Can machine learning algorithms be used to find the best tympanic membrane (eardrum) images from an endoscopic ear examination video?

Background

AI is transforming the world around it, including healthcare, with limitless possibilities for its use. Several studies have used a form of AI called machine learning to identify different ear pathologies from the best images taken with an endoscope by an expert user. However, no such study had identified the best image of the tympanic membrane (TM) from a video to subsequently allow a diagnosis to be made. This is the first step towards developing a complete machine-learning package that non-experts can use. The project aimed to collect videos of TMs and run the videos through a machine learning algorithm to find the images which showed the best view of the TM.

Impact of ENT UK Foundation funding

The Foundation grant has funded the purchase of an endoscope to facilitate the collection of videos of normal TMs.

Current Status

The endoscope has been purchased, and all data has been collected and analysed. We are now fine-tuning the dataset/neural network to improve the algorithm’s accuracy.