Personalised lung cancer treatment through outcome predictions and patient stratification
Principal Investigator
Name
Charles-Antoine Collins Fekete
Degrees
Ph.D.
Institution
University College London
Position Title
Senior Research Fellow
Email
About this CDAS Project
Study
NLST
(Learn more about this study)
Project ID
NLST-881
Initial CDAS Request Approval
Feb 16, 2022
Title
Personalised lung cancer treatment through outcome predictions and patient stratification
Summary
The staging of non-small cell lung cancer (NSCLC) aims at classifying the operability of lung tumours based on organ involvement and size but does not highlight the significant anatomical variation that can be present within each staging group. Thus, there is a clear and unmet need for personalised patients’ stratification and treatment based on a comprehensive analysis of the risk and outcomes. Concurrently, machine learning algorithms have recently demonstrated a unique capacity to unveil correlation features in a large imaging dataset. This project aims to solve this recently identified flaw of lung cancer dose planning strategy using an algorithmic solution which exploits the massive worldwide growth in AI. A new way of planning radiotherapy treatments will be promoted, using a machine-learning predicted outcome to adapt the treatment strategy and improve the patient's survival. This will involve the development of a platform based methodological approach to change the current approach in radiotherapy planning of lung cancer patients.
Aims
1. Define a representation model to map the patient into a common ATLAS space to enhance a machine learning (ML) algorithm performance
2. Predict the most likely treatment outcome from the patient multi-modal representation in the ATLAS space.
3. Perform a thorough validation and investigation of the limits of the predictive model, and produce a Code of Practice for clinical usage
4. Introduce the most likely predicted outcome to personalise the treatment strategy in radiotherapy
5. Integrate multiomics data in the prediction pipeline, including histopathological data, genomics, and genetic markers
Collaborators
Professor Gary Royle, University College London
Professor Maria Hawkins, University College London