Diagnostically relevant facial gestalt information from ordinary photos

Abstract

Craniofacial characteristics are highly informative for clinical geneticists when diagnosing genetic diseases. As a first step towards the high-throughput diagnosis of ultra-rare developmental diseases we introduce an automatic approach that implements recent developments in computer vision. This algorithm extracts phenotypic information from ordinary non-clinical photographs and, using machine learning, models human facial dysmorphisms in a multidimensional 'Clinical Face Phenotype Space'. The space locates patients in the context of known syndromes and thereby facilitates the generation of diagnostic hypotheses. Consequently, the approach will aid clinicians by greatly narrowing (by 27.6-fold) the search space of potential diagnoses for patients with suspected developmental disorders. Furthermore, this Clinical Face Phenotype Space allows the clustering of patients by phenotype even when no known syndrome diagnosis exists, thereby aiding disease identification. We demonstrate that this approach provides a novel method for inferring causative genetic variants from clinical sequencing data through functional genetic pathway comparisons.DOI:http://dx.doi.org/10.7554/eLife.02020.001

Genetic heterogeneity in Cornelia de Lange syndrome (CdLS) and CdLS-like phenotypes with observed and predicted levels of mosaicism

“…We have not yet done a similar analysis with the other known genes, because the number of facial images is not sufficient and the typical face of the mutation-positive cases is not yet fully appreciated. Statistical approaches to the similarity of averaged faces are being developed and may ultimately replace the human gestalt assessment 34. The NIPBL -like group are clearly an interesting group on whom to focus future research, as they are likely to be enriched for mosaic cases.…”

Using facial images for the diagnosis of genetic syndromes: A survey

“…Appearance based features include learned image filters from Independent component analysis (ICA), Principal Component Analysis (PCA), Local Feature Analysis (LFA), Gabor filters, Gabor wavelet transform (GWT) [13], Scale-Invariant Feature Transform (SIFT) [14], and Local Binary Patterns (LBP) [15] are the most popular ones. Experiments demonstrate that the fusion of geometric and appearance features can provide better accuracy than using either of them alone [2][8]. …”

“…Clinical geneticists and pediatricians are closely involved in the identification of the correspondent features. 1 in 17 people worldwide suffer from genetic disorder like Down Syndrome (DS), 22q11.2 Deletion Syndome (22q11.2DS), Cri-du-chat, Cornelia de Lange, fragile X, Mucopolysaccharidosis III, Noonan, Prader-Willi, Smith-Lemli-Opitz, Sotos, Williams-Beuren and Wolf-Hirschhorn syndrome [2]. There are about 7,000 known genetic disorders, 30 to 40% of cases cause a physical change showing abnormalities in the face and skull [3].…”

“…An early diagnosis permits to provide an early medical care, alert the medical team to manage the related complications and develop their skills as fully as possible. The child will get better follow up: an urgent treatment like an operation in the heart is sometimes needed, may have digestive problems that require a lifelong special diet, all of which can benefit from early intervention [2] [8].…”

On the Construction of Intelligent Diagnostic System Cognitive Images Based

“…They are not included in the traditional computer-assistance systems support diagnostic decisions. Movement in this direction can be seen in the English study [11], unites three approaches: (a) train a computer vision algorithm for automatic annotation of 36 feature points of interest across the face on foto, (b) the use of active probabilistic models of face dysmorphism within the nosological groups in a multidimensional space, (c) machine learning using the nearest neighbor method. Clusters of patients in the space known syndromes facilitate the generation of diagnostic hypotheses.…”