Vocal folds paralysis study using a pre-processing stage of Gabor filtering and Chan-Vese segmentation

Abstract

This paper describes the methodology proposed by the authors to diagnose vocal folds paralysis automatically using vocal folds stroboscopic images. Previously, a segmentation stage is necessary to obtain correctly and quickly the glottal space from healthy and pathological vocal folds video sequences captured by the laryngoscope. The designed segmentation method combines Gabor filtering and Chan-Vese algorithm. To evaluate the quality of the segmentation algorithms some parameters have been measured: the segmentation of the region of interest, the frames incorrectly segmented (in %) and the number of iterations used by the designed algorithm (in %). This study is specifically focused on measuring the opening angles in healthy and vocal folds with paralysis. This kind of parameters help the specialist to express a diagnosis based on something more robust than the observation of medical images. It shows a great advance in design, and in the nearby future, a complete method to diagnose vocal folds pathologies.