Don't miss our weekly PhD newsletter | Sign up now Don't miss our weekly PhD newsletter | Sign up now

  Develop a machine learning approach to multimodal imaging


   Department of Physics and Astronomy

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

Click here to search FindAPhD.com for PhD studentship opportunities
  Dr A Cadby, Dr J King  Applications accepted all year round  Competition Funded PhD Project (UK Students Only)

About the Project

The emergence of antimicrobial resistance (AMR) threatens to undo the advances of modern medicine, making commonplace interventions like major surgery or cancer chemotherapy effectively impossible. Our immune system has evolved over billions of years to provide robust defence against invading unicellular organisms, particularly bacteria. Yet, despite this complexity and evolutionary advance, our defences are regularly breached by bacteria, leading to high antibiotic usage, hospitalisations, and death.

A key step in neutralising threats from bacteria is the internalisation and subsequent killing of these bacteria by professional phagocytes - neutrophils or macrophages. This step can be evaded by bacteria, which have developed an array of mechanisms for evading host killing, by escape from the intracellular killing pathways, division, and subsequent killing of the phagocyte. Because this occurs at the level of a single cell and often a single bacterium, there remains a key gap in our knowledge around how diverse bacteria escape host killing in phagocytes. The best way to address this gap is to use optical microscopy. Optical microscopy allows us to study these processes at the single cell level.

Modern optical microscopy consists of a powerful set of techniques which led to dramatic increases in our understanding of biological systems. There is a huge variety of different microscopy techniques, and each of these modalities has its advantages and disadvantages. Some techniques provide high spatial resolution, some high temporal resolution. However, the increase in resolution often comes with the requirement to use higher illumination intensities. These higher illumination intensities are damaging to biological samples especially neutrophils or macrophages. This makes following processes over time in high resolution difficult. Ideally, we would want to image these biological systems with a low damage technique until the process we are interested in studying starts, then we switch to a higher resolution technique.

Current commercial microscopes are design to performing a single modality meaning that to swap modalities we would have to move the sample during the experiment. We have developed a versatile imaging platform which can perform many different imaging modalities. With the ability to swap between the different modalities during imaging we can switch modalities nearly instantaneously. The versatile imaging platform allows us to following biological processes over time with several different modalities.

We would like to automate this system, such that using a machine learning approach we can teach the microscope to switch between modalities depending on the stage of the process we are following.

 What will you do.

You will use machine vision and neural network classifier to teach the microscope how to follow a process and which technique to use for each stage. This will require collecting training data, optimising imaging conditions and training neural networks.

Who will you work with.

You will work in a multidisciplinary team based in the department of physics and astronomy and working alongside colleagues in the Biosciences and medical sciences.

What will you learn.

You will get advanced training in optical design, optical microscopy, the development, and training of neural networks.

 Students must be able to start by October 2023.

Physics (29)

References


https://www.sheffield.ac.uk/physics/people/academic/ashley-cadby
https://www.sheffield.ac.uk/lmf

Where will I study?

Search Suggestions
Search suggestions

Based on your current searches we recommend the following search filters.

 About the Project