Developing AI platforms for rapid diagnosis of adult and childhood blood cancer

Artificial intelligence (AI) and machine learning (ML) could greatly change how doctors diagnose and treat cancer. One important part of this process is looking at pictures of tissue samples (called histopathology images) to identify cancer and its type.

Plasma cell myeloma (PCM) is a type of blood cancer that affects around 6,200 people in the UK each year, and it usually has a poor outlook for patients. Treatments are changing quickly as researchers discover new ways to target the disease. The basic criteria for diagnosing PCM have only been slightly updated in the past ten years, but analysing bone marrow is still very important for making a diagnosis. When doctors take a bone marrow sample, they prepare slides from the blood to examine it. The amount of plasma cells - specifically, more than 10% of abnormal plasma cells - helps determine the treatment options and can affect how long patients survive. However, analysing these samples can be slow and take a lot of work for doctors.

The goal of this PhD project is to create AI and machine learning techniques to more accurately and quickly identify plasma cells in bone marrow samples to assist with diagnosing myeloma. The project includes three aims:

1.  Train AI and machine learning systems using data from about 200 previously collected bone marrow samples that include detailed information from five blood specialists.

2.  Analyse new bone marrow samples as they are obtained and compare the speed and accuracy of these AI tools with the assessments made by experienced specialists.

3.  Use insights from the research above and analyse the environment around the bone marrow cells (an aspect not currently considered in standard diagnostics) to develop a pilot AI tool for diagnosing and predicting outcomes for childhood leukaemia based on bone marrow samples

Click here to access the application form