Detection of recurrent nerve paralysis: Development of a Computer Aided Diagnosis system

Abstract Recurrent nerve paralysis (RP) is one of the most frequent complications of thyroid surgery. It reduces vocal fold mobility. Nasal endoscopy, a mini-invasive procedure, is the reference procedure to detect RP, and is based on the examination of vocal fold mobility. A new approach based on laryngeal ultrasound acquisition and using a dedicated data analysis was designed to help with the automated detection of RP. One hundred and fifty subjects were enrolled for this feasibility study: 50 controls, 50 patients with RP and 50 patients without RP, according to nasal endoscopy. The ultrasound protocol was based on a ten-second B-mode acquisition in an axial plane during normal breathing. Image processing included three steps: 1) the detection of two consecutive closing and opening images corresponding to extreme positions of vocal folds in the sequence of B-mode images; 2) the positioning of three landmarks and the robust tracking of these points using multi-pyramidal refined optical flow approaches; 3) the estimation of quantitative parameters indicating left and right fractions of mobility and motion symmetry. Results provided by the first two image processing steps were compared to those obtained by an expert. Motion symmetry and fraction of mobility indices were systematically computed using the automated procedures. Associated sensitivity and specificity values for detecting RP were then calculated. To optimize the performances of the system, a mixed CAD system, which integrates the automatic steps of image processing and a possible correction of its results by a trained operator, was developed. Laryngeal ultrasound combined with appropriate image processing helped in the diagnosis of recurrent nerve paralysis and could be proposed as a first-line method.