Cochlear nerve canal stenosis and associated semicircular canal abnormalities in paediatric sensorineural hearing loss: a single centre study

Training and validation of a deep learning U-net architecture general model for automated segmentation of inner ear from CT

BACKGROUND

The intricate three-dimensional anatomy of the inner ear presents significant challenges in diagnostic procedures and critical surgical interventions. Recent advancements in deep learning (DL), particularly convolutional neural networks (CNN), have shown promise for segmenting specific structures in medical imaging. This study aimed to train and externally validate an open-source U-net DL general model for automated segmentation of the inner ear from computed tomography (CT) scans, using quantitative and qualitative assessments.

METHODS

In this multicenter study, we retrospectively collected a dataset of 271 CT scans to train an open-source U-net CNN model. An external set of 70 CT scans was used to evaluate the performance of the trained model. The model's efficacy was quantitatively assessed using the Dice similarity coefficient (DSC) and qualitatively assessed using a 4-level Likert score. For comparative analysis, manual segmentation served as the reference standard, with assessments made on both training and validation datasets, as well as stratified analysis of normal and pathological subgroups.

RESULTS

The optimized model yielded a mean DSC of 0.83 and achieved a Likert score of 1 in 42% of the cases, in conjunction with a significantly reduced processing time. Nevertheless, 27% of the patients received an indeterminate Likert score of 4. Overall, the mean DSCs were notably higher in the validation dataset than in the training dataset.

CONCLUSION

This study supports the external validation of an open-source U-net model for the automated segmentation of the inner ear from CT scans.

RELEVANCE STATEMENT

This study optimized and assessed an open-source general deep learning model for automated segmentation of the inner ear using temporal CT scans, offering perspectives for application in clinical routine. The model weights, study datasets, and baseline model are worldwide accessible.

KEY POINTS

A general open-source deep learning model was trained for CT automated inner ear segmentation. The Dice similarity coefficient was 0.83 and a Likert score of 1 was attributed to 42% of automated segmentations. The influence of scanning protocols on the model performances remains to be assessed.

[Cochlear nerve canal stenosis: a review of recent research]

Some patients with severe-profound sensorineural hearing loss （SNHL） with normal cochlear anatomical structure received cochlear implantation （CI） and the hearing and speech rehabilitation effect was not ideal. Through retrospective analysis, it was found that some of these patients had cochlear never canal （CNC） stenosis, or atresia in severe cases.This article reviews the development of the CNC, the diagnostic criteria of CNC stenosis and the results of hearing and speech rehabilitation in these patients after CI.

Cochlear Nerve Deficiency in Pediatric Unilateral Hearing Loss and Asymmetric Hearing Loss

INTRODUCTION

The rates of cochlear nerve abnormalities and cochlear malformations in pediatric unilateral hearing loss (UHL) are conflicting in the literature, with important implications on management. The aim of this study was to investigate the incidence of cochlear nerve deficiency (CND) in pediatric subjects with UHL or asymmetric hearing loss (AHL).

METHODS

A retrospective chart review of pediatric subjects <18 years of age evaluated for UHL or AHL with fine-cut heavily T2-weighted magnetic resonance imaging (MRI) between January 2014 and October 2019 (n = 291) at a tertiary referral center was conducted. MRI brain and computed tomography temporal bone were reviewed for the presence of inner ear malformations and/or CND. Status of the ipsilateral cochlear nerve and inner ear was evaluated. Pure tone average (PTA) at 500, 1,000 and 2,000 Hz was assessed.

RESULTS

204 subjects with UHL and 87 subjects with AHL were included. CND (aplasia or hypoplasia) was demonstrated in 61 pediatric subjects with UHL (29.9%) and 10 with AHL (11.5%). Ipsilateral cochlear malformations were noted in 25 subjects with UHL (12.3%) and 11 with AHL (12.6%), and ipsilateral vestibular malformations in 23 (11.3%) and 12 (13.8%) ears, respectively. Median PTA was statistically significantly higher in ears with CND (98.33) than ears with normal nerves (90.84).

DISCUSSION/CONCLUSION

Imaging demonstrated a high incidence of inner ear malformations, particularly CND, in pediatric subjects with UHL. Auditory findings indicated CND cannot be ruled out by thresholds alone as some CND ears did demonstrate measurable hearing. Radiologic evaluation by MRI should be performed in all patients within this population to guide counseling and management of hearing loss based on etiology, with implications on candidacy for cochlear implantation.

Partial Duplication of the Lateral Semicircular Canal—A Novel Anatomical Malformation in a Child with Barakat Syndrome

Several known genetic causes of sensorineural deafness are associated with dysplasia of inner ear structures, including the cochlea and labyrinth. Here, we present a child with Barakat syndrome and sensorineural hearing loss, found to have multiple inner ear anomalies including partial duplication of the posterior limb of the left lateral semicircular canal. To our knowledge, duplication of the semicircular canal has not previously been reported. This finding expands our understanding of the range of anatomical variations observed in congenital inner ear malformations, and further characterizes the phenotypic manifestations of Barakat syndrome.