Feature-based assessment of visuospatial neglect severity in a computer-based line cancellation task

Visuospatial neglect is a complicated disorder that affects a large number of stroke patients and may occur in different degrees of severity. Conventional pen and paper tests used for detection of neglect rely on a battery of tests and many studies have reported that a single test is not enough for neglect detection. To give a clearer diagnosis and reduce patient testing fatigue, increasing test sensitivity is an important task. This study presents a computer-based approach to visuospatial neglect assessment whereby the importance of timing features in the detection of different degrees of neglect severity is highlighted. Results obtained from a line cancellation test showed that while static features are important for the detection of severe neglect cases, they are insufficient for the detection of some moderate and the majority of mild neglect cases among the stroke population. An in-depth static and dynamic feature assessment is carried out offering potential increase in test sensitivity.