CC BY-NC-ND 4.0 · Laryngo-Rhino-Otol 2019; 98(S 02): S192
DOI: 10.1055/s-0039-1686812
Abstracts
Miscellaneous
Georg Thieme Verlag KG Stuttgart · New York

International validation of an automated algorithm to classify the severity of facial palsy

O Guntinas-Lichius
1  Univ. HNO-Klinik, Gebäude A1, Jena
,
GF Volk
2  Univ. HNO-Klinik, Jena
,
J Denzler
3  Lehrstuhl für Digitale Bildverarbeitung, Uni Jena, Jena
› Author Affiliations
This work was supported by the German Federal Ministry of Education and Research (BMBF; project IRESTRA grant no. 16SV7209).
Further Information

Publication History

Publication Date:
23 April 2019 (online)

 

Subjective classification of patients with facial palsy using grading systems like House-Brackmann-Scale or the Sunnybrook Grading scale still is the standard in clinical routine. Because subjective grading has a limited intra-rater reliability and inter-rater reliability automated classification systems are urgently needed. This would also allow a better comparison of studies (for instance to compare different types of facial nerve reconstructive surgery) and patients with different ethnic background. In Jena, Germany, we have established an interdisciplinary Facial Nerve Center dealing with different kind of facial palsies (peripheral, central palsy) and different types of therapies (conservative, surgery). We have developed and validated a video tutorial to standardize the recording of facial nerve movements in patients with different types of facial palsy. The videos allow an evaluation of all important facial nerve functions. Furthermore, we habe developed together with the chair of image analysis a prototype of a new software based on machine learning algorithms to classify facial palsies automatically based on standardized photographs. At the moment we transfer the technology on video data. We would like to perform a bi-national Chinese-German clinical trial to train the algorithm also for Chinese patients and validate the results in this international setting.