Lip Shape Analysis in Patients with Repaired Cleft Lip
Michelle Novoa, MS – Craniofacial Research Fellow
Kayla Prezelski, MSE, Cortney Van’t Slot, MS, CCC-SLP, Rami R. Hallac, PhD, Alex A. Kane, MD, James R. Seaward FRCS (Plast.)
Introduction:
Patients with cleft lip may have abnormal lip movements secondary to altered muscle anatomy and post-operative scarring.
•Traditional 2D and 3D imaging methods capture static expressions but fail to reflect dynamic motion.
4D imaging (3D over time) allows real-time visualization of facial movements during speech.
•4D analysis of lip movement during speech can:
oHelp evaluate functional outcomes
oGuide further intervention
oProvide objective, quantifiable measures
Purpose: To analyze dynamic lip movement in children with repaired cleft lip versus controls using 4D stereophotogrammetry and shape analysis.
Shape Analysis Challenge: developing methods that can accurately and efficiently interpret how facial shapes change over time, across specific regions of interest, and between different patient groups, given the massive and complex 4D datasets.
Previous Work analyzed:
•Left to right facial asymmetry at different timepoints during facial expressions (e.g. Smiling)
•Magnitude and displacement of key landmarks across different facial expressions
•Vector analysis, measuring both magnitude and direction of facial movements
This study analyzed:
•Procrustes shape analysis on lip and perioral regions during 4 different speech sounds
•Stretch and compression of facial skin during target speech sounds.
Methods and materials:
Age range: 6-14 years old
Patient Group: Non-syndromic patients with repaired unilateral cleft lip, with or without cleft palate
Control Group: Individuals without craniofacial or speech conditions
•4D imaging captured synchronized video and audio while subjects recited:
“Joshua Points to Every Brown Feather.”
•Sentence phonemes to increase lip movement included:
/p/, /b/, /f/ and /v/
•Processing of 4D imaging via 3dMD tempus software generated participant face masks from a standard mask.
•Meshes were then aligned
•Average mesh created for each group for group comparisons
•Mesh regions of interest (ROIs) were selected for analysis
Analysis & Algorithms
•Procrustes: allows us to compare two shapes in space by aligning them through translation, rotation and scaling to achieve the best possible fit 3 dimensionally.
oRequires a well-defined shape for analysis
•T-tests: identified statistically significant differences
•Traveling Salesman Algorithm: Connects facial landmarks into closed loops. Used to create shapes for Procrustes Analysis.
•Stretch/Compression: Tracks change in triangle area on the mesh surface
Results:
Procrustes: P values for T tests between the cleft and control groups for all the ROIs in our 4 speech sounds.
•Statistically significant motion differences in the cleft upper lip roll for plosive sounds “Brow” and “Poi”, as well as differences in multiple dimensions and rotation axes in both sides of the lower lip.
•Also, “Poi” sound showed significant differences in both commissures and non-cleft lower vermillion. In contrast, “Brow” had more statistically significant differences in both L & R upper vermillion.
•For labiodental sounds in the upper lip, there was nothing significant for ‘vry’ but for ‘fea’ there were significant differences in the way the lip moves left to right and the way it rotates in the yaw axis.
Stretch/Compression:
•Non-cleft upper lip showed greater movement—possible compensatory mechanism
•Statistically significant stretch patterns varied by sound and region
Complications:
•4D imaging of facial expressions reveals persistent motion differences between control and cleft lip patients, even after repair.
•Asymmetry is greater and more uneven in cleft patients, especially during dynamic speech, emphasizing the need for dynamic functional assessments in post-surgical evaluation
Acknowledgements: Analytical Imaging & Modeling Center
