The use of image processing to quantitatively analyze oromotor movements : a feasibility study with nasopharyngeal cancer survivors

Abstract

Nonspeech oromotor movements are important to diagnosis and treatments of speech disorders and dysphagia. Yet, there is a lack of objective measurement systems. Therefore, this study proposed an image processing system to assess oromotor movements. Nasopharyngeal cancer survivors, who often report restricted oromotor movements post-cancer treatments, and healthy participants were recruited and grouped into impaired and normal speech intelligibility groups. Participants were video-recorded while they performed oromotor tasks including tongue lateralization and elevation, lip protrusion and retraction, and jaw opening. With the Fiji software, pixel-wise displacement of the targeted structures during movements was measured. Group difference between speakers with impaired and normal intelligibility and correlations between the oromotor and speech measures were established as precise speech performance requires adequate range of oromotor movements. This study found significant group differences between the two groups in four oromotor measurements: lip retraction ratio (left and right), tongue elevation and jaw opening values; and significant correlations across seven out of eight oromotor measures with speech measures. High inter- and intra-rater reliability, diagnostic and convergent validity were established. The results revealed the potential of this system. Yet, further investigations have to be done for validation and revision of the method.