A Proof-of-Concept Computer Vision Approach for Measurement of Tympanic Membrane Perforations

OBJECTIVE

Size, an important characteristic of a tympanic membrane perforation (TMP), is commonly assessed with gross estimation via visual inspection, a practice which is prone to inaccuracy. Herein, we demonstrate feasibility of a proof-of-concept computer vision model for estimating TMP size in a small set of perforations.

METHODS

An open-source deep learning architecture was used to train a model to segment and calculate the area of a perforation and the visualized tympanic membrane (TM) in a set of endoscopic images of mostly anterior and relatively small TMPs. The model then computed relative TMP size by calculating the ratio of perforation area to TM area. Model performance on the test dataset was compared to ground-truth manual annotations. In a validation survey, otolaryngologists were tasked with estimating the size of TMPs from the test dataset. The primary outcome was the average absolute error of model size predictions and clinician estimates compared to sizes determined by ground-truth manual annotations.

RESULTS

The model's average absolute error for size predictions was a 0.8% overestimation for all test perforations. Conversely, among the 38 survey respondents, the average clinician error was a 11.0% overestimation (95% CI, 5.2-16.7%, p = 0.003).

CONCLUSIONS

In a small sample of TMPs, we demonstrated a computer vision approach for estimating TMP size is feasible. Further validation studies must be done with significantly larger and more heterogenous datasets.

LEVEL OF EVIDENCE

N/A Laryngoscope, 134:2906-2911, 2024.

Assessment of tympanic membrane perforation size using ImageJ software by ENT clinicians of different grades

OBJECTIVE

To compare visual estimation versus ImageJ calculation of tympanic membrane perforation size in the paediatric population between clinicians of different experience.

METHODS

Five images of tympanic membrane perforations in children, captured using an otoendoscope, were selected. The gold standard was the ImageJ results by one consultant otologist. Consultants, registrars and Senior House Officers or equivalent were asked to visually estimate and calculate the perforation size using ImageJ software.

RESULTS

The mean difference in variation from gold standard between visual estimation and ImageJ calculation was 12.16 per cent, 95 per cent CI (10.55, 13.78) p < 0.05, with ImageJ providing a more accurate estimation of perforation. Registrars were significantly more accurate at visual estimation than senior house officers. There was no statistically significant difference in ImageJ results between the different grades.

CONCLUSION

Using ImageJ software is more accurate at estimating tympanic membrane perforation size than visual assessment for all ENT clinicians regardless of experience.

Quantitative Measurements in Otological Surgery: Use of an Endoscopic Integrated Multipoint Laser System

OBJECTIVES

Assess the clinical utility of an Endoscopic Integrated Multipoint Laser System (EMLS) to otology. This is an emerging technology from automotive engineering that may offer the ability to accurately measure anatomy and pathology using an endoscope while undertaking ear surgery.

PATIENTS

Simulated otology patients were used incorporating the Phacon Temporal Bone synthetic models and Kyoto Kagaku Ear Examination Simulator models to allow assessment of the EMLS technology in evaluating external ear and middle ear pathology, e.g., perforation or prosthesis sizing.

INTERVENTION

Eight otolaryngology resident and fellows at a tertiary university teaching hospital were given training in EMLS and reviewed simulated anatomy and pathology within the models including tympanic membrane perforation, ossicular discontinuity, and a cochleostomy.

MAIN OUTCOME MEASURE

Variance in measurement was assessed in relation to those made manually by an independent surgeon using surgical calipers (0.1 mm).

RESULTS

The 8 participants produced 47 mean measurements. The mean difference from independently made manual measurement was 0.294 mm (standard error of the mean 0.033). Maximum variance was 0.98 mm and minimum 0.01 mm.

CONCLUSION

Use of an integrated endoscopic laser measurement tool allows reliable, easy-to-obtain measurements to be obtained within a simulated otological surgical environment. Translation of the technology to a thinner delivery system through a rigid endoscope offers further promise for routine use in a clinical setting.

Myringoplasty: Impact of Size and Site of Perforation on the Success Rate

Myringoplasty aims to reconstruct the tympanic membrane, restoring protection to the middle ear and improve hearing. Success of Myringoplasty in terms of anatomical closure is influenced by many factors. This study focuses on the influence of size and site of tympanic membrane perforation on success rate of Myringoplasty. This is a prospective study of 60 patients who underwent myringoplasty and were then followed up for 6 months from the date of operation. Patients with dry perforation, with good cochlear reserve, intact and mobile ossicular chain, functioning Eustachian tube were selected randomly for the operation. Video Otoendoscopy of all cases with storz 0 degree endoscope were done. All images were recorded on the computer [DELL VOSTRO 3400]. Using 'Image J' [version 1.35j] geometrical package, the area of perforation (P) and the entire area of the tympanic membrane (T) were calculated. Then, the percentage area of the perforation [P/T × 100 %] for each ear was obtained. Site of perforation also documented. Success rate for pin-point and small perforations was 100 %, for medium size 80 %, and for large & subtotal perforations 69.2 and 42.9 % respectively. We regard size of the perforation as a major factor that influences success of Myringoplasty. Site of perforation failed to be a statistically significant influencing factor in this study.

Tympanic membrane perforation: size, site and hearing evaluation

OBJECTIVE

To assess different clinical scales of TM perforation size; to evaluate the effect of the size and the site of a perforation on the hearing level and frequencies.

METHODS

Prospective study. Observers had subjectively estimated the size in millimeter and in percentage of a particular perforation; objectively computerized measures of TM perforations area were analyzed. Agreement between different measures was studied. Cases with postoperative intact TM and an air-bone gap (ABG) of 10 dB or less were studied.

RESULTS

Global mean preoperative ABG was 21.8 ± 17 dB. Preoperative ABG was different between small-large and small-total perforations (p=0.001). Difference of the preoperative ABG was statistically significant between perforations filling up the four quadrants and perforation limited to one quadrant in the postero-inferior, antero-superior and antero-inferior site. No statistically difference between perforation sites was identified for each affected frequency. Difference is statistically significant (p=0.001) between the 250 Hz and the other frequencies for the medium, large and total perforations. After myringoplasty bone conduction improvement was statistically significant for the frequencies 500 (p=0.04), 1000 (p=0.04) and 2000 Hz (p=0.011). Agreement was large enough when TM perforation size was expressed in percentage and absent when expressed in millimeter.

CONCLUSION

TM perforation can be clinically estimated quite precisely as a percentage of the TM area. Conductive hearing loss is frequency dependent; with the greatest loss occurring at the lowest sound frequencies. Hearing loss does not depend on the perforation's location.