Computer aided feature extraction in the paper version of Luria's alternating series test in progressive supranuclear palsy

Luria’s Alternating Series Test (AST) is a popular bedside graphomotor task used to assess set-shifting both in patients with focal brain damage and with neurodegenerative disease. Several approaches to automate its assessment are available, yet they require digitizers or other electronic devices. However, archived data from the past 60 years are stored only on paper. This pilot study concerns the extraction of the features from the paper version of AST using image processing algorithms. In this study, AST data from twenty two cases with progressive supranuclear palsy are processed. Each case contains reference data (a pattern sequence drawn by the examiner) and diagnostic data (continuation of the pattern drawn by the patient). Spatial-based features as well as (novel) erosion bar charts (EBC) are extracted from the initial characters of the examiner and patient sequence. The results are evaluated to find parameters that are statistically different between the groups. The results indicate that patients typically write larger characters than examiners, yet their pen-paper pressure is smaller.