Medical Imaging and Interpretation Group:
Overview

Overview

Colour retinal photograph Parametric map showing the distribution of blood within the retina.

Left: Color photograph of the human ocular fundus (eye).
Right: “Parametric map” showing the distribution of blood within the retina.


3D Finite element model of a mouse togther with simulated NIR light propagation.

Research in Medical Image and Interpretation Group (MIG) is focused on finding novel and non-invasive methods of acquiring data and extracting information from biological tissue. Some of the recent work has been focused on developing novel techniques that allow us to use the features and properties of an optical image to infer quantitative information about the structure and composition of the tissue being imaged. Specific areas of work include:
  • Optical imaging for the detection of skin cancer:
    This work relies on a detailed understanding of the physics of image formation, and has led to the development of the successful skin imaging system, the SIAScope.
  • Detection of early signs of retinopathies:
    A recently completed project has successfully demonstrated the ability of these methods to detect and quantify retinopathies from images of the human eye.
  • Optical diagnosis of colon cancer.
  • Detection and characterisation of breast cancer using optical imaging.
  • Optical tomographic imaging of the human brain.
  • Small animal imaging using molecular markers.
  • Multi-modality imaging and image coregistration.
We are also developing new techniques for modelling light transport in tissues. A finite element model based image reconstruction package (NIRFAST), freely available and based on MATLAB, has been developed and is being used for applications in optical imaging using intrinsic and extrinsic molecular markers. We are also especially interested in the application of cutting-edge techniques from natural computation to help us to explore and optimise our models and imaging methods.

The grouping is also interested in statistical methods of image analysis, especially the identification of the boundaries of lesions, and the extraction of textural information from an image. This information can be of great benefit to a clinician when making a diagnosis.

The grouping hosts and maintains a fully-equipped multispectral imaging laboratory, part of the University's Collaborative Research Network in Imaging and Visualisation.

Further Information