Publications

Automatic analysis of asymmetry in facial paralysis patients using landmark-based measures

Abstract

Methods
This study was approved by the IRB of Keck School of Medicine (HS-19-00076). Two data sets were utilized:(1) the KECK facial palsy data set (KECK data set) and (2) the MEEI facial palsy photo and video data set (MEEI data set). 4 The facial landmarks from the KECK and MEEI data sets were automatically extracted using the opensource dlib toolkit 5 and the MEEI-specific dlib model, respectively. 3 These landmarks were used to derive the differences of the left–right (LR) parts of the face for eye opening symmetry ratios, brow elevation, mid-facial symmetry, and oral commissure symmetry. The five measurements, illustrated in Figure 1, are referred to as the Landmark-based facial asymmetry (L-FAM) measures (Supplementary Video S1).

Date
December 1, 2022
Authors
Digbalay Bose, Krishna Somandepalli, Tymon Tai, Courtney Voelker, Shrikanth Narayanan, Amit Kochhar
Journal
Facial Plastic Surgery & Aesthetic Medicine
Volume
24
Issue
6
Pages
491-493
Publisher
Mary Ann Liebert, Inc., publishers