Improving the Differential Diagnosis of Otitis Media With Effusion Using Wideband Acoustic Immittance

Abstract

Objectives: The objective of this work is to determine whether there is a systematic effect of middle ear effusion volume on wideband acoustic immittance in children with surgically confirmed otitis media with effusion. Design: Wideband acoustic immittance was measured in 49 ears from children (9 months to 11 years) who had a diagnosis of otitis media with effusion and compared to 14 ears from children (10 months to 10 years) without a recent history of otitis media. For children with otitis media with effusion, wideband acoustic immittance testing took place in the child’s preoperative waiting room before surgical placement of tympanostomy tubes. Testing was completed in a pressurized condition (wideband tympanometry) for all ears as well as in an ambient condition in a subset of ears. Intraoperative findings regarding effusion volume were reported by the surgeons immediately before tube placement and confirmed following myringotomy. This classified the volume of effusion as compared to middle ear volume categorically as either full, partial, or clear of effusion. The type of wideband acoustic immittance explored in this work was absorbance. Absorbance responses were grouped based on effusion volume into one of four groups: full effusions, partial effusions, ears clear of effusion at the time of surgery, and normal control ears. Standard tympanometry was also completed on all ears. Results: Absorbance is systematically reduced as the volume of the middle ear effusion increases. This reduction is present at most frequencies but is greatest in the frequency range from 1 to 5 kHz. A multivariate logistic regression approach was utilized to classify ears based on effusion volume. The regression approach classified ears as effusion present (full and partial ears) or absent (clear ears and normal control ears) with 100% accuracy, ears with effusion present as either partial or full with 100% accuracy, and ears without effusion as either normal control ears or ears clear of effusion with 75% accuracy. Regression performance was also explored when the dataset was split into a training set (70% of the data) and a validation test set (30% of the data) to simulate how this approach would perform on unseen data in a clinical setting. Accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve are reported. Overall, this approach demonstrates high sensitivity and specificity for classifying ears as effusion being present or absent and as present effusions being full or partial with areas under the curve ranging from 1 to 0.944. Despite the lack of effusion present in both clear ears and normal control ears, this approach was able to distinguish between these ears, but with a more moderate sensitivity and specificity. No systematic effect of effusion volume was found on standard tympanometry. Conclusions: Wideband acoustic immittance, and more specifically, absorbance, is a strong and sensitive indicator of the volume of a middle ear effusion in children with otitis media with effusion.

A Deep Learning Approach for Detecting Otitis Media From Wideband Tympanometry Measurements

“…So far, this has been limited to the binary classification of OM and NOE. Merchant et al [15] created a multivariate prediction model based on the three first principal components using logistic regression, showing good results. Their study concludes that wideband absorbance is a strong and sensitive indicator of the effusion volume.…”

“…Besides our proposed methods, we have also run experiments with the methods proposed by Merchant et al [15] and Grais et al [20] for comparison. These methods were trained and tested using our dataset to ensure a proper comparison.…”

“…These methods were trained and tested using our dataset to ensure a proper comparison. Merchant et al [15] propose an approach based on a multivariate logistic classification model based on the three first principal components of the WBT measurements. We trained the binary classification model to predict OM or NOE, and tested it on our test dataset.…”

The Role of Wideband Tympanometry in the Diagnosis of Meniere's Disease

“…Currently, WBT has shown significant advantages in the diagnosis of ME diseases. The absorbance in the WBT systematically decreases with increasing ME effusion accumulation, which is a powerful and sensitive indicator of the volume of ME effusion in children with secretory otitis media (16). However, there are few studies using WBT to diagnose MD, and the sample sizes included in these studies are generally small.…”

“…Keefe et al developed a test method for WBT and obtained data for clinical measurements (15). WBT can provide more sensitive and specific results for middle ear (ME) pathologies such as secretory otitis media, otosclerosis, and ossicular chain disruption, thus helping clinicians differentiate diagnoses (16)(17)(18). Changes in IE compartment pressure can induce symptoms in some IE diseases, and WBT has been found to be helpful in the pathological diagnosis of elevated intracranial pressure and IE pressure (8).…”