Automated classification of otitis media with OCT: augmenting pediatric image datasets with gold-standard animal model data.

Otitis media (OM) is an extremely common disease that affects children worldwide. Optical coherence tomography (OCT) has emerged as a noninvasive diagnostic tool for OM, which can detect the presence and quantify the properties of middle ear fluid and biofilms. Here, the use of OCT data from the chinchilla, the gold-standard OM model for the human disease, is used to supplement a human image database to produce diagnostically relevant conclusions in a machine learning model. Statistical analysis shows the datatypes are compatible, with a blended-species model reaching ∼95% accuracy and F1 score, maintaining performance while additional human data is collected.

Assessing the Effect of Middle Ear Effusions on Wideband Acoustic Immittance Using Optical Coherence Tomography.

OBJECTIVES: Wideband acoustic immittance (WAI) noninvasively assesses middle ear function by measuring the sound conduction over a range of audible frequencies. Although several studies have shown the potential of WAI for detecting the presence of middle ear effusions (MEEs), determining the effects of MEE type and amount on WAI in vivo has been challenging due to the anatomical location of middle ear cavity. The purpose of this study is to correlate WAI measurements with physical characteristics of the middle ear and MEEs determined by optical coherence tomography (OCT), a noninvasive optical imaging technique. DESIGN: Sixteen pediatric subjects (average age of 7 ± 4 years) were recruited from the primary care clinic at Carle Foundation Hospital (Urbana, IL). A total of 22 ears (normal: 15 ears, otitis media with effusion: 6 ears, and acute otitis media: 1 ear, based on physician's diagnosis) were examined via standard otoscopy, tympanometry, OCT imaging, and WAI measurements in a busy, community-based clinical setting. Cross-sectional OCT images were analyzed to quantitatively assess the presence, type (relative turbidity based on the amount of scattering), and amount (relative fluid level) of MEEs. These OCT metrics were utilized to categorize subject ears into no MEE (control), biofilm without a MEE, serous-scant, serous-severe, mucoid-scant, and mucoid-severe MEE groups. The absorbance levels in each group were statistically evaluated at α = 0.05. RESULTS: The absorbance of the control group showed a similar trend when compared with a pediatric normative dataset, and the presence of an MEE generally decreased the power absorbance. The mucoid MEE group showed significantly less power absorbance from 2.74 to 4.73 kHz (p < 0.05) when compared with the serous MEE group, possibly due to the greater mass impeding the middle ear system. Similarly, the greater amount of middle ear fluid contributed to the lower power absorbance from 1.92 to 2.37 kHz (p< 0.05), when compared with smaller amounts of fluid. As expected, the MEEs with scant fluid only significantly affected the power absorbance at frequencies greater than 4.85 kHz. A large variance in the power absorbance was observed between 2 and 5 kHz, suggesting the dependence on both the type and amount of MEE. CONCLUSIONS: Physical characteristics of the middle ear and MEEs quantified from noninvasive OCT images can be helpful to understand abnormal WAI measurements. Mucoid MEEs decrease the power absorbance more than serous MEEs, and the greater amounts of MEE decreases the power absorbance, especially at higher (>2 kHz) frequencies. As both the type and amount of MEE can significantly affect WAI measurements, further investigations to correlate acoustic measurements with physical characteristics of middle ear conditions in vivo is needed.

Automated classification platform for the identification of otitis media using optical coherence tomography.

The diagnosis and treatment of otitis media (OM), a common childhood infection, is a significant burden on the healthcare system. Diagnosis relies on observer experience via otoscopy, although for non-specialists or inexperienced users, accurate diagnosis can be difficult. In past studies, optical coherence tomography (OCT) has been used to quantitatively characterize disease states of OM, although with the involvement of experts to interpret and correlate image-based indicators of infection with clinical information. In this paper, a flexible and comprehensive framework is presented that automatically extracts features from OCT images, classifies data, and presents clinically relevant results in a user-friendly platform suitable for point-of-care and primary care settings. This framework was used to test the discrimination between OCT images of normal controls, ears with biofilms, and ears with biofilms and middle ear fluid (effusion). Predicted future performance of this classification platform returned promising results (90%+ accuracy) in various initial tests. With integration into patient healthcare workflow, users of all levels of medical experience may be able to collect OCT data and accurately identify the presence of middle ear fluid and/or biofilms.

In vivo detection of nanometer-scale structural changes of the human tympanic membrane in otitis media.

Otitis media (OM) is a common ear infection and a leading cause of conductive hearing loss in the pediatric population. Current technologies such as otoscopy, pneumatic otoscopy, tympanometry, and acoustic reflectometry are used to diagnose OM, which can reasonably diagnose the infection with a sensitivity and specificity of 50-90% and 60-90%, respectively. However, these techniques provide limited information about the physical architecture of the tympanic membrane (TM), or what may lie behind it. Here, we report the detection of nanometer-scale structural changes of the TM using nano-sensitive optical coherence tomography (nsOCT). In total, an image dataset from 65 pediatric subjects from three different groups (normal, acute OM, and chronic OM) and with longitudinal image-based analysis of ear infections were included in this study. The nsOCT data were correlated with physician diagnosis and with OCT thickness measurements and were found to be in good agreement with these results. We report that nsOCT detects in vivo structural deformations of the TM earlier than OCT alone, and enhances the detection sensitivity of OCT measurements. This unique technique for early detection of nano-scale structural modifications in the TM has the potential to aid in our understanding of microbiological effects, and possibly for early diagnosis and more effective treatment of OM.

Pneumatic low-coherence interferometry otoscope to quantify tympanic membrane mobility and middle ear pressure.

Pneumatic otoscopy to assess the mobility of the tympanic membrane (TM) is a highly recommended diagnostic method of otitis media (OM), a widespread middle ear infection characterized by the fluid accumulation in the middle ear. Nonetheless, limited depth perception and subjective interpretation of small TM displacements have challenged the appropriate and efficient examination of TM dynamics experienced during OM. In this paper, a pneumatic otoscope integrated with low coherence interferometry (LCI) was adapted with a controlled pressure-generating system to record the pneumatic response of the TM and to estimate middle ear pressure (MEP). Forty-two ears diagnosed as normal (n = 25), with OM (n = 10), or associated with an upper respiratory infection (URI) (n = 7) were imaged with a pneumatic LCI otoscope with an axial, transverse, and temporal resolution of 6 µm, 20 µm, and 1 msec, respectively. The TM displacement under pneumatic pressure transients (a duration of 0.5 sec with an intensity of ± 150 daPa) was measured to compute two metrics (compliance and amplitude ratio). These metrics were correlated with peak acoustic admittance and MEP from tympanometry and statistically compared via Welch's t-test. As a result, the compliance represents pneumatic TM mobility, and the amplitude ratio estimates MEP. The presence of a middle ear effusion (MEE) significantly decreased compliance (p<0.001). The amplitude ratio of the OM group was statistically less than that of the normal group (p<0.01), indicating positive MEP. Unlike tympanometry, pneumatic LCI otoscopy quantifies TM mobility as well as MEP regardless of MEE presence. With combined benefits of pneumatic otoscopy and tympanometry, pneumatic LCI otoscopy may provide new quantitative metrics for understanding TM dynamics and diagnosing OM.

Noninvasive depth-resolved optical measurements of the tympanic membrane and middle ear for differentiating otitis media.

OBJECTIVE/HYPOTHESIS: In this study, optical coherence tomography (OCT) is used to noninvasively and quantitatively determine tympanic membrane (TM) thickness and the presence and thickness of any middle-ear biofilm located behind the TM. These new metrics offer the potential to differentiate normal, acute, and chronic otitis media (OM) infections in pediatric subjects. STUDY DESIGN: Case series with comparison group. METHODS: The TM thickness of 34 pediatric subjects was acquired using a custom-built, handheld OCT system following a traditional otoscopic ear exam. RESULTS: Overall thickness (TM and any associated biofilm) was shown to be statistically different for normal, acute, and chronic infection groups (normal-acute and normal-chronic: P value < 0.001; acute-chronic: P value = 0.0016). Almost all observed scans from the chronic group had an accompanying biofilm structure. When the thickness of the TM and biofilm were considered separately in chronic OM, the chronic TM thickness correlated with the normal group (P value = 0.68) yet was still distinct from the acute OM group (P value < 0.001), indicating that the TM in chronic OM returns to relatively normal thickness levels. CONCLUSION: Identifying these physical changes in vivo provides new metrics for noninvasively and quantitatively differentiating normal, acute, and chronic OM. This new diagnostic information has the potential to assist physicians to more effectively and efficiently screen, manage, and refer patients based on quantitative data. LEVEL OF EVIDENCE: 4.