Purpose

Diffuse large B-cell lymphoma (DLBCL) is a heterogeneous entity, with patients exhibiting a wide range of outcomes. Pre-therapeutic stratification is important because patients who are R-CHOP refractory have a very poor prognosis. Patients with poor prognostic characteristics should be offered first-line treatment with new targeted drugs in addition to R-CHOP. However, identification of poor responders is not possible with the Revised-International-Prognostic-Index(R-IPI) or IPI risk scores. Radiomics-analysis on [18]FDG-PET provides semi-quantitative features of tumor characteristics as intensity, shape, volume, localization and texture. Importantly, this [18]FDG-PET radiomics approach permits in-depth whole-body intra- and inter-tumor analyses, which is advantageous to histology analyses, which are typically based on a single biopsied lesion. Recent data suggest that radiomics features carry prognostic information in addition to clinical and genetic information

Objectives

  • To identify where quantitative PET/CT data can guide treatment decisions to determine an individualized approach to first-line and second-line therapy

  • To generate a pipeline for the delineation of tumor burden and identify quantitative and automated methods to do so

  • To develop a prediction model for tumor progression in FL patients using radiomic features

  • To evaluate the predictive capabilities of radiomic features in several (inter)national patient cohorts

  • To explore the potential of artificial intelligence on DLBCL PET scans for the generation of predictive models