Optimization of chronic rhinosinusitis diagnosis using artificial intelligence

Abstract

Objective. To analyze the possibilities of artificial intelligence application in order to optimize the diagnosis of chronic rhinosinusitis. Material and methods. The review includes materials published in MEDLINE, PubMed information databases. The selection of material was carried out according to the following keywords: chronic rhinosinusitis, nasal polyps, artificial intelligence, machine learning, neural networks; with subsequent systematization in accordance with the task. Results. The review includes 18 publications that meet the inclusion criteria. The publications were devoted to the possibilities of artificial intelligence use in order to solve the problems of differential diagnosis of chronic rhinosinusitis endotypes, differential diagnosis of nasal polyps and inverted papillomas, detection of nasal polyps during endoscopy, segmentation of the paranasal sinuses, determination of the osteomeatal complex occlusion degree, differential diagnosis of fungal rhinosinusitis, assessment of sinus wall remodeling. Conclusion. Software models created using machine learning have high accuracy and low operating costs, which will allow to use them for solution of the problems of comprehensive (clinical, radiation and endoscopic) diagnosis of chronic rhinosinusitis, simplifying the doctor’s work and reducing the frequency of expensive invasive techniques (for example, biopsy with subsequent pathohistological examination) application.

Ähnliche Arbeiten