PET radiomics-based lesions representation in Hodgkin lymphoma patients

Keywords

Statistical learning
Health Analytics
Code:
85/2020
Title:
PET radiomics-based lesions representation in Hodgkin lymphoma patients
Date:
Wednesday 23rd December 2020
Author(s):
Cavinato, L.; Sollini, M.; Kirienko, M.; Biroli, M.; Ricci, F.; Calderoni, L.; Tabacchi, E.; Nanni, C.; Zinzani, P. L.; Fanti, S.; Guidetti, A.; Alessi, A.; Corradini, P.; Seregni, E.; Carlo-Stella, C.; Chiti, A.; Ieva, F.;
Download link:
Abstract:
As medical image analysis has been proven to entail tumor-specific in- formation, the so-called radiomics paradigm holds the promise to characterize the disease and infer long term outcomes of chemotherapy. In this work, we propose an insightful framework for disease characterization in Hodgkin lymphoma which could inform future research. Particularly, an intra-patient similarity index (ISI) was built to represent the homogeneity of the patients’ disease, while a radiomics-based fingerprint was create for local lesion description. Through descriptive statistics and classification algorithms, ISI-weighted fingerprint has been showed to be discriminatory between responders and relapsing patients.
This report, or a modified version of it, has been also submitted to, or published on
Cavinato, L., Sollini, M., Kirienko, M., Biroli, M., Ricci, F., Calderoni, L., ... & Ieva, F. (2020). Pet radiomics-based lesions representation in Hodgkin lymphoma patients. In Book of Short Papers SIS 2020 (pp. 474-479).