Delineation and Visualisation of Congenital Abnormality using 3D Facial Images

INTRODUCTION One in fifty children is born with significant congenital abnormality [1]. Some have multiple anomalies constituting a specific dysmorphic syndrome. Of 5,000 or so dysmorphic syndromes, over 700 involve dental, oral or craniofacial differences. Dysmorphic craniofacial features include extended distance between eyes (telecanthus) or pupils (hypertelorism); inner/outer eye corners at different levels (sloping palpebral fissures); unusual nose width (interalar distance), bridge of nose, skull shape or position/orientation of ears; occurrence of skin folds; and, drooping eyelids (ptosis). Syndrome diagnosis is based on clinical observation of abnormal body parts and proportions, and unusual appearance. Early diagnosis is important if it is to guide clinical investigation and treatment, and genetic counseling of parents. The condition focused on here, Noonan syndrome [2], involves short stature (72%), mild to moderate learning disability (61%), congenital heart disease, low set ears, hypertelorism, neck webbing, sloping palpebral fissures and ptosis. Noonan syndrome affects approximately 1 child in 2000 and on average a diagnosis is made by four years of age. Figure 1 illustrates characteristic facial features of Noonan syndrome. Visual clues are important in dysmorphology. The age of individuals being assessed is also significant, since some features are present at birth and others evolve with time [3]. Recognition of facial characteristics of particular syndromes lags behind somatic and behavioural characteristics and may delay diagnosis [4]. Characteristic features may not develop simultaneously but become apparent as a child develops, and may even become less obvious in later life [5]. In Noonan syndrome, physical manifestations in adults may be subtle and some without a known heart defect or other medically significant problem may have gone unrecognised [6]. Rarity of syndrome occurrence is a problem for clinicians with infrequent exposure to congenital disorders. Thus, an analysis of craniofacial features and their change over time is important in the delineation of dysmorphic syndromes and training in their detection. For this reason, more than 10,000 2D photographs already form an important component of the London Dysmorphology Database (LDDB), which catalogues some 3,500 dysmorphic syndromes [7]. Although 2D photographs can depict typical craniofacial features of syndromes, there are problems with pose that are eliminated by employing 3D images. Once captured, 3D images can be rotated and viewed from any angle to support clinical assessment, whereas 2D photographs are fixed in the same pose forever. We report on 6 months’ use of a photogrammetric scanner to capture 3D facial shape and appearance and give an analysis of data gathered from 62 children, of whom 22 have Noonan syndrome. The objective is to evaluate the use of 3D images to visualise and delineate facial features in dysmorphic syndromes. The next section introduces craniofacial assessment [8], summarises our data collection and describes problems encountered. The penultimate section is a preliminary analysis of the data collected. The final section contains an interpretation of the results and a summary of future plans. All image manipulations, other than capture, are provided by ShapeFind, a shape analysis system developed in-house.