A Novel Ear Impression-Taking Method Using Structured Light Imaging and Machine Learning: A Pilot Proof of Concept Study with Patients' Feedback on Prototype.

INTRODUCTION: Taking an ear impression is a minimally invasive procedure. A review of existing literature suggests that contactless methods of scanning the ear have not been developed. We proposed to establish a correlation between external ear features with the ear canal and with this proof of concept to develop a prototype and an algorithm for capturing and predicting ear canal information. METHODS: We developed a novel prototype using structured light imaging to capture external images of the ear. Using a large database of existing ear impression images obtained by traditional methods, correlation analyses were carried out and established. A deep neural network was devised to build a predictive algorithm. Patients undergoing hearing aid evaluation undertook both methods of ear impression-taking. We evaluated their subjective feedback and determined if there was a close enough objective match between the images obtained from the impression techniques. RESULTS: A prototype was developed and deployed for trial, and most participants were comfortable with this novel method of ear impression-taking. Partial matching of the ear canal could be obtained from the images taken, and the predictive algorithm applied for a few sample images was within good standard of error with proof of concept established. DISCUSSION: Further studies are warranted to strengthen the predictive capabilities of the algorithm and determine optimal prototype imaging positions so that sufficient ear canal information can be obtained for three-dimensional printing. Ear impression-taking may then have the potential to be automated, with the possibility of same-day three-dimensional printing of the earmold to provide timely access.

Classification, registration and segmentation of ear canal impressions using convolutional neural networks.

Today, fitting bespoke hearing aids involves injecting silicone into patients' ears to produce ear canal molds. These are subsequently 3D scanned to create digital ear canal impressions. However, before digital impressions can be used they require a substantial amount of effort in manual 3D editing. In this article, we present computational methods to pre-process ear canal impressions. The aim is to create automation tools to assist the hearing aid design, manufacturing and fitting processes as well as normalizing anatomical data to assist the study of the outer ear canal's morphology. The methods include classifying the handedness of the impression into left and right ear types, orienting the geometries onto the same coordinate system sense, and removing extraneous artifacts introduced by the silicone mold. We investigate the use of convolutional neural networks for performing these semantic tasks and evaluate their accuracy using a dataset of 3000 ear canal impressions. The neural networks proved highly effective at performing these tasks with 95.8% adjusted accuracy in classification, 92.3% within 20° angular error in registration and 93.4% intersection over union in segmentation.