Application of image analysis to neurological screening through figure-copying tasks

Abstract

Approaches to the detection and monitoring of neurological dysfunction are generally resource-intensive and often difficult to evaluate objectively. This paper describes an approach to neurological screening which is based on the on-line analysis of figure copying tasks, and a variety of features are measured to characterise task execution. It is shown how both qualitative and quantitative measures of individual performance can be obtained in real time. A preliminary experimental study is described to identify appropriate performance indicators for a reference asymptomatic population and for a population with a known abnormal clinical history. The results obtained experimentally demonstrate that the approach proposed, while convenient and efficient in terms of the resources required in implementation, nevertheless offers a high degree of robustness and sensitivity which supports an optimistic view of its clinical viability.