RESUMO
OBJECTIVES: This study aims to assess utility of pixel-valued movement software in detecting arytenoid dislocation preoperatively. STUDY DESIGN: This is a retrospective analysis. METHODS: Twenty-seven patients diagnosed with unilateral arytenoid dislocation were included. Diagnosis of arytenoid dislocation was confirmed by lack of vocal fold paralysis on preoperative laryngeal electromyography and by intraoperative findings of cricoarytenoid dislocation. A region-tracking software algorithm developed by Zhuang et al was used to analyze 27 preoperative endoscopic videos of patients diagnosed with arytenoid dislocation. Vector analysis measuring cuneiform movement during inspiration was used as an indirect measure of arytenoid movement. Values were normalized using vocal fold length. Two raters blinded to diagnosis of arytenoid dislocation measured vocal fold length and cuneiform movement on both the dislocated and the nondislocated sides. RESULTS: A Wilcoxon signed-rank test indicated that the mean pixel-valued cuneiform movement and standard deviation (SD) were greater for nondislocated (159.24, SD = 73.35) than for dislocated (92.49, SD = 72.11) arytenoids (Z = 3.29, P = 0.001). The interrater correlation coefficient was 0.87 for the dislocated side and 0.75 for the nondislocated side. The intrarater correlation coefficient was 0.87 for the dislocated side and 0.91 for the nondislocated side. The receiver operating characteristic curve revealed an area under the curve between 0.76 and 0.83 (95% confidence interval 0.63-0.90). Analysis by the first and second raters revealed misdiagnosis of laterality of arytenoid dislocation in four and six patients, respectively. CONCLUSIONS: The software program developed by Zhuang et al provides a high-degree of precision, with good interrater and intrarater correlation coefficients. However, high rates of misdiagnosis of arytenoid dislocation and the laborious analysis process using this software program make it of limited utility as a clinical diagnostic tool in its present state.
