FindAPhD Weekly PhD Newsletter | JOIN NOW FindAPhD Weekly PhD Newsletter | JOIN NOW

Diagnosing cancer early: validation of artificial intelligence prediction models in cancer detection

   Barts and The London School of Medicine and Dentistry

This project is no longer listed on and may not be available.

Click here to search for PhD studentship opportunities
  Dr Judith Offman, Dr Oleg Blyuss  No more applications being accepted  Funded PhD Project (Students Worldwide)

About the Project

Applications are invited from graduates with a BSc (First or Upper Second) or MSc (Merit or Distinction), or equivalent, to work within the Wolfson Institute of Population Health. This 4 year studentship will commence in January 2023 and will be based at the Charterhouse Square Campus. This is an exciting opportunity for a graduate from disciplines related to epidemiology, medical statistics, and public health.

Project description


This project is about ensuring that new machine learning tools for detecting cancers early can be applied in day to day clinical practice and therefore improving cancer survival in patients.

Around 375,000 new cancers are diagnosed in the UK every year (2016-18) and only about 50% of cancer patients survive for ten years or more [1]. Diagnosing cancer earlier, when it is easier to treat effectively and with fewer side effects, can improve patient survival and quality of life. Artificial intelligence has the potential to revolutionise cancer diagnosis by allowing earlier detection of cancers from medical images, reducing workforce bottlenecks, providing diagnostic support and using multi-modality approaches, including big data to identify patients at highest risk of developing cancer. New machine learning models for cancer diagnostics are being developed and published at an increasing rate. However, hardly any of these models have been evaluated extensively enough to demonstrate real-world medical utility. Furthermore, where external evaluation has been carried out, these studies have been at high risk of bias and of low generalisability [2, 3].

In a collaboration with machine learning experts who are part of the PRISM research group, we are currently working on AI algorithms that can identify morphological pre-malignant changes in mesothelial cell images. Malignant Mesothelioma (MM) is an aggressive cancer of the pleural lining primarily caused by asbestos exposure. Due to non-specific clinical manifestation, diagnosis usually occurs late when it is difficult to treat, resulting in the 5-year survival rate being less than 5%. Therefore, there is an urgent need to diagnose MM earlier. MM is currently diagnosed by histopathologists recognising morphological cell features on histopathology images and is dependent on their correct interpretation. It can be classified into three subtypes: epithelioid, sarcomatoid, and biphasic. This distinction is crucial to patient treatment, management, and prognosis. PRISM colleagues have developed a machine learning approach for malignant subtyping, which has so far performed well in initial tests [4].

In addition to pattern recognition in medical images machine learning can also be used for pattern recognition in clinical data by utilizing AI approaches including Natural Language Processing to extract structured and unstructured data from electronic healthcare systems, for example to identify individuals at high risk for a specific type of cancer [5].

Aims and Objectives

This PhD studentship is aimed at understanding how to best validate new machine learning models in cancer detection following the initial developmental phase up to first implementation in a clinical setting. This is not about developing new tools, but you will be working with already existing models.

As part of this PhD you will:

1)    Review published studies on artificial intelligence model validation in cancer early diagnosis with a specific focus on biases

2)    Design and carry out an early-stage retrospective validation study of a machine learning model for mesothelioma detection on digital pathology images.

3)    Design and carry out a pilot study of a prospective trial of a machine learning model for clinical data, for example in primary care (GP surgeries).

You will furthermore contribute to the creation of a framework for the validation of machine learning models in computational pathology as part of a multidisciplinary team.

Due to the fast-moving nature of the field, the exact two machine learning algorithms at different stages of validation for these studies will be identified as part of existing collaborations with artificial intelligence experts Dr Oleg Blyuss and Dr Jan Lukas Robertus (PhD co-supervisors).

You will be supervised by an interdisciplinary supervisory team and work closely with experts in epidemiology, medical statistics, machine learning, pathology and primary care. You will be based at the Centre for Prevention, Detection and Diagnosis, which is at the forefront of research internationally into the prevention, detection and control of cancer. We have particularly active research programmes in breast, cervical, colorectal, lung, pancreatic and prostate cancer prevention, cancer screening, awareness and early diagnosis, and statistical methods for clinical trials and epidemiology.

You will develop skills in systematic reviews, study design including clinical trial design, managing studies and analysing datasets. Training in the use of Stata or R will be provided as necessary. By the end of this PhD, you will have obtained a very good understanding of the application of machine learning models in health care.

Informal enquiries can be made via email to:

Dr Judith Offman, Wolfson Institute of Population Health, QMUL ([Email Address Removed])

How to apply

Your application should consist of a CV and contact details of two academic referees. You must also include a personal statement (1,000 words maximum) describing your suitability for the selected project including how your research experience and interests relate to the project.

Please submit your application to: Patrick Mullan ([Email Address Removed]).

Successfully shortlisted candidates will be invited to an interview.

Funding Notes

This 4 year PhD studentship is funded by Barts Charity and comes with a tax-free stipend of £24,278. It is open to all students. Those who are not UK Nationals or non-UK nationals with indefinite leave to remain in the UK will need to acquire a visa ahead of the start of the studentship. University tuition fees (at UK levels) will be met by the funding body.


1. Cancer Research UK: Cancer Statistics for the UK Accessed on 06.06.2022 2022.
2. Anderson AW, Marinovich ML, Houssami N, Lowry KP, Elmore JG, Buist DSM, Hofvind S, Lee CI: Independent External Validation of Artificial Intelligence Algorithms for Automated Interpretation of Screening Mammography: A Systematic Review. Journal of the American College of Radiology 2022, 19(2):259-273.
3. Freeman K, Geppert J, Stinton C, Todkill D, Johnson S, Clarke A, Taylor-Phillips S: Use of artificial intelligence for image analysis in breast cancer screening programmes: systematic review of test accuracy. BMJ 2021, 374:n1872.
4. Eastwood M, Marc ST, Gao X, Sailem H, Offman J, Karteris E, Fernandez AM, Jonigk D, Cookson W, Moffatt M et al: Malignant Mesothelioma Subtyping of Tissue Images via Sampling Driven Multiple Instance Prediction. In: Artificial Intelligence in Medicine: 2022// 2022; Cham: Springer International Publishing; 2022: 263-272.
5. Richter AN, Khoshgoftaar TM: A review of statistical and machine learning methods for modeling cancer risk using structured clinical data. Artificial Intelligence in Medicine 2018, 90:1-14.

How good is research at Queen Mary University of London in Clinical Medicine?

Research output data provided by the Research Excellence Framework (REF)

Click here to see the results for all UK universities
PhD saved successfully
View saved PhDs