Utility of Facial Images in Treatment Planning for Patients With Cleft Lip/Palate: 2- and 3-Dimensional Static and Dynamic Images.

Background

To support clinical and research applications of facial imaging, a multimodal framework integrating static and dynamic facial images termed the Standardized Assessment for Facial Surgery (SAFS) was developed.

Purpose

The study purpose was to evaluate SAFS for facial feature identification and mobility in support of treatment planning for lip revision surgery.

Study design, setting, sample

This prospective cross-sectional study included subjects with repaired cleft lip/palate (CL/P), with and without lip revision recommendations, recruited from 6 craniofacial centers, and surgeons experienced in cleft care. Surgeons evaluated SAFS for subjects from institutions other than their own.

Independent variable

The independent variable was SAFS with 4 categorical levels: 2-dimensional (2D) static, 3-dimensional (3D) static, 2D dynamic facial images, and 3D dynamic objective visualizations.

Main outcome variable

The main outcome was surgeons' perspective of facial features related to appearance and mobility across successive imaging modalities coded as change or not from the previous modality. Secondary outcomes were themes from surgeons' interviews.

Covariates

Covariates for the subjects were age, sex, and unilateral versus bilateral CL/P.

Analyses

One-sided superiority binomial tests assessed whether each modality identified additional features relative to the preceding modality and changed surgical recommendations. Interviews were analyzed using grounded theory.

Results

The sample included 24 subjects (mean age, 12.3 years [SD, 4.5]); 10 (41%) were female, 16 (67%) had unilateral CL/P, and 12 (50%) identified as White, 7 (29%) as Asian, and 5 (21%) as multiracial. Eight surgeons participated. Among patients not initially recommended for surgery, surgeons changed recommendations for 6 of 7; the transition from 2D to 3D static imaging changed recommendations for 4 additional patients. Changes favored more extensive surgery for a significant overall change of 43%. Transitioning from 2D to 3D static imaging increased feature detection by more than 10% for all animations (P < .05), and from 2D dynamic video to 3D dynamic visualization yielded detection rates significantly greater than 50%.

Conclusions and relevance

Compared with clinical examination, 2D and 3D static images led surgeons to alter treatment plans and perceive greater deformity severity. Relative to 2D static imaging, 3D static and 3D dynamic visualizations improved feature detection. Thematic analysis supported SAFS as a useful diagnostic aid.