Lung cancer: Artificial Intelligence for risk stratification
A project that, thanks to the synergy between diagnostic imaging and Artificial Intelligence, aims to develop a predictive model to stratify the risk of the individual patient with lung cancer.
Lung cancer is the fourth most frequently diagnosed malignant form of cancer in Europe and is the leading cause of death from cancer in both sexes. The identification of diagnostic and predictive biomarkers is essential in the choice of patient management strategy. Genetic information can be of great help in clinical decisions because it can be used to predict survival – prognostic biomarker – or to predict treatment response – predictive biomarker.
Diagnostic imaging provides significant data on tumour characteristics. In recent years, important progress has been made through advanced imaging analysis using Radiomics and Artificial Intelligence programs. Radiomics is a sophisticated technique that allows to extract quantitative information from imaging (CT and PET) and integrate it with the patient’s clinical data and the genetic characteristics of the tumor (Radiogenomics).
Aim of the research
We aim to develop a platform based on artificial intelligence that would be able to analyze medical images, extract clinical and instrumental data – with automatic or semi-automatic methods – from electronic medical records, de-identify patient data (with due respect for privacy) and build a predictive model to stratify the individual patient’s risk. To realize this project, we are going to use a clinical scenario based on lung cancer. Therefore, we select patients with non-small cell lung cancer (NSCLC) by extracting the clinical and biological information from clinical records. We will also analyze CT and PET images using dedicated algorithms and deep neural networks based on the amount of data available. The results will be integrated with mutational and clinical data to build a predictive model. The approach can provide relevant information on the patient’s prognosis and can be used prospectively in patients with non-small cell lung cancer and, in the future, in case of other neoplasms.
Five researchers are involved in the project. The first phase will be completed in two years, while the entire project will end in five years.
How are the medical skills and the competences of the data scientists integrated? Doctors identify clinical needs and priorities. Based on this information, they design a research protocol and implement it with the support of the engineers involved in the project.