Automated detection of alveolar arches for nasoalveolar molding in cleft lip and palate treatment

Abstract

Nasoalveolar moulding (NAM) has become a widely accepted and evidence-based treatment strategy for newborns with cleft lip and palate (CLP), attempting to reduce the cleft gap and to form an appropriately shaped alveolar arch by an intraoral patient-specific NAM plate and to erect the usually flattened nostrils towards a natural nose wing occurrence. The generation of such an appropriately shaped NAM plate requires, besides 3d information of the patient’s initially cleft lip and palate, an estimated target model of the maxilla. Previous studies showed the applicability of curve-based approaches to describe the maxilla during early infancy. We have developed an automated algorithm implemented with the programming language Python, describing the alveolar arch by an approximated ellipse. Therefore, the digitalized data sets of human maxillae were aligned to a global coordinate system with a total least square method and subsequently analyzed with the curvature-based algebraic point set surfaces (APSS) algorithm. The gathered information of height ratio and curvature allows the detection of points on the alveolar segments and therewith the fit of an ellipse describing the human maxilla. In 84.5% of 193 maxilla impressions of healthy newborns the fitted ellipses described the course of the maxilla within defined margins. Applying the algorithm to 38 newborns suffering from unilateral cleft lip and palate in 76.3% the fitted ellipses bridge the CLP alveolar segments, so that a harmonic alveolar arch can be deduced. Describing the alveolar arch by one or multiple ellipses allows (i) to automatically measure the dimensions of the maxilla, (ii) to derive a growth model during early infancy, (iii) to derive a healthy harmonic arch from CLP alveolar segments and (iv) to automatically generate a basic NAM device on the basis of the virtually modified maxilla.