Expert recommendations on collection and annotation of otoscopy images for intelligent medicine

Middle and outer ear diseases are common otological diseases worldwide. Otoscopy and otoendoscopy examinations are essential first steps in the evaluation of patients with otological diseases. Misdiagnosis often occurs when the doctor lacks experience in interpreting the results of otoscopy or otoendoscopy, leading to delays in treatment or complications. Using deep learning to process otoscopy images and developing otoscopic artificial-intelligence-based decision-making systems will become a significant trend in the future. However, the uneven quality of otoscopy images is among the major obstacles to development of such artificial intelligence systems, and no standardized process for data acquisition, and annotation of otoscopy images in intelligent medicine has yet been fully established. The standards for data storage and data management are unified with those of other specialties and are introduced in detail here. This expert recommendation criterion improved and standardized the collection and annotation procedures for otoscopy images and fills the current gap in otologic intelligent medicine; it would thus lay a solid foundation for the standardized collection, storage, and annotation of otoscopy images and the application of training algorithms, and promote the development of automatic diagnosis and treatment for otological diseases. The full text introduced image collection (including patient preparation, equipment standards, and image storage), image annotation standards, and quality control.

Abstract licence: CC BY-NC-ND

1. 25.

   Reporting guidelines for clinical trial reports for interventions involving artificial intelligence: the CONSORT-AI Extension · 2020

2. 27.

   Guidelines for clinical trial protocols for interventions involving artificial intelligence: the SPIRIT-AI extension · 2020

27 more references, whose wording the paper’s licence keeps private.

A person counts once a day per paper. We keep no cookie and no address: a reader is a hash that changes every day. Any month under five reads shows no number.