Enhancing lung cancer screening with low-dose CT: a comprehensive study on quantitative tools, data standardization, and reproducibility
Moreover, given the large volume and variety of data collected in screening trials, data standardization is critical. It ensures consistent collection, analysis, and sharing, promoting reproducibility across different trials and platforms. Lastly, generating reliable synthetic data can help handle imbalances in datasets, providing robust and representative sets for training and validating AI algorithms extensively.
This research project thus aims to enhance the cost-effectiveness and personalization of lung cancer screening programs using LDCT. The project focuses on defining and validating biomarkers and predictive models for lung cancer, cardiovascular disease, and COPD, individually and in combination. Additionally, the study aims to profile lung cancer aggressiveness through quantitative descriptors.
To ensure reliability and collaboration, the project emphasizes the importance of data standardization, proposing a unified data structure for consistent analysis across different trials and platforms. Furthermore, we plan to tackle imbalances in datasets by generating reliable synthetic data using generative deep learning models and maintaining statistical properties while addressing privacy concerns.
The research project concludes with a commitment to reproducibility studies, employing robust statistical methods to confirm the consistency and replicability of identified biomarkers, aggressiveness profiles, and personalized screening approaches. Overall, this project seeks to advance the field of lung cancer screening by integrating quantitative tools, standardized data practices, and reproducibility studies for improved effectiveness and personalized patient care.
The project aims to study quantitative and reproducible tools for lung cancer screening with LDCT to improve the cost-effectiveness of lung cancer screening programs. Our specific objectives include:
1. Definition and validation of quantitative biomarkers and predictive models. We will explore and analyze quantitative descriptors of lung cancer, cardiovascular disease, and COPD, individually and in combination. The goal is to improve the prediction of lung cancer incidence and subjects’ mortality.
2. Profiling lung cancer aggressiveness. Through the analysis of quantitative descriptors, we aim to create profiles that shed light on the aggressiveness of lung cancer.
3. Improving data standardization. We plan to establish a unified data structure to enhance the reliability of analyses and promote collaboration. This ensures consistency and compatibility across datasets through proper data transformation and metadata definition.
4. Generating lung cancer synthetic data. Using generative deep learning models, we aim to create synthetic data that maintains the statistical properties of the original data. This approach increases data availability while addressing privacy concerns.
5. Reproducibility studies. We will employ robust statistical methods and validation techniques to confirm the consistency and replicability of identified biomarkers, aggressiveness profiles, and personalized screening approaches.
Dr. Giulia Raffaella De Luca, Department of Electrical, Electronic, and Information Engineering "Guglielmo Marconi", University of Bologna
Dr. Andrea Espis, Department of Electrical, Electronic, and Information Engineering "Guglielmo Marconi", University of Bologna
Dr. Matteo Lai, Department of Electrical, Electronic, and Information Engineering "Guglielmo Marconi", University of Bologna
Dr. Chiara Marzi, Department of Statistics, Computer Science and Applications “Giuseppe Parenti”, University of Florence
Prof. Mario Mascalchi, "Mario Serio" Department of Experimental and Clinical Biomedical Sciences, University of Florence
Dr. Riccardo Scheda, Department of Electrical, Electronic, and Information Engineering "Guglielmo Marconi", University of Bologna