy
    Toekennende instantie
    • Rijksuniversiteit Groningen
    • van Ooijen, Peter, Supervisor
    • Oudkerk, Matthijs, Supervisor
    • Veldhuis, Ruurd, Supervisor
    Datum van toekenning22-jun-2021
    Plaats van publicatie[Groningen]
    StatusPublished - 2021

    Citeer dit