Explainable Multi-Modal Deep Learning for Precision Non-Small Lung Cancer Treatment

About the Project

A full-time PhD studentship is available within the Department of Surgery and Cancer at Imperial College London. This is an exciting interdisciplinary opportunity to develop and apply cutting-edge machine learning techniques to address a key challenge in personalised lung cancer treatment.

Project Background:

Precision oncology has transformed the way cancer is treated, moving from “one-size-fits-all” protocols toward individualised strategies. Non-small cell lung cancer (NSCLC) is the most common cause of cancer-related mortality worldwide. While checkpoint blockade immunotherapy (CBI) has significantly improved survival in a subset of its patients, identifying such patients remains difficult due to the imperfect PD-L1 immunohistochemistry method currently in use. Despite efforts to use AI on imaging data to find an alternative non-invasive solution, most computational tools developed for NSCLC CBI do not integrate biological data, and are thus limited in their explainability, hindering their clinical acceptance and deployability.

To address this gap, we propose to develop a novel multi-modal biomarker for NSCLC that integrates biological (transcriptomics, histology and blood biomarkers) data with imaging features to construct a shared latent space that captures complementary disease information. This latent representation will be reinforced with causal modelling, enabling the disentanglement of mechanistic relationships between tumour biology, CBI response, treatment adverse events, and patient prognosis. By adding biological information and a casual basis to an imaging-derived model, we aim to improve its predictive performance and explainability, providing clinicians with an interpretable non-invasive tool for patient stratification and therapeutic decision-making in NSCLC CBI.

The project aim is develop a latent space model that enables multiple modal data integration including genomic, histopathologic, imaging, and clinical data to guide precision CBI for NSCLC, with direct applications in patient stratification, treatment planning, and prognostic counselling.

The successful student will be registered for an Imperial College PhD degree in the Faculty of Medicine where they will be based within the group led by Dr. Mitch Chen. We welcome applications from excellent candidates with a strong background in computing, mathematics or physics. High proficiency in coding (Python, R) is expected. Prior experience in cancer research is desirable but not essential.

The studentship is for 36 months. Starting date and PhD enrolment is negotiable, but a preferred starting date is 1st October 2026.

Applicants should submit their CV and a covering letter, including a statement on how overseas fee difference will be covered (for non-UK applicants), and full contact details of two referees, to Dr. Mitch Chen (mitchell.chen@imperial.ac.uk). Imperial College PhD entry requirements must be met and the successful applicant will subsequently need to apply on-line via the Imperial College application process.

Application is accepted on a rolling basis (CSC applicants must adhere to program application deadlines). Interviews will be carried out via Microsoft Teams (first round) and in-person (final round). Successful applicants in the first-round should be able to attend the final round interview on site in London.

Funding Notes

There are two funding routes:

Eligible clinically qualified applicants (GMC-registered) can apply for this opportunity via Imperial CSC.

Non-clinical applicants must have been awarded individual scholarship, or self-secured funding to cover tuition fees, living expenses, and any associated research costs.