A new method for the assessment of adenoid hypertrophy: Respirdynamicsgram (RDG)

Abstract

Adenoid hypertrophy (AH) is a common cause of airway obstruction in children, which is the etiology of changes in facial growth, obstructive sleep apnea syndrome, and/or more serious problems. Most sequelae can be avoided or reversed by timely diagnosis and treatment. It is difficult to frequently assess the changes in adenoids among children using the commonly available diagnostic tools owing to invasiveness and radiation exposure. A new method called respirdynamicsgram (RDG) is proposed for the assessment and diagnosis of AH. Specifically, RDG is generated by visualizing the respiratory dynamics extracted from nasal airflow through dynamic learning. The morphology of RDG remarkably differs between children with adenoidal hypertrophy (irregular shapes) and healthy children (regular butterfly-like shapes). Three features are utilized to represent the morphology of RDG, including temporal heterogeneity, spatial heterogeneity, and wing shape. These morphology features of RDG are fed into the linear support vector machine for the AH classification task. As a result, the proposed method achieves a mean accuracy of 86.6% over 100 resample folds on a total of 74 subjects. It is shown that RDG has the capability of distinguishing children with AH from healthy children. As it is simple, non-invasive, and secure, it is hopeful that RDG may become an effective method for timely and frequent assessment and diagnosis of AH.