Facial mimicry after conservative parotidectomy: a three-dimensional optoelectronic study


Facial nerve paralysis is one of several possible complications following conservative parotidectomy. To assess three-dimensional facial movements non-invasively in patients with unilateral facial palsy following parotidectomy for benign tumours, the three-dimensional coordinates of 21 soft-tissue facial landmarks were recorded in 32 patients (21 HB I, 5 HB II, 6 HB III-IV; 3 months post-surgery follow-up), and 40 control subjects, during the performance of facial movements (smile, ‘surprise’, eye closure, single eye closure). For all symmetric animations, control subjects had larger total mobility than patients; mobility progressively decreased in patients with larger clinical grades. For asymmetric eye closures, HB I patients and control subjects had similar total movements, while HB II patients had smaller movements, especially for the paretic side eye closure; smaller total movements were found in HB III-IV patients. The method allowed the quantitative detection of alterations in facial movements. Significant differences between patients and control subjects in the magnitude and asymmetry of movements were found.

The movements of facial muscles play a key role in social communication and interaction, as well as in speech articulation . Altered facial movements can also cause damage to visceral functions, with reduced corneal protection and labial competence, and insufficient cheek and lip mobility during feeding .

Various pathologies can alter the activity of facial muscles in static and dynamic conditions, examples are central and peripheral nervous system diseases, muscle disorders, drug administration, scars . Facial and mandibular movements may also be altered by surgical procedures, either by damage to the facial muscles and soft tissues, or by peripheral nerve lesions .

Facial nerve paralysis is a possible complication following parotidectomy. When surgery is performed for benign diseases such as pleomorphic adenoma, temporary nerve palsy has been reported in 18–44% of patients, depending on the type of surgical removal (total conservative parotidectomy vs. superficial parotidectomy) patients regain normal function within 1 year of surgery, but permanent facial palsy occurs in 1–4% .

The best method for the quantification of facial motion and of its impairments is debated. Both clinical and instrumental assessments have been proposed to grade spontaneous and instructed facial movements . While clinical assessments focus on total and local facial motion, synkinesis and movement asymmetries , quantitative methods can assess the movements of selected facial landmarks and their trajectories .

The authors have developed a method for assessing facial movements three-dimensionally using an optoelectronic motion analyser . The method is reliable, and correctly detects total and local motion during the performance of standardized symmetric and asymmetric facial animations . Sex- and age-related reference values for several non verbal movements have recently been published .

The aim of the current investigation was quantitatively to assess facial movements in a group of patients with unilateral facial palsy following parotidectomy for benign tumours. The patients had various grades of facial palsy according to clinical scales , and were recorded while performing a set of standardized symmetric and asymmetric facial animations, and the displacement of selected facial landmarks was measured in three dimensions. Data were compared to those previously collected in healthy men and women . The authors hypothesized that there are differences between patients and healthy people in the magnitude of facial movements, as well as in their asymmetry.

Materials and methods

32 patients (13 men, 19 women; age range 38–66 years, mean 56.2 years, SD 11.1), submitted to unilateral parotidectomy with complete preservation of the facial nerve for benign tumours were analysed approximately 3 months after surgery. The patients were separated into three groups according to a clinical classification made following the House-Brackmann (HB) scale : 21 patients were classified HB grade I (10 men, 11 women, mean age 56.4 years, SD 13.8); 5 patients were classified HB grade II (1 man, 4 women, mean age 54.4 years, SD 17.2); and 6 patients were classified HB grade III–IV (2 men, 4 women, mean age 55 years, SD 11.9).

The diagnoses were: pleomorphic adenoma (14 patients with HB I; 2 patients with HB II; 4 patients with HB III–IV), Warthin tumour (7 patients with HB I; 2 patients with HB II; 1 patient with HB III–IV), cavernous hemangioma (1 patient HB II) and schwannoma (1 patient with HB III–IV).

All patients came from the Ear Nose Throat Unit of an oncology public hospital. After the nature and possible risks of the study had been described, written informed consent was obtained from each patient. The protocol used in the current study was not invasive, and did not involve dangerous or painful activities, in accord with the Helsinki Declaration.

Data collected from the patients were compared with those previously obtained in a group of 40 healthy adults (20 men, 20 women) of the same ethnic group (white Caucasians), and aged 20–50 years . All subjects had clinically normal facial function, no previous facial trauma, paralysis or surgery, and no known neurological diseases.

The same protocol previously described was used for data collection. In brief, facial movements were recorded using an optoelectronic three-dimensional (3D) motion analyser with a 60 Hz sampling rate (SMART System, BTS, Milano, Italy). The instrument uses 6 high-resolution infrared sensitive charge-coupled device video cameras coupled with a video processor that define a working volume of 44 (width) cm × 44 (height) cm × 44 (depth) cm; metric calibration and correction of optical and electronic distortions are performed before each acquisition session using a 20 cm wand, with a resulting mean dynamic accuracy of 0.121 mm (SD 0.086), corresponding to 0.0158% .

The patient sat inside the working volume on a stool, and was asked to perform a series of standardized facial movements. During the execution of the movement, for any camera special software identified the coordinates of 21 passive markers positioned on facial landmarks ( Fig. 1 ). Subsequently, all the coordinates were converted to metric data, and a set of 3D coordinates for each landmark in each frame that constituted each animation was obtained.

Fig. 1
Soft tissue markers used for the analysis of facial movements (tr, trichion; n, nasion; prn, pronasale; ls, labiale superius; sl, sublabiale; pg, pogonion; sci, right and left superciliare; ex, right and left exocanthion; or, right and left orbitale; ac, right and left nasal alar crest; ch, right and left cheilion; li, right and left lower lip points halfway between cheilion and sublabiale; t, right and left tragion; v, vertex).

For each patient, a set of 2 mm squared reflective markers was positioned in correspondence with 21 soft tissue landmarks ( Fig. 1 ). The positions of the markers were carefully controlled to avoid interference with facial movements.

The three head markers t r , t l and v defined a head reference plane that was used to eliminate head movements mathematically during facial animations, and to standardize head position within and between subjects.

Each patient performed seven standardized, maximum facial animations from rest : instructed (maximum) smile (bite on the back teeth, smile as much as possible, and then relax); free (natural) smile; ‘surprise’ with closed mouth (bite on the back teeth, make a surprise expression without opening the mouth, with a prevalent movement of the forehead and eyes); ‘surprise’ with open mouth (make a surprise expression opening the mouth, with a global facial movement); symmetrical eye closure; right side eye closure (maximal closure of the eye); left side eye closure (maximal closure of the eye). Each animation was explained and shown to the patients, and they were allowed to practise before data acquisition. For each patient, three repetitions of each expression were recorded without modifications of the marker positions.

Data analysis

The same method detailed by Sforza et al. was used. This included the subtraction of head and neck movements from the raw facial movements using the three cranial markers. Therefore, only movements occurring in the face (activity of mimic muscles, and mouth opening during the ‘surprise with open mouth’ expression) were considered further. Subsequently, for each of the 18 facial markers, the 3D movements during each facial animation were computed, and the modulus (intensity) of the 3D vector of maximum displacement from rest was calculated. The origin of axes was set in the nasion.

The total facial movement was obtained as the sum of the movement of the 18 facial markers: the larger the value, the larger the facial movement.

In a previous study performed in healthy adults , the authors found that facial size did not significantly influence facial movements for six of the seven facial animations analysed. Only instructed smile was significantly related to facial size (total facial movement = 0.793 × facial size − 73.586 mm, p = 0.025). The relevant values were therefore corrected by subtracting from each total facial movement the difference between the regression line and the mean estimate of facial size .

Considering the unilateral facial palsy in patients with a left-side paresis, all paired landmarks were mirrored on the other side of the face, and all movements were considered relative to the healthy or paretic side.

To assess differential movements between the two hemi-faces, percentage indices of asymmetry were computed as: (paretic side displacement − healthy side displacement)/(paretic side displacement + healthy side displacement) × 100; in particular, markers sci, ex, or, gave the eye asymmetry index; and markers ch, li gave the mouth asymmetry index. The indices range between −100 (complete healthy side prevalence during the movement) and +100 (complete paretic side prevalence) .

Method error

Within- and between-session repeatability had been assessed in healthy subjects . To assess the within-session error, the technical error of the measurement (random error, TEM) was computed separately for each sex, movement and landmark (three repetitions of each expression). TEM for the single landmarks ranged between 0.3 and 9.42 mm (on average 0.5–3.38 mm), showing sufficient reproducibility. To assess the between-session error, the standard deviation between the mean displacements of each landmark (four independent sessions) was computed for each movement. All movements, except left eye closure, had standard deviations lower than 1 mm.

Statistical calculations

For each patient, the three series of facial animations were averaged. Descriptive statistics were obtained for the maximum displacement of each marker, the total facial movement, and the asymmetry indices separately for the three palsy groups (HB I, HB II, and HB III–IV). To assess if the asymmetry indices significantly deviated from the expected value of 0, Student’s t -tests for paired samples (HB I group) and Wilcoxon signed rank tests (HB II and III–IV groups) were carried out.

To compare the asymmetry indices, the single landmark displacements and the total facial movements computed in the patients with those obtained in control subjects, from the reference values collected in the healthy subjects were studied; 95% confidence limits were computed . The level of significance was set at 5% ( p < 0.05).


On average, during the execution of the two asymmetric facial movements (single side eye closures), all patients had movements of the orbital landmarks that were smaller on the side being moved, and larger on the contralateral side, than those recorded in the healthy subjects ( Figs. 2 and 3 ). A similar effect was seen for the ala nasi and cheilion landmarks, while no major differences were observed in the other parts of the face. Overall, HB I patients had total movements very similar to those found in control subjects, while HB II patients had smaller movements, especially when the eye of the paretic side was closed ( Tables 1 and 2 ; total mobility smaller than the 95% confidence limits of the healthy controls). Even smaller total movements were found in HB III–IV patients.

Fig. 2
Paretic side eye closure, single landmark displacement in patients with unilateral facial palsy. Triangle, HB I; square, HB II; diamond, HB III–IV; bars 95% confidence limits for control subjects.

Fig. 3
Healthy side eye closure, single landmark displacement in patients with unilateral facial palsy. Triangle, HB I; square, HB II; diamond, HB III–IV; bars 95% confidence limits for control subjects.

Table 1
Asymmetry indices (%) and total facial mobility (mm) during the execution of standardized movements.
Paretic side eye closure Healthy side eye closure Eye closure Max smile Free smile Surprise-closed mouth Surprise-open mouth
HB I Asymmetry eye Mean 43.38 * −46.05 * −1.22 −0.44 −2.59 −1.28 −0.04
SD 17.81 21.71 11.19 11.99 9.47 10.92 10.66
Asymmetry mouth Mean 23.85 * −24.24 * −4.94 −2.21 −2.08 2.58 0.10
SD 17.31 23.74 16.18 9.62 11.51 8.85 3.68
Total mobility Mean 64.73 61.72 60.05 69.82 92.63 48.04 153.31
SD 36.86 21.43 47.66 24.21 18.47 21.00 58.32
HB II Asymmetry eye Mean 44.36 * −42.84 * −6.84 6.32 −0.03 −6.70 −1.15
SD 15.54 13.35 11.43 12.11 10.19 12.21 9.23
Asymmetry mouth Mean 36.78 * −27.99 −0.63 −0.35 −4.67 −0.18 −2.32
SD 10.19 24.08 28.25 13.87 6.77 18.96 4.24
Total mobility Mean 47.97 65.25 65.59 63.31 85.57 32.79 142.11
SD 20.38 37.97 36.74 19.56 17.36 14.85 29.82
HB III–IV Asymmetry eye Mean −18.38 −54.14 * −13.65 −11.25 −17.26 −25.22 −18.16
SD 29.39 16.24 8.02 28.88 21.86 14.42 12.99
Asymmetry mouth Mean 14.12 −18.94 14.69 −25.51 −26.61 1.66 −10.22
SD 24.35 22.85 9.11 13.38 19.24 15.31 7.55
Total mobility Mean 49.11 45.50 74.27 72.42 66.97 52.52 141.06
SD 20.88 23.07 6.48 16.76 23.89 19.81 59.45
Controls Asymmetry eye Mean 41.27 * −44.14 * −1.37 3.12 1.71 1.87 1.88
SD 15.17 16.32 7.56 11.17 8.42 8.57 9.64
Asymmetry mouth Mean 13.42 * −16.81 * −2.61 3.22 0.04 −3.34 * −2.22 *
SD 21.29 18.97 11.70 10.70 8.46 8.51 3.72
Total mobility Mean 67.53 66.30 92.43 85.13 104.48 68.97 162.99
SD 24.15 27.43 31.47 21.53 18.94 30.39 57.50
Asymmetry indices, negative values: prevalent healthy (patients)/left side (control subjects); positive values: prevalent paretic (patients)/right side (control subjects).

* Significantly different from the expected value of 0 (Student’s t -test for paired samples, p < 0.05).

Significantly different from the expected value of 0 (Wilcoxon signed rank test, p < 0.05).

Table 2
Asymmetry indices and total facial mobility compared with normal subjects.
Paretic side eye closure Healthy side eye closure Eye closure Max smile Free smile Surprise-closed mouth Surprise-open mouth
HB I Asymmetry eye <95% CL
Asymmetry mouth >95% CL >95% CL >95% CL >95% CL
Total mobility <95% CL <95% CL <95% CL <95% CL
HB II Asymmetry eye >95% CL <95% CL <95% CL
Asymmetry mouth >95% CL >95% CL >95% CL
Total mobility <95% CL <95% CL <95% CL <95% CL <95% CL <95% CL
HB III–IV Asymmetry eye <95% CL <95% CL <95% CL <95% CL <95% CL <95% CL <95% CL
Asymmetry mouth <95% CL >95% CL >95% CL <95% CL <95% CL >95% CL <95% CL
Total mobility <95% CL <95% CL <95% CL <95% CL <95% CL <95% CL <95% CL
Only gold members can continue reading. Log In or Register to continue

Jan 26, 2018 | Posted by in Oral and Maxillofacial Surgery | Comments Off on Facial mimicry after conservative parotidectomy: a three-dimensional optoelectronic study
Premium Wordpress Themes by UFO Themes