Scaled measurement of improvement in lip excursion (SMILE) technique. Preoperative and postoperative photographs of the left free gracilis flap powered by the masseteric branch of the trigeminal nerve. The glabella-radix-menton line bisects the intercanthal line. The iris diameter (u) measures 14 mm. This is divided by the human mean (11.7 mm) so that the x and y measurements are divided by v, a factor of 1.21. The z value is derived by Pythagorean theorem z = √(x2+ y2). Trigonometric analysis of x and y for each side enables objective calculation of smile angle and symmetry. Preoperative (A) and postoperative (B) views.
Screenshot of MatLab (MathWorks, Natick, Massachusetts) interface for automated scaled measurement of improvement in lip excursion (SMILE) analysis.
Bray D, Henstrom DK, Cheney ML, Hadlock TA. Assessing Outcomes in Facial Reanimation: Evaluation and Validation of the SMILE System for Measuring Lip Excursion During Smiling. Arch Facial Plast Surg. 2010;12(5):352-354. doi:10.1001/archfacial.2010.69
There is no current consensus for objectively evaluating postoperative smiling outcome. Several objective measurement systems have been proposed,1- 10 but, owing to their complexity, cumbersome nature, or cost, no single system has been adopted. With such variation in the reporting of results, it is difficult to make comparisons and draw conclusions about the outcome of different surgical techniques. We propose a simple facial analysis system, the scaled measurement of improvement in lip excursion (SMILE), that obviates the need for expensive equipment, facial marking, or prolonged patient cooperation. The system uses the iris diameter (corneal white-to-white diameter) as a scale reference for all facial dimensions on the same photograph. Rüfer et al1 measured the corneal white-to-white diameter in 390 healthy subjects aged 10 to 80 years with the Orbscan II system (Bausch & Lomb, Rochester, New York). They found the mean (SD) corneal diameter was 11.77 (0.42) mm. With such small variation in human iris diameter, this built-in scale—in the same plane as the oral commissure—allows “real-life” millimetric measurements of horizontal to vertical commissure excursion to be extrapolated from frontal photographs using readily available photoediting software.
Preoperative and postoperative frontal photographs of the patient at rest and with a full smile are imported into Photoshop (Adobe Systems Inc, San Jose, California). Using the measuring tool, the corneal white-to-white diameter (u) is measured horizontally through the pupillary center and divided by the standardized human average (11.7 mm) to produce a photograph-specific scaled vector (v) for an adjusted iris diameter value. This vector is then applied to additional measurements in the same photograph to calculate facial dimensions, proportions, and angles, as follows.
The medial canthi are marked electronically and connected by a straight intercanthal line. This line is bisected perpendicularly through the radix and inferiorly through the menton. The intersection of this line and the lower lip vermillion (lower vermillion midpoint) is assigned the zero point on the ruler tool. A horizontal line (x-axis) is drawn from the lower vermillion midpoint perpendicular to the glabella-radix-menton line laterally to a point that vertically bisects the lateral commissure on either side. The ruler tool is then reset on this new point, and a vertical line (y-axis) is drawn to the lateral commissure. This delineates the adjacent (x-axis) and opposite (y-axis) sides of a right-angled triangle
Pythagorean theorem calculates the hypotenuse (true smile excursion from the midline), and trigonometry, the smile angle from the horizontal midline. The same measurements are made on the contralateral side. Correlation between angles and hypotenuses on the normal side and affected side, both before and after facial reanimation, provides an objective analysis of smile symmetry and change in commissure excursion (Figure 1).
Microcalipers were used to measure the true lower vermillion midpoint to lateral commissure distance at full smile in 10 normal individuals (20 hemismiles). Clinical photographs of the same patients at full smile were imported into Photoshop, and, after a brief SMILE technique tutorial, 3 independent lay testers were asked to measure the lower vermillion midpoint to the lateral commissure distance on each side in millimeters. Having measured the iris-scale–corrected x- and y-axes, each tester calculated the hypotenuse (z) on each side using Microsoft Excel software (Redmond, Washington). Class correlation coefficients were calculated for each tester's measured distances and compared with the true excursion. A strong correlation was found for each tester, respectively (R = 0.99, 0.96, and 0.99). Each tester's calculated excursion distances over 20 independent measurements were compared against each other to test intertest reliability. There was strong correlation between the measurements of all 3 testers (tester 1 to tester 2, R = 0.96; tester 2 to tester 3, R = 0.98; tester 1 to tester 3, R = 0.88). The following week, to determine intratest reliability, the same 3 testers repeated the analysis of 4 randomly chosen photographs. The results of their second analyses were compared with their first, and a strong correlation was found for each tester (R = 0.99, 0.99, and 0.99).
From our database of 84 free gracilis transfer procedures in the past 5 years, we selected 20 procedures with subjectively excellent results and follow-up of 4 to 12 months following single-stage surgery or 12 to 18 months following second-stage surgery for SMILE technique analysis.
Measurement of lip excursion parameters x, y, and z and excursion angles were calculated using the SMILE technique and are reported in the Table. Mean paralyzed x, y, and z values increased significantly postoperatively (P < .05). Comparison of postoperative excursion distances between affected and unaffected sides revealed a significant improvement in overall symmetry (P < .05).
There were statistically significant increases in the degree of change from rest to smile in all smile parameters on the paralyzed side following successful smile surgery. Interestingly, the data also showed improved smile excursion on the healthy side.
On the paralyzed side, the mean (SD) preoperative critical change in commissure excursion (Δ z) was −0.25 (4.0) mm, increasing significantly to 11.4 (7.1) mm postoperatively (P <.001).
Using the lower vermillion midpoint as a reference enables the creation of simple Pythagorean triangles to analyze smile excursion, angle, symmetry, and volume. This mathematical model lends itself to automation. Clicking on frontal photographs in 4 anatomical sites produces calculated data in tabulated format. The preliminary interface using MatLab software (MathWorks, Natick, Massachusetts) is shown in Figure 2.
When comparing postoperative outcomes, it is easy to see the change in measured commissure position. However, the most notable change is shown by measuring the degree of change in the commissure position from rest to smile and the comparison of its position before and after surgery. Mean preoperative Δ x and Δ z values of the paralyzed side were negative secondary to the contralateral pull of the healthy side. For all values on the paralyzed side, there was a significant improvement (P < .05) in the excursion change from rest to smile both preoperatively and postoperatively. In addition, the improvement following surgery in the normal side parameters shows that we are not compromising the healthy side.
In patients with an excellent postoperative bilateral smile, we closed the change in Δ z gaps. It is possible that the Δ z gap closure between affected and unaffected sides could be the new standard for measuring successful reanimation of the smile. By focusing on the Δ z gaps, we incorporate both horizontal and vertical excursion, equally weighting both improvements. Because we base our comparisons to the healthy side, the SMILE technique is sensitive to the hypertonic, overcorrected smile, as well as the undercorrected result. In addition, our system is sensitive to an excellent static result without movement as well as a poor static result because it includes both measures at rest and with maximal smile.
In conclusion, accurate measurement of facial movement is essential to all clinicians involved in facial paralysis rehabilitation. To advance the field in such a niche subspecialty, it is essential that outcomes are reported uniformly. In addition to the creation of a “global” database, standardization of smile analysis would further this goal. To achieve widespread acceptance, the system needs to be simple, intuitive, and inexpensive. It must be reproducible across different international centers, and requires strong interrater and intrarater correlation. We believe the SMILE system is valid and satisfies these criteria. To our knowledge, it is the only smile analysis technique that permits remote measurement of digitally transmitted photographs. This development alone represents an improvement for patients, who often travel long distances for review of smile excursion, and for surgeons, who can now analyze smile outcome without the patient physically present. It also provides a useful tool for retrospective photographic facial analysis. Comparing preoperative to postoperative change in measured parameters from rest to smile may be the most useful determinant of successful facial reanimation surgery.
Correspondence: Dr Hadlock, Facial Nerve Center, Department of Otolaryngology–Head and Neck Surgery, Massachusetts Eye and Ear Infirmary, 243 Charles St, Boston, MA 02114 (firstname.lastname@example.org).
Financial Disclosure: None reported.