Predicting 2-year time to progression in diffuse large B cell lymphoma using 3D CNNs on whole-body PET/CT scans.
Authors
Affiliations (14)
Affiliations (14)
- Cancer Center Amsterdam, Department of Radiology and Nuclear Medicine, Amsterdam UMC, Vrije Universiteit Amsterdam, Amsterdam, Netherlands. [email protected].
- Cancer Center Amsterdam, Imaging and Biomarkers, Amsterdam, Netherlands. [email protected].
- , Amsterdam, The Netherlands. [email protected].
- Cancer Center Amsterdam, Department of Radiology and Nuclear Medicine, Amsterdam UMC, Vrije Universiteit Amsterdam, Amsterdam, Netherlands.
- Cancer Center Amsterdam, Imaging and Biomarkers, Amsterdam, Netherlands.
- Cancer Center Amsterdam, Department of Hematology, Amsterdam UMC, Vrije Universiteit Amsterdam, Amsterdam, Netherlands.
- Department of Epidemiology and Data Science, Amsterdam Public Health Research Institute, Amsterdam UMC, Vrije Universiteit Amsterdam, Amsterdam, The Netherlands.
- Department of Hematology, Erasmus MC Cancer Institute, University Medical Center Rotterdam, Rotterdam, The Netherlands.
- Department of Nuclear Medicine, Clinic and Polyclinic for Nuclear Medicine, University of Leipzig, Leipzig, Germany.
- Department of Hematology, West German Cancer Center, University Hospital Essen, University of Duisburg-Essen, Essen, Germany.
- School of Biomedical Engineering and Imaging Sciences, King's College London and Guy's and St Thomas' PET Centre, King's Health Partners, King's College London, London, UK.
- Department of Clinical Oncology, Guy's Cancer Centre, School of Cancer and Pharmaceutical Sciences, King's College London University, London, UK.
- SAKK Swiss Group for Clinical Cancer Research, Bern, Switzerland.
- Department of Nuclear Medicine, Medical Imaging Centre, Semmelweis University, Budapest, Hungary.
Abstract
The aim of this study was to develop 3D convolutional neural networks (CNN) for the prediction of 2 years' time to progression using PET/CT baseline scans from diffuse large B-cell lymphoma (DLBCL) patients. The predictive performance of the 3D CNNs was compared to that of the International Prognostic Index (IPI) and a previously developed 2D CNN model using maximum intensity projections (MIP-CNN). 1132 DLBCL patients were included from 7 independent clinical trials. Two 3D CNN models were developed using a training dataset of 636 patient scans merged from two trials, one CNN model trained on lesion-only PET (L-PET3D-CNN) and the second model trained on both lesion-only and whole body PET scans (LW-PET3D-CNN). The 3D models were cross-validated and performance was independently tested on 496 patient scans merged from five external trials, using the area under the curve (AUC). Performance was compared to the IPI and MIP-CNN using DeLong test. Occlusion maps were implemented to gain insights about the models' decision-making process. The IPI and the MIP-CNN yielded an AUC of 0.53 and 0.65 respectively on external test data. The L-PET3D-CNN and the LW-PET3D-CNN yielded a significantly higher AUC, 0.65 and 0.64 respectively, compared to the IPI. For each individual external clinical trial, the models were consistently better than IPI. The MIP-CNN and the 3D CNNs showed equivalent performance on external test data. The 3D CNN models remained predictive of outcome on all external test datasets, outperforming the IPI. Although these models perform similarly to the MIP-CNN, the main advantage of the 3D CNN is the use of 3D occlusion maps to better understand the decision-making process of the models.