Rationale

In December 2022, the Council of the EU recommended EU member states to explore the feasibility and effectiveness of lung cancer screening (LCS) using low-dose computed tomography (LDCT). In a lung cancer screening program, a population at high risk for lung cancer is scanned every year using LDCT and the scans are interpreted by radiologists for the occurrence of lung cancer related abnormalities (lung nodules). If those are detected, the size is preferably assessed using computerized tools that perform volumetric measurement of the nodules. Current computerized tools primarily use Artificial intelligence (AI) models for the detection, segmentation and measurement of pulmonary nodules.

In the past the performance of AI-based detection algorithms has already been evaluated in publications and public challenges such as LUNA16 (publication link), the Kaggle Data Science Bowl 2017 (publication link). In addition, nodule malignancy probability estimation has been benchmarked in the LUNA25 challenge. The quantification of the nodule size, however, as a measurement task has not been assessed in a public benchmarking challenge. The nodule size can be measured as the maximum diameter of a nodule or as the nodule volume. Publications from the NELSON trial and other publications suggest that the volumetric assessment is superior to diameter assessment.

LUNOVO26 challenge

The LUNOVO26 (Lung Nodule Volumetry 2026) challenge is a public benchmarking challenge organized by EIBALL (European Imaging Biomarkers Alliance), QMIC (Quantitative Medical Imaging Coalition) and Radboud University Medical Center. LUNOVO26 is supported and co-funded by the SOLACE project. LUNOVO26 aims to provide an objective framework for the evaluation of AI algorithms for volumetric measurement of pulmonary nodules in lung cancer screening. LUNOVO26 has the following objectives:

  • Provide an objective framework for transparent validation and comparison of the accuracy, variability and reproducibility of both academic and commercial AI algorithms for volumetric measurement of pulmonary nodules.
  • Attract attention of the community to the task of volumetric measurement of pulmonary nodules, a crucial component for running an effective lung cancer screening program.
  • Provide a good basis for reliable quality assurance and quality control of AI algorithms for volumetric measurement of pulmonary nodules.

Organizers

  • Prof. Aad van der Lugt, Department of Radiology & Nuclear Medicine, Erasmus University Medical Center, Rotterdam, the Netherlands (member of EIBALL)
  • Dr. Gudrun Zahlmann, Quantitative Medical Imaging Coalition, Madison, Wisconsin, United States (member of EIBALL)
  • Dr. Colin Jacobs, Department of Medical Imaging, Radboud University Medical Center, Nijmegen, the Netherlands (PI for Radboud UMC in the SOLACE project)
  • Dr. Joanna Bidzińska, Department of Radiology, University Clinical Centre, Gdansk; Department of Radiology, Medical University of Gdansk, Gdansk, Poland
  • Dr. Gareth Iball, Faculty of Health and Social Care, University of Bradford, England, UK