School of Engineering and Digital Arts

Medical Image Analysis

Intelligent Interactions Research Group

Much of this theme is concerned with application-specific work which builds directly on the fundamental tools and techniques developed from more generic studies. This is typified by an area of work which the Group has pioneered, concerned with the computer-based analysis of writing and drawing activity as a means of assessing medical conditions which arise from neurological dysfunctioning in patients.

Evaluative tests based on the analysis of writing and, especially drawing, are very common in medical assessment. These are often characterised as "pencil-and-paper" tasks, and typically involve the drawing (from memory) or copying of geometric shapes, or related activity such as written line bisection or object cancellation tests. We have developed a powerful technique which can capture the activity of the patient in carrying out such tasks but which, because the data capture is carried out using on-line pen movement monitoring, makes available both static information (related to the visible image produced) and, importantly, the dynamic information which describes the actual execution of the drawing process itself.

We have adapted this type of technique to the study of patients who have suffered cerebral vascular accident (CVA or stroke) and, working with expert clinical staff, have carried out some ground-breaking work on the study of visuo-spatial neglect in stroke patients, shedding new light on the ability easily to assess the extent and nature of the damage caused to a patient’s perceptual abilities. We have used similar techniques to develop simple but more objective tests to evaluate the performance of patients with Parkinson’s disease.

A significant area of research recently has been a major project to follow up previous pilot studies developing a system to automatically administer and analyse a common evaluation test used with dyspraxic children. Dyspraxia is a common condition in young children which results in coordination difficulties and a range of other debilitating symptoms. A commonly used assessment technique uses the Visual Motor Integration (VMI) test, requiring the child to copy a sequence of increasingly complex geometrical drawings. We have shown how our automated techniques can extract important evaluative information from this type of testing, and how it is possible to disambiguate motor, cognitive and strategy-related performance indicators from the task during the completion of this standard test procedure. This is important on-going work which requires further research to develop more efficient and validated diagnostic procedures, to refine the feature extraction processes underlying diagnostics, and to provide a greater degree of automated support for clinicians in applying the VMI test. We are also interested in the possibilities which our work to date provides for developing new and more effective rehabilitation strategies which exploit the information extracted from the testing process.

This broad area of work has led to a collaboration with the University of Rouen to implement and evaluate a generic system for this type of testing which can effectively integrate the individual test elements which are required when investigating different medical conditions. We are developing a test structure which can identify both universally required features and those which are condition-specific, integrating these to provide a clinical tool for routine use for the assessment and monitoring of a wide range of clinical conditions.

Other areas in which the Group has interests include the automated processing of echocardiogram images and digital mammography.

This research theme has been supported by EPSRC, the EU INTERREG Programme, and a number of medical charities.

Selected Projects

School of Engineering and Digital Arts, Jennison Building, University of Kent, Canterbury, Kent, CT2 7NT

Enquiries: contact us

Last Updated: 30/08/2017