Affiliation:
1. Università Cattolica del Sacro Cuore
2. Maastricht University Medical Centre+
3. Real World Data Facility, Gemelli Generator, Fondazione Policlinico Universitario Agostino Gemelli IRCCS
4. Fondazione Policlinico Universitario Agostino Gemelli IRCCS
5. Radboud University Medical Center
Abstract
Abstract
Predictive modelling of cancer outcomes using radiomics faces dimensionality problems and data limitations, as radiomics features often number in the hundreds, and multi-institutional data sharing is often unfeasible. Federated learning (FL) and feature selection (FS) techniques combined can help overcome these issues, as one provides the means of training models without exchanging sensitive data, while the other identifies the most informative features, reduces overfitting, and improves model interpretability. Our proposed FS pipeline based on FL principles targets data-driven radiomics FS in a multivariate survival study of non-small cell lung cancer patients. The pipeline was run across datasets from three institutions without patient-level data exchange. It includes two FS techniques, Correlation-based Feature Selection and LASSO regularization, and Cox Proportional-Hazard regression with Overall Survival as endpoint. Trained and validated on 828 patients overall, our pipeline yielded a radiomic signature comprising "intensity-based energy" and "mean discretised intensity". Validation resulted in a mean Harrell C-index of 0.59, showcasing fair efficacy in risk stratification. In conclusion, we propose an approach for distributed radiomics involving preliminary FS to reduce the available feature set to the most important on a data-driven basis and reduce dimensionality problems other than those related to data limitations and interpretability.
Publisher
Research Square Platform LLC
Reference49 articles.
1. Radiomics: the process and the challenges;Kumar V;Magn Reson Imaging,2012
2. High-Dimensional Cluster Analysis with the Masked EM Algorithm;Kadir SN;Neural Comput,2014
3. Wu, Y. et al. Robust Feature Selection Method of Radiomics for Grading Glioma. Proceedings of the 2nd International Conference on Healthcare Science and Engineering null, null (2018).
4. Feature selection methods and predictive models in CT lung cancer radiomics;Ge Jie GZ;J Appl Clin Med Phys,2022
5. Feature Selection and Its Use in Big Data: Challenges, Methods, and Trends;Rong Dunwei;IEEE Access,2019