Interest of Using Deep Learning Algorithm for Otosclerosis Detection on Temporal Bone High Resolution CT (OtoIA)

Otosclerosis is a relatively frequent pathology, of multifactorial origin with genetic and hormonal part, predominantly in women. This disease causes a disorder of the bone metabolism of the middle and inner ear, responsible for a progressive deafness, which can become severe.

Several elements are necessary to make the diagnosis of otosclerosis: the clinical examination and questioning, the audiometric assessment, and finally the temporal bone CT.

The CT scan allows to detect foci of otosclerosis within the bone of the middle or inner ear. This diagnosis is sometimes difficult and requires interpretation by a trained radiologist.

The investigators would like to evaluate the ability of a deep learning algorithm to detect these foci of otosclerosis, and to compare its diagnostic performance with a trained radiologist.

Study Overview

• Status: Active, not recruiting
• Conditions: Otosclerosis
• Intervention / Treatment: Combination product: Radiologic diagnosis; Diagnostic test: Artificial intelligence diagnosis

Detailed Description

Otosclerosis is a relatively frequent pathology, of multifactorial origin with genetic and hormonal part, predominantly in women. This disease causes a disorder of the bone metabolism of the middle and inner ear, responsible for a progressive deafness, which can become severe.

Several elements are necessary to make the diagnosis of otosclerosis: the clinical examination and questioning, the audiometric assessment, and finally the temporal bone CT.

The CT scan allows to detect foci of otosclerosis within the bone of the middle or inner ear. This diagnosis is sometimes difficult and requires interpretation by a trained radiologist.

The investigators would like to evaluate the ability of a deep learning algorithm to detect these foci of otosclerosis, and to compare its diagnostic performance with a trained radiologist.

• Study Type: Observational
• Enrollment (Estimated): 240

Contacts and Locations

Study Locations

• France
  • Pierre-Bénite, France, 69310
    • Hospices Civils de Lyon, Centre Hospitalier Lyon sud, Service d'ORL, d'otoneurchirurgie et de chirurgie cervico-facaile

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Adult; Older Adult
• Accepts Healthy Volunteers: Yes

• Sampling Method: Non-Probability Sample

Study Population

The study population will include cases from a group of patients who have undergone surgery for otosclerosis at the Lyon Sud hospital in France, and controls who have had a high resolution temporal bone CT scan at the Lyon Sud hospital with no clinical suspicion for otosclerosis and considered normal.

Description

• Inclusion Criteria * :

  • age over 18
  • high resolution temporal bone CT scan available for analysis
  • for the "case" group : surgical confirmation of positive diagnosis for otosclerosis
  • for the "control" group : a first radiological analysis in favor of a normal temporal bone CT scanner and an initial radiologic report considered normal as well

• Exclusion Criteria * :

  • age under 18
  • no high resolution temporal bone CT scan available for analysis
  • unwillingness to participate in the study

Study Plan

How is the study designed?

Number of groups / cohorts

2

Cohorts and Interventions

Group / Cohort	Intervention / Treatment
CASE; Patients with surgically confirmed otosclerosis who initially consulted for conductive hearing loss with normal otoscopy, and with a high resolution computed tomography of temporal bone available	Combination product: Radiologic diagnosis; Each CT scan is interpreted by a radiologist and is assigned as positive or negative for the diagnosis of otosclerosis; Diagnostic test: Artificial intelligence diagnosis; Each CT scan is screened by the deep learning algorithm and is assigned as positive or negative for the diagnosis of otosclerosis
CONTROL; Random patients with a high resolution computed tomography scan of temporal bone performed without suspicion of otosclerosis and considered normal	Combination product: Radiologic diagnosis; Each CT scan is interpreted by a radiologist and is assigned as positive or negative for the diagnosis of otosclerosis; Diagnostic test: Artificial intelligence diagnosis; Each CT scan is screened by the deep learning algorithm and is assigned as positive or negative for the diagnosis of otosclerosis

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Diagnostic performance of the artificial intelligence algorithm compared to the diagnostic performance of the radiologist : sensitivity, specificity, positive and negative predictive value, area under the ROC curve	These diagnostic performances will be established from the positive or negative diagnoses of the algorithm and the radiologist, compared to the "case" or "control" status of each patient included in the study	through study completion, an average of 5 months

Collaborators and Investigators

• Sponsor: Hospices Civils de Lyon
• Investigators: Principal Investigator: Maxime FIEUX, MD, Hospices Civils de Lyon

Study record dates

Study Major Dates

• Study Start (Actual): July 1, 2022
• Primary Completion (Actual): May 1, 2023
• Study Completion (Estimated): October 1, 2023

Study Registration Dates

• First Submitted: May 17, 2023
• First Submitted That Met QC Criteria: August 9, 2023
• First Posted (Actual): August 14, 2023

Study Record Updates

• Last Update Posted (Actual): August 14, 2023
• Last Update Submitted That Met QC Criteria: August 9, 2023
• Last Verified: May 1, 2023

More Information

Terms related to this study

• Keywords: deep learning; IA; otosclerosis; temporal bone CT
• Additional Relevant MeSH Terms: Otorhinolaryngologic Diseases; Ear Diseases; Otosclerosis
• Other Study ID Numbers: 22-5019 / 69HCL22_1193

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No