Analysis of facial muscles movements V02
Research type
Research Study
Full title
Development of Mathematical Index for the assessment of facial expressions using Dynamic stereophotogrammetry
IRAS ID
229233
Contact name
Mahmoud Amir Alagha
Contact email
Sponsor organisation
NHS Greater Glasgow and Clyde
Duration of Study in the UK
5 years, 0 months, 1 days
Research summary
Research Summary
Facial expressions can be altered in various pathologic conditions and malformations. Quantifying the degree of distortions of facial expressions is crucial in evaluating the impact of various treatment modalities and in monitoring over time.Assessment of facial expressions to date has been subjective which lacks reproducibility and is prone to human error. Methods include the use of grading scales, panel assessments and two dimensional (2D) still photographic and video techniques. Objective methods have been suggested which include rulers and callipers, but these methods have limitations. Evaluation of the severity of facial paralysis deformity and the effectiveness of reconstructive surgery have been attempted using a hand held ruler. The method does not describe the dynamics of facial expressions. The House-Brackmann (HB) and the Sunnybrook Facial Grading System (SFGS) are the two main scales to evaluate patients with Facial Nerve Paralysis. However these rely on the clinician to accurately and reproducibly landmark the 3D image of the animated face on several occasions. This is dependent upon precise landmark definitions for both the pre- and post-operative images
The aim of this study is to utilize an objective assessment tool, four-dimensional imaging, that provides a comprehensive assessment of the full face at rest and with various facial expressions to develop a mathematical index for the assessment of facial expressions. Also to explore the correlation between subjective grading of facial expressions and the mathematical calculation of the same facial muscle movements
Summary of Results
The aim of this study was the assessment of distorted facial muscles movements in unilateral facial palsy based on a mathematically validated subjective clinical grading systems.Material & Method
Facial paralysis patients were assessed by 7 expert assessors based on two clinical grading systems, the study introduced, for the assessment of unilateral facial palsy, the modified Sunnybrook facial palsy grading and the Glasgow Index. The reproducibility of the clinical gradings between two rating sessions was examined.
Advanced geometric morphometric approach using Dense Surface Models was applied for the analysis of the 3D facial dysmorphology over time. Asymmetries at 10 facial regions have been calculated from the 4D images of 6 expressions of the patients (rest, smile, lip purse, cheek puff, eyebrow raise, eye closure). The asymmetries calculated from 4D images were treated as the gold standard to evaluate the performance of the subjective grading systems.
Cross-correlations between the mathematical measurements and the subjective grades were calculated. The Modified Sunnybrook system assessed 8 parameters (3 at rest and 5 at individual facial expression). The Glasgow index assessed 29 parameters for the assessment of dynamic facial abnormalities with considerations for the directionality and severity of asymmetry. The similarities and dissimilarities between the two clinical assessments and to the mathematical measurements were investigated.
Results
No statistically significant differences were found in the indexing method of the Modified Sunnybrook system and, the Glasgow indexing method II following the exclusion of three parameters (GI18, GI21 at cheek puff, GI26 at eye closure).
In the Modified Sunnybrook system, the first three rest parameters showed high correlations at the cheek (-0.76) and the upper lip (-0.69) but low correlation at the eye and forehead regions (-0.42, -0.48 respectively). The first 6 rest parameters of the refined Glasgow index showed high correlations at the forehead (-0.72), cheek (-0.84) and at the corner of the mouth (-0.79) but low correlation at the eye region (-0.49). The clinical assessment of the eyes based on 4D image data was not ideal due to the 4D imaging defects on reflective surface of the cornea.The correlation coefficients between the clinical assessments of facial dynamics (5 MSB parameters and 23 RGI parameters) and the mathematical calculations of the same regions regarding to asymmetry and directional asymmetry showed a large range of variations, especially for RGI parameters assessing the directionality of asymmetry and its severity at the 5 facial expressions. Significant correlations (R > 6) were detected during:
Smile expression for assessment of
o cheek asymmetry (RGI8 -0.73, -0.67 MSB8).
o nasolabial region (RGI8 -0.67, RGI10 -0.60, MSB6 -0.64).
o upper lip asymmetry (RGI8 -0.65, MSB6 -0.66).
o lower lip (MSB6 -0.61).
o Corner of mouth asymmetry (MSB6 -0.60).
Lip purse for assessment of
o full face asymmetry (RGI13 -0.74).
o upper lip asymmetry (RGI14-15 -0.61, MSB8 -0.65).
o lower lip asymmetry (RGI14, MSB8 -0.62 for both).
o corner of mouth (RGI14 -0.62, MSB8 -0.62).
Cheek puff for the assessment of cheek asymmetry (MSB7 -0.69).
Eye closure for the assessment of forehead asymmetry (RGI27 -0.71, MSB5 -0.67).
The clinical assessment of the directionality and severity of the dynamic facial dysmorphology was sensitive enough and correlated with the objective measurements at:
Smile expression showed 68.75% at nasolabial and upper lip regions (-0.61, -0.62, respectively).
Lip purse showed 75% agreement at lower lip -0.66.Conclusion
The mathematical assessment of the dynamics of facial expressions in unilateral facial palsy using advanced geometric morphometrics provides state-of-art approach for the quantification and visualization of facial dysmorphology. It allows the accurate measurement of asymmetry at individual facial regions to underpin the contribution of each group of facial muscles and the asymmetry of expressions and facilitates the identification of the average patterns of facial asymmetry.The introduced clinical grading systems of the asymmetric facial expressions in unilateral facial palsy, the refined Glasgow Index and the Modified Sunnybrook Index, were reproducible.
The significant correlations between the clinical grading of facial palsy and the mathematical calculation of the same facial muscle movements provided a satisfactory evidence of objectivity to the clinical assessments. The refined Glasgow index provided more validated parameters for the assessment of facial palsy in comparison to the modified Sunnybrook index.
REC name
South Central - Berkshire Research Ethics Committee
REC reference
17/SC/0541
Date of REC Opinion
13 Oct 2017
REC opinion
Favourable Opinion