CC BY-NC-ND 4.0 · Laryngorhinootologie 2019; 98(S 02): S195
DOI: 10.1055/s-0039-1686848
Abstracts
Salivary Glands/Thyroid Gland

Machine learning based classification of facial palsies using standard still photografies

O Guntinas-Lichius
1   Univ. HNO-Klinik, Gebäude A1, Jena
,
O Mothes
2   Lehrstuhl Digitale Bildverarbeitung, Friedrich-Schiller-Universität Jena, Jena
,
GF Volk
3   Univ. HNO-Klinik, Jena
,
CM Klingner
4   Univ. Klinik für Neurologie, Jena
,
OW Witte
4   Univ. Klinik für Neurologie, Jena
,
P Schlattmann
5   Univ. Institut für Medizinische Statistik, Informatik und Datenwissenschaften, Jena
,
J Denzler
2   Lehrstuhl Digitale Bildverarbeitung, Friedrich-Schiller-Universität Jena, Jena
› Author Affiliations
› Further Information
Also available at
 
 

    Introduction:

    An automated, objective, fast and simple system for the classification of facial palsy (FP) is urgently needed for the use in clinical routine.

    Methods:

    233 patients with unilateral peripheral facial palsy (in total 4572 standard still photographs using 9 posed facial expressions) were subjectively rated by an observer and objectively rated applying a machine learning approach including the Supervised Descent Method. All photograph series were rated according to House-Brackmann grading scale (HB), Sunnybrook grading system (SB), and Stennert index (SI).

    Results:

    First assessment was performed after a median time of 6 days after onset. At first examination, the median objective HB, total SB, and total SI were grade 3, 45, and 5, respectively. The best correlation between the subjective and objective grading was seen for the SB movement score, total SB, and for the SI movement score (r = 0.746; r = 0.698; r = 0.732, respectively). No sufficient agreement was found between subjective and objective HB grading (Test for symmetry 80.61, df015, p < 0.001, weighted kappa =-0.0105; 95% confidence interval [CI]=-0.0542 to 0.0331; p = 0.6541). Also no agreement was found between subjective and objective total SI (test for symmetry 166.37, df055, p < 0.001). The multinomial logistic regression showed that the probability for higher SI scores was higher for subjective compared to objective classification (Odds Ratio = 1.608; CI = 1.202 to 2.150; p = 0.0014). The best agreement was observed between subjective and objective SB (Intraclass coefficient ICC = 0.34645).

    Conclusions:

    Automated SI grading delivered with fair agreement fast and objective global and regional data on facial motor function. We recommend applying this new tool in clinical routine and clinical trials.


    #
    Prof. Dr. med. Orlando Guntinas-Lichius
    Univ. HNO-Klinik, Gebäude A1,
    Am Klinikum 1, 07747
    Jena

    Publication History

    Publication Date:
    23 April 2019 (online)

    © 2019. The Author(s). This is an open access article published by Thieme under the terms of the Creative Commons Attribution-NonDerivative-NonCommercial-License, permitting copying and reproduction so long as the original work is given appropriate credit. Contents may not be used for commercial purposes, or adapted, remixed, transformed or built upon. (https://creativecommons.org/licenses/by-nc-nd/4.0/).

    Georg Thieme Verlag KG
    Stuttgart · New York